Handbook Of Geospatial Approaches To Sustainable

Cities 1st Edition Qihao Weng download

https://ebookbell.com/product/handbook-of-geospatial-approaches-
to-sustainable-cities-1st-edition-qihao-weng-56484220

Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.

Handbook Of Geospatial Artificial Intelligence Song Gao Yingjie Hu

https://ebookbell.com/product/handbook-of-geospatial-artificial-
intelligence-song-gao-yingjie-hu-54993888

The Routledge Handbook Of Geospatial Technologies And Society 1st

Edition Alexander J Kent

https://ebookbell.com/product/the-routledge-handbook-of-geospatial-
technologies-and-society-1st-edition-alexander-j-kent-50795592

Handbook Of Big Geospatial Data 1st Edition Martin Werner Yaoyi Chiang

https://ebookbell.com/product/handbook-of-big-geospatial-data-1st-
edition-martin-werner-yaoyi-chiang-26746742

Handbook Of Research On Geospatial Science And Technologies 1st

Edition Joyce Gosata Maphanyane

https://ebookbell.com/product/handbook-of-research-on-geospatial-
science-and-technologies-1st-edition-joyce-gosata-maphanyane-54012514
Handbook Of Psychology Developmental Psychology Richard M Lerner M A
Easterbrooks Jayanthi Mistry

https://ebookbell.com/product/handbook-of-psychology-developmental-
psychology-richard-m-lerner-m-a-easterbrooks-jayanthi-mistry-44869188

Handbook Of Home Language Maintenance And Development Andrea C

Schalley

https://ebookbell.com/product/handbook-of-home-language-maintenance-
and-development-andrea-c-schalley-44875220

Handbook Of What Works With Sexual Offenders Contemporary Perspectives

In Theory Assessment Treatment And Prevention 1st Edition Jean Proulx

https://ebookbell.com/product/handbook-of-what-works-with-sexual-
offenders-contemporary-perspectives-in-theory-assessment-treatment-
and-prevention-1st-edition-jean-proulx-44880746

Handbook Of Dialysis Therapy 6th Edition Allen R Nissenson Richard N

Fine Rajnish Mehrotra Joshua Zaritsky

https://ebookbell.com/product/handbook-of-dialysis-therapy-6th-
edition-allen-r-nissenson-richard-n-fine-rajnish-mehrotra-joshua-
zaritsky-44887338

Handbook Of Revolutions In The 21st Century The New Waves Of

Revolutions And The Causes And Effects Of Disruptive Political Change
1st Edition Jack A Goldstone Editor

https://ebookbell.com/product/handbook-of-revolutions-in-the-21st-
century-the-new-waves-of-revolutions-and-the-causes-and-effects-of-
disruptive-political-change-1st-edition-jack-a-goldstone-
editor-44887720
 Handbook of Geospatial
Approaches to Sustainable Cities

This comprehensive handbook presents the current state of knowledge on geospatial

technologies, techniques, and methods that are imperative for providing solutions to
sustainable cities. It addresses the role of geospatial big data and AI techniques and
how they are applied when analyzing the sustainability of urban development, land
use, urban planning, and resource management, as well as monitoring the impact
urbanization has on the environment and the ecosystem. Taking an interdisciplinary
approach to sustainable cities, and with contributions from renowned experts
around the world, this holistic handbook is a toolbox for geospatial, urban, and
sustainability professionals, the artificial intelligence community, and those who
work in related fields.

Features:

• Explores cutting-edge geospatial and AI techniques in support of efficient,

resilient, digital, and smart cities
• Bridges urban science and sustainability science via geospatial methods
• Contributes to the efforts of GEO by addressing and exemplifying
pertinent societal benefit areas and engagement priorities
• Includes 16 case studies with a broad geographic scope that integrate
societal needs with technological advances
• Draws expertise in geospatial technology, big data, and artificial intelli­
gence from leading experts in the world

This book is intended for researchers and scientists interested in learning techniques
in GeoAI, including the technologies for collecting, analyzing, managing, processing,
and visualizing geospatial datasets.
Imaging Science Series
Qihao Weng
Hong Kong Polytechnic University

Imaging science is a multidisciplinary field concerned with the generation,

collection, duplication, analysis, modification, and visualization of images,
including imaging things that the human eye cannot detect. As an evolving field, it
includes research and researchers from physics, mathematics, electrical engi­
neering, computer vision, computer science, remote sensing, and perceptual
psychology. The book series will review the subfields within imaging science that
include: image processing, computer vision, 3D computer graphics, animations,
atmospheric optics, astronomical imaging, digital image restoration, digital
imaging, color science, digital photography, holography, magnetic resonance
imaging, medical imaging, microdensitometry, optics, photography, remote
sensing, radar imaging, radiometry, silver halide, ultrasound imaging, photoacoustic
imaging, thermal imaging, visual perception, and various printing technologies. The
books in this series assume that readers have a basic knowledge of remote sensing
and imaging science, and therefore will not cover very basic materials but
concentrate on those aspects of imaging science and the application discipline(s)
that are pertinent for a ‘how to’ knowledge. This provides academic authors and
leading professionals an opportunity to combine their research, consultancy, and
interesting project work into a book that is then relevant to professional lines of
work or teaching modules focusing on imaging science applications.

High Spatial Resolution Remote Sensing: Data, Analysis, and Applications

Edited by Yuhong He and Qihao Weng
Remote Sensing Time Series Image Processing
Qihao Weng
Handbook of Geospatial Approaches to Sustainable Cities
Edited by Qihao Weng, in collaboration with Cheolhee Yoo
https://www.routledge.com/Imaging-Science/book-series/CRCIS
 Handbook of Geospatial
Approaches to Sustainable
Cities

Edited by
Qihao Weng
in collaboration with Cheolhee Yoo

The Hong Kong Polytechnic University

The Hong Kong Institute of Surveyors

Designed cover image: © Tavarius, Shutterstock

First edition published 2024

by CRC Press
2385 NW Executive Center Drive, Suite 320, Boca Raton FL 33431
and by CRC Press
4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN
CRC Press is an imprint of Taylor & Francis Group, LLC

© 2024 selection and editorial matter, Qihao Weng; individual chapters, the contributors

Reasonable efforts have been made to publish reliable data and information, but the author
and publisher cannot assume responsibility for the validity of all materials or the
consequences of their use. The authors and publishers have attempted to trace the copyright
holders of all material reproduced in this publication and apologize to copyright holders if
permission to publish in this form has not been obtained. If any copyright material has not
been acknowledged please write and let us know so we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted,
reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other
means, now known or hereafter invented, including photocopying, microfilming, and
recording, or in any information storage or retrieval system, without written permission from
the publishers.
For permission to photocopy or use material electronically from this work, access
www.copyright.com or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. For works that are not available on CCC please
contact [email protected]
Trademark notice: Product or corporate names may be trademarks or registered trademarks
and are used only for identification and explanation without intent to infringe.

ISBN: 978-1-032-15481-7 (hbk)

ISBN: 978-1-032-15534-0 (pbk)
ISBN: 978-1-003-24456-1 (ebk)
DOI: 10.1201/9781003244561

Typeset in Times
by MPS Limited, Dehradun
Contents
Preface.....................................................................................................................xiii
Acknowledgments...................................................................................................xvi
Editors ....................................................................................................................xvii
Contributors ..........................................................................................................xviii

PART I Artificial Intelligence and Big Data Analytics:

Needs and Requirements
Chapter 1 Sensing Urban Physical Environment with GeoAI
and Street-Level Imagery....................................................................3
Ce Hou, Yong Li, and Fan Zhang
Introduction..........................................................................................3
Street-Level Imagery (SI).................................................................... 5
Street-Level Imagery Sources ..................................................... 5
Street-Level Imagery Attributes.................................................. 6
Methods for Sensing the Urban Physical Environment.....................7
Sensing Tasks .............................................................................. 7
Approaches Based on Techniques ............................................ 10
Applications of Urban Sensing .........................................................12
Observational Sensing: Sensing the Physical Entity................ 13
Urban Implicit Relationship Sensing: Sensing Beyond
the Image ................................................................................... 16
Human Perception Sensing: Sensing with
Human-Centric Perspective....................................................... 18
Future Trends..................................................................................... 19
References..........................................................................................20

Chapter 2 Geospatial Big Data for Urban Sustainability Science....................31

Zidong Yu and Xintao Liu
Urban Sustainability Science ............................................................31
Geospatial Big Data .......................................................................... 32
Application of Using Geospatial Big Data.......................................33
Agglomeration Economies ........................................................ 33
Urban Freight............................................................................. 35
Urban Functions and Transit-Oriented Development
(TOD)......................................................................................... 37
Conclusions........................................................................................40
Future Challenges ..............................................................................41
References..........................................................................................42
v
vi Contents

Chapter 3 Urban Flooding Monitoring and Management in Geospatial

Perspective: Data, Techniques, and Platforms ................................. 44
Zhiwei Li, Cheolhee Yoo, and Qihao Weng
Urban Sustainability Science ............................................................44
Geospatial Data for Urban Flooding Monitoring.............................46
Geospatial Techniques for Urban Flooding Management ...............46
Flood Modeling and Forecasting .............................................. 46
Flood Mapping and Risk Assessment ...................................... 47
Flooding Impact Assessment, Recovery, and Adaptation........ 48
Geospatial Platforms for Urban Flooding Management ..................48
Problems and Prospects.....................................................................50
Data Availability and Quality Issues ........................................ 50
Technical Limitations and Gaps ............................................... 50
Geospatial Artificial Intelligence-Enabled Flood
Monitoring and Management .................................................... 51
Integration of Geospatial Platforms with Other Urban
Management Systems ................................................................ 51
Conclusions........................................................................................52
References..........................................................................................53

PART II Geospatial Techniques for Renewable Cities

Chapter 4 Assessing Sustainability in China’s Resource-Based Cities:

Leveraging Remote Sensing Techniques for Evaluating
Environment-Related SDG Progress................................................. 61
Huijuan Xiao and Song Xue
Introduction........................................................................................61
Methods..............................................................................................62
City Categorization.................................................................... 62
Data Collection .......................................................................... 63
Multi-Criteria Decision Making................................................ 66
Results................................................................................................68
Weight of SDG Indicators ........................................................ 68
Score of Each Environment-Related SDG ............................... 69
Overall Score of Environment-Related SDGs.......................... 71
Conclusion .........................................................................................72
References..........................................................................................73
Contents vii

Chapter 5 Urban Green Space Maps Based on GeoAI..................................... 75

Yang Chen
Introduction: Background and Driving Forces................................. 75
GeoAI Method................................................................................... 77
Data Preprocessing .................................................................... 78
Multi-Scale Feature Extraction Module.................................... 79
Multi-Modal Information Fuse Module.................................... 81
Boundary Enhancement Module............................................... 82
Study Area and Datasets ...................................................................83
Study Area ................................................................................. 83
Datasets ...................................................................................... 84
Mapping Urban Green Spaces with Different Methods ..................84
Conclusion .........................................................................................88
References..........................................................................................89

Chapter 6 Contrasting Pattern of Urban Expansion and Urban Land

Use Intensification of Global Megacities Using Nighttime
Light Time Series Data .....................................................................94
Qiming Zheng and Qihao Weng
Introduction........................................................................................94
Study Area and Datasets ...................................................................95
Methods..............................................................................................96
Time Series Modeling ............................................................... 96
Urban Built-Up Area Mapping ................................................. 98
Urban Land Change Classification and Analysis..................... 99
Results.............................................................................................. 100
Time Series Modeling ............................................................. 100
Urban Built-Up Area Mapping ............................................... 101
Urban Land Change Pattern.................................................... 102
Discussions ......................................................................................104
Conclusion .......................................................................................105
References........................................................................................106

Chapter 7 The Potential of Nature-Based Solutions in Urban Heat

Mitigation and Building Energy Savings .......................................109
Siqi Jia, Qihao Weng, and Yuhong Wang
Introduction......................................................................................109
Nature-Based Solutions for Urban Heat Mitigation
and Building Energy Savings.......................................................... 110
Green Roofs............................................................................. 111
Evaporative and Water-Retentive Pavements......................... 112
Street Trees .............................................................................. 112
viii Contents

Methodology .................................................................................... 114

Assessment of Thermal Comfort ............................................ 114
Microclimatic Modeling .......................................................... 115
Building Energy Simulation.................................................... 119
Results.............................................................................................. 123
Performance of Nature-Based Solutions in Urban
Heat Mitigation........................................................................ 123
Performance of Nature-Based Solutions in Building
Energy Savings ........................................................................ 126
Discussion ........................................................................................126
The Effectiveness of NBS in Cooling the Environment
and Improving Thermal Comfort............................................ 126
The Effectiveness of Green Roofs in Reducing
Building Energy Use ............................................................... 128
Conclusion .......................................................................................129
References........................................................................................129

PART III Geospatial Techniques for Resilient Cities

Chapter 8 GeoAI for High-Resolution Urban Air Temperature

Estimation and Urban Heat Island Monitoring .............................. 139
Cheolhee Yoo and Qihao Weng
Introduction......................................................................................139
Study Area and Data ....................................................................... 141
Study Area ............................................................................... 141
Satellite and In-Situ Meteorological Data .............................. 142
Local Climate Zone Map ........................................................ 144
Methods............................................................................................ 144
Variable Processing ................................................................. 144
Air Temperature Estimation.................................................... 144
Urban Heat Islands Monitoring .............................................. 147
Results and Discussion....................................................................148
LST Downscaling Results....................................................... 148
Hourly Air Temperature Estimation Accuracy
Assessment............................................................................... 149
Hourly Air Temperature Mapping Analysis........................... 151
Diurnal Cycle of UHI Intensity .............................................. 153
Novelty and Limitation ........................................................... 154
Conclusion .......................................................................................155
References........................................................................................156
Contents ix

Chapter 9 Satellite-Based Assessment of Urban Thermal Environments ......159

Yuhei Yamamoto
Introduction......................................................................................159
Acquisition and Preprocessing of Satellite LST Data.................... 160
Data Selection.......................................................................... 160
Overview of LST Retrieval..................................................... 162
SC Algorithm........................................................................... 164
SW Algorithm.......................................................................... 165
TES Algorithm ........................................................................ 165
Urban-Specific LST Errors and Thermal Anisotropy ............ 166
Scaling and Classification of LST .......................................... 166
Evaluation of SUHI......................................................................... 167
Overall SUHII.......................................................................... 168
SUHII Distribution .................................................................. 171
Challenges and Future Perspectives................................................ 174
Conclusions......................................................................................175
References........................................................................................175

Chapter 10 Urban Air Pollution Mitigation for Sustainable Cities:

Observation, Modeling, and Control Strategies .............................184
Hai Guo, Yu Wang, Yangzong Zeren, and Xiaoyu Jiao
Introduction: Background and Driving Forces............................... 184
O3 and Its Primary Emitted Precursors .................................. 184
Airborne Particulate and Aerosols .......................................... 185
Methodology .................................................................................... 188
Sampling Site Selection .......................................................... 188
Measurement Techniques ........................................................ 189
Modeling and Calculation Techniques ................................... 194
Observations of Urban Air Pollutants ............................................ 201
Intensive Samplings of Urban Air Pollutants......................... 201
Long-Term Variations of Urban Air Pollutants ..................... 208
Numerical Simulations on Urban Air Pollution.............................211
Simulation of Atmospheric Dynamics.................................... 211
Simulation of Atmospheric Chemistry ................................... 213
Source Apportionments of Urban Air Pollutants ................... 214
Control Strategies on Urban Air Pollution.....................................216
Potential Control Measures for O3 Pollution ......................... 216
References........................................................................................220
x Contents

Chapter 11 Geospatial Analysis of Emerging Drought Risk in a

Warming Climate ............................................................................234
Shuo Wang and Yamin Qing
Introduction......................................................................................234
Spatiotemporal Evolution of Flash Drought...................................235
Underlying Mechanism of Flash Drought ......................................238
Drought-Related Compound Extremes ........................................... 240
Summary and Conclusions.............................................................. 242
References........................................................................................243

Chapter 12 Urban Summer Air Temperature Forecasting through

a Fusion of a Numerical Weather Prediction Model
with Machine Learning ...................................................................245
Dongjin Cho, Siwoo Lee, and Jungho Im
Introduction......................................................................................245
Machine Learning Algorithms ........................................................247
Machine Learning.................................................................... 247
Deep Learning ......................................................................... 248
Overview of Study Area and Numerical Weather
Prediction Model for Case Studies .................................................249
Study Area: Seoul.................................................................... 249
Local Data Assimilation and Prediction System Model ........ 249
Case Study 1: Machine Learning-Based Spatial
Distribution Production of Bias-Corrected Urban
Air Temperature Forecast................................................................ 250
Data .......................................................................................... 250
Method ..................................................................................... 250
Results and Discussion............................................................ 252
Case Study 2: Deep Learning-Based
Post-Processing
of Urban Air Temperature Forecast................................................ 256
Data .......................................................................................... 256
Method ..................................................................................... 256
Results and Discussion............................................................ 258
Conclusion .......................................................................................260
References........................................................................................261

PART IV Digital Cities

Chapter 13 GIS-Based Modeling for Estimating Urban Carbon Emissions ....267

Cai Meng and Ren Chao
Introduction......................................................................................267
Contents xi

Data Sources and Methodologies for GIS-Based

Emissions Modeling ........................................................................ 268
Top-Down Method Using Nighttime Lights .......................... 268
Bottom-Up Method.................................................................. 269
Hybrid Method ........................................................................ 269
Cross-Comparison of the Methods ......................................... 270
Case Studies of GIS-Based Modeling for Urban Carbon
Emissions Estimation ...................................................................... 271
Modeling Spatiotemporal Carbon Emissions for
Two Mega-Urban Regions ...................................................... 271
Modeling High-Resolution of Carbon Emissions for
Two Mega-Urban Regions ...................................................... 275
Challenges and Opportunities in GIS-Based
Emissions Modeling ........................................................................ 277
Conclusion .......................................................................................280
References........................................................................................281

Chapter 14 The Relationship between Visual Space and Elders’

Mental Health within 15-Minute Life Circles Using SVIs:
A Case Study of Beijing ................................................................. 285
Jiang Ying
Introduction......................................................................................285
Study Area .......................................................................................286
Data ..................................................................................................287
Mental Health Data ................................................................. 287
Residence Communities Data in Haidian District.................. 288
Street View Image Data .......................................................... 288
Methodology .................................................................................... 291
Semantic Segmentation for Street View Images.................... 292
Street Visual Perception Assessment...................................... 294
The Spatial Clustering of Visual Indicators for
Residential Communities......................................................... 296
Regression Analysis between Visual Perception of
Urban Environment and Mental Health of the Elderly.......... 296
Results.............................................................................................. 299
Street Visual Perception within 15-Minute Walking
Circles of Residence Communities......................................... 299
Relationship between Visual Perception within
15-Minute Walking Circle of Residence
Communities and Mental Health Outcomes........................... 302
Discussion ........................................................................................307
Urban Planning and the Visual Environment around
Residential Blocks ................................................................... 307
Interpretation of the Impacts of Urban Street Visual
Environment on Mental Health of the Elderly....................... 308
xii Contents

Strengths and Limitations........................................................ 308

Conclusions......................................................................................309
References........................................................................................310

Chapter 15 Comparisons of Deep Learning Models for Dynamic

Local Climate Zone Mapping .........................................................313
Haojie Chen, Cheolhee Yoo, and Qihao Weng
Introduction......................................................................................313
Methodology .................................................................................... 314
GIS-Based Approach............................................................... 314
EO-Based Approaches............................................................. 315
Hybrid Models......................................................................... 315
Convolutional Neural Networks (CNN) Models.................... 316
Data Processing and CNN Models Training .......................... 317
LCZ Mapping Results .....................................................................320
Discussion and Conclusion ............................................................. 321
References........................................................................................321

Chapter 16 The Potential Diffusion and Economic Impacts of Private

Autonomous Vehicles: A Spatial Agent-Based Model..................325
Chengxiang Zhuge and Justin Hayse Chiwing G. Tang
Introduction......................................................................................325
Literature Review ............................................................................326
Analysis and Modeling of AV Adoption ............................... 326
Impacts of AV Adoption......................................................... 327
Research Gaps and Aims ........................................................ 328
A Spatial Agent-Based Autonomous Vehicle (AV)
Diffusion Model ..............................................................................328
Model Framework ................................................................... 328
Stage 1 – Screening: Generating Consumer Agents .............. 329
Stage 2 – Decision Making: Choose a Vehicle Type............ 330
A Case Study of Beijing, China ..................................................... 331
Study Area: Beijing................................................................. 331
Description of AV Diffusion Scenarios.................................. 331
The Potential Diffusion of AVs.............................................. 333
Potential Economic Impacts of the AV Diffusion ................. 335
Conclusion .......................................................................................344
References........................................................................................345

Index......................................................................................................................348
Preface
Geospatial technology is one of the three most important, evolving fields in the
21st century, along with nanotechnology and biotechnology. Earth Observation (EO)
technologies, in conjunction with field surveys, have been used to observe, monitor,
measure, and model many of the components that comprise the ecosystems, contributing
to the sustainability and productivity of natural and human systems. Driven by societal
needs and improvement in geospatial technology, we have witnessed a great increase in
research and development, technology transfer, and engineering activities worldwide.
Their integration with the Internet and communications technologies, as well as
Geographic Information System, have led to the emergence of geospatial big data over
the web and on hands. These geo-referenced big data and imagery, in conjunction with
GPS, mobile mapping, and modern telecommunication technologies, have sparked
increasing interest in governments, industries, and the public.
Artificial Intelligence (AI), especially deep learning, has transformed the field of
computer vision, and now rivals human-level performance in certain tasks, such as
image recognition and semantic segmentation. In this context, there is a strong need
for research and development of new approaches to multi-sensory and multi-modal
deep learning within the new processing and recognition frameworks of remotely
sensed imagery. Since the 21st century, with the improvement of spatial resolution
of remote sensing images, image classification methods have gradually developed
into three parallel branches at different levels: pixel-level, object-level, and scene-
level classification. Many AI-based methods employ scene classification, which
uses the layout of objects within the scene, in addition to the ambient context, for
classification. It requires solving the problems in data training and validation, model
generalization across different modalities of EO images, and supercomputing
power. In recent years, AI has shown great promise to fulfil the challenging need
for image processing. AI can aid in the collecting and processing of image data
using neural networks and deep learning models and through Computer Vision
models to better understand and interpret data more efficiently. Since circa 2016,
we have witnessed the convergence of the improved capacities in geospatial and
AI technology, leading to a new field of Geospatial AI (abbr. GeoAI).
This book explores the convergence of geospatial technology, GeoAI, big data,
and sustainable urban development, providing insights into the nexus of these fields,
the current state, and future potentials. It consists of four sections, each addressing a
unique aspect of these synergies. This book would be of great value for researchers,
students, and professionals in the fields of geospatial technologies, AI, urban
science, and sustainability science.

PART I – ARTIFICIAL INTELLIGENCE AND BIG DATA

ANALYTICS: NEEDS AND REQUIREMENTS
This section covers three chapters discussing the current state of AI and big data
analytics in understanding and managing urban environments. Chapter 1 provides
xiii
xiv Preface

an overview of urban physical background sensing via GeoAI and street-level

imagery, discussing the available datasets, processing methodologies, current
applications, and potential future trajectories of urban remote sensing. Chapter 2
explores using geospatial big data to tackle urban sustainability issues, providing
case studies on agglomeration economies, urban freight logistics, and urban
functionalities. Chapter 3 reviews the recent advancements in geospatial data,
techniques, and platforms dedicated to urban flood monitoring and management. It
covers mitigation, preparation, response, and recovery strategies, underlining the
crucial role of multi-sensor data and integrating geospatial techniques with
advanced AI. This chapter also points out existing challenges and prospects in
the field.

PART II – GEOSPATIAL TECHNIQUES FOR RENEWABLE CITIES

This section explores using geospatial techniques to create sustainable, renewable
urban environments, emphasizing their potential to improve urban living. Chapter 4
investigates the challenges resource-based cities in China face in achieving the
UN’s Sustainable Development Goals (SDGs) and the role of geospatial
techniques’ remote sensing data in assessing environment-related SDG indicators.
The findings suggest that resource-based cities typically struggle more with
environmental challenges and sustainable development than non-resource-based
cities. Chapter 5 presents a GeoAI model that leverages crowdsourced data for
mapping urban green spaces from Sentinel-2A satellite images, demonstrating how
integrating remote sensing with crowdsourced geospatial big data can boost urban
green space mapping. Chapter 6 uses nighttime light time series data to analyze
changes in urban lands across 34 global mega-cities, categorizing them into five
groups: urban growth, deurbanization, urban land use intensification, land use
degradation, and no significant change. This chapter offers new insights into global
urbanization, highlighting the broader extent and greater magnitude of changes in
developing countries’ mega-cities. Chapter 7 studies the impact of nature-based
solutions, such as green roofs, permeable pavements, and street trees, on reducing
urban heat islands, improving thermal comfort, and lessening building energy
consumption. The analysis shows that green roofs can significantly reduce energy
demand in hot, sunny regions and potentially eliminate annual energy demands for
cooling or heating in low-rise urban buildings.

PART III – GEOSPATIAL TECHNIQUES FOR RESILIENT CITIES

This section provides an in-depth overview of how geospatial techniques contribute
to identifying, assessing, and managing various urban risks, encompassing both
natural and anthropogenic threats, towards creating resilient cities. Chapter 8
unveils a two-step GeoAI algorithm for mapping hourly urban air temperature with
a detailed 250 m spatial resolution in Seoul, South Korea. This algorithm addresses
the limitations inherent in satellite imagery and ground-based observation stations,
producing air temperature maps that effectively capture intricate urban temperature
patterns and their temporal variations, aiding in monitoring urban heat island (UHI)
Preface xv

effects. Chapter 9 introduces various satellite land surface temperature (LST)

products, LST retrieval algorithms, and the unique biases associated with urban
surfaces. It further discusses different methods for quantifying UHI and outlines the
contexts where these methods are best employed. Chapter 10 outlines strategies for
mitigating urban air pollution, encompassing observation, modeling, and control
aspects. Chapter 11 assesses emerging drought risks through geospatial analysis,
unveiling underlying mechanisms from a land-atmosphere coupling perspective.
Chapter 12 presents methodologies to enhance urban air temperature forecasting
performance by merging numerical weather prediction (NWP) models with
machine learning, featuring case studies in Seoul. This chapter underscores how
machine learning can augment urban air temperature forecasting accuracy for
practical applications.

PART IV – DIGITAL CITIES

This section emphasizes the transformative potential of geospatial techniques, big
data, and AI in sculpting the future of urban landscapes into digitally driven cities.
Chapter 13 conducts a comprehensive review and analysis of existing datasets and
spatial modeling methods to estimate and map carbon emissions from urban areas.
This effort underpins policy decisions concerning urban energy resilience and
promoting low-carbon development. Chapter 14 formulates visual indicators,
namely greenness, openness, and enclosure, derived from street view data using a
GeoAI-based semantic segmentation approach. Through multilevel regression
analysis, the study investigates the relationship between these indicators and
depression and anxiety levels among the elderly. Chapter 15 delves into applying
deep learning AI models for mapping Local Climate Zones (LCZ), presenting an
innovative method to evaluate urban form across varied landscapes objectively.
This chapter highlights the potential of deploying GeoAI in LCZ mapping, paving
the way for future breakthroughs in urban climate studies. Chapter 16 introduces a
spatial agent-based model developed to simulate the diffusion of private
autonomous vehicles (AVs) across time and space. It investigates their potential
impact on housing expenses saved via residential relocation and in-vehicle work.
The findings from this chapter provide valuable insights for AV-related
stakeholders, including local authorities, urban planners, and vehicle manufacturers.
Qihao Weng and Cheolhee Yoo
Hong Kong
October 20, 2023
Acknowledgments
We wish to thank all the contributors for making this book possible. Furthermore,
we offer our sincere appreciation to all the reviewers, who have taken precious time
from their busy schedules to review the chapters submitted to this book. Further, we
appreciate the members at the JC STEM Lab of Earth Observations and Research
Centre for Artificial Intelligence in Geomatics, The Hong Kong Polytechnic
University, for their support. Last but not least, we acknowledge the financial
support from The Hong Kong Polytechnic University and The Hong Kong Institute
of Surveyors.
We hope that the publication of this book will provide much stimulation to
students, researchers, and practitioners to conduct more in-depth studies on
geospatial technology, GeoAI, urban science, and sustainability science, and will
open up a new way of thinking for years ahead in these communities.
The reviewers of this book are listed below in alphabetical order: Yinxia Cao, Yue
Chang, Dongjin Cho, Siqi Jia, Zhiwei Li, Xiaoyan Lu, Pir Mohammad, Seonyoung
Park, Huijuan Xiao, Cheolhee Yoo, Qiming Zheng, and Qingwei Zhong.

xvi
Editors
Qihao Weng, a Foreign Member of The Academy of Europe (Academia Europaea)
and a Fellow of IEEE, AAAS, AAG, ASPRS, and AAIA, is currently a Chair
Professor at the Hong Kong Polytechnic University and has worked as the Director
of the Center for Urban and Environmental Change and a Professor of Geography
at Indiana State University, 2001–2021, and a Senior Fellow at the National
Aeronautics and Space Administration from 2008 to 2009. He received his Ph.D. in
geography from the University of Georgia in 1999. Weng is the Lead of GEO
Global Urban Observation and Information Initiative, 2012–2022, and an Editor-in-
Chief of ISPRS Journal of Photogrammetry and Remote Sensing. Additionally, he
serves as the Series Editor of Taylor & Francis Series in Remote Sensing
Applications, and Taylor & Francis Series in Imaging Science. Weng has been
the Organizer and Program Committee Chair of the biennial IEEE/ISPRS/GEO
sponsored International Workshop on Earth Observation and Remote Sensing
Applications conference series since 2008; a National Director of American Society
for Photogrammetry and Remote Sensing from 2007 to 2010; and a panelist of U.S.
DOE’s Cool Roofs Roadmap and Strategy in 2010.

Cheolhee Yoo is currently a Research Assistant Professor at the Hong Kong

Polytechnic University. He earned his Bachelor of Engineering and Ph.D. from the
Department of Urban and Environmental Engineering at the Ulsan National
Institute of Science and Technology (UNIST), South Korea, in 2017 and 2022,
respectively. In 2019, he further expanded his research experience as a research
intern at the RIKEN Center for Advanced Intelligence Project (AIP) in Japan.
Following the completion of his doctoral studies, Dr. Yoo served as a Postdoctoral
Fellow at the Hong Kong Polytechnic University from April to September 2022. He
is currently an Editorial Board Member of the ISPRS Journal of Photogrammetry
and Remote Sensing and GIScience & Remote Sensing.

xvii
Contributors
Ren Chao Siwoo Lee
The University of Hong Kong Ulsan National Institute of Science
Hong Kong S.A.R., China and Technology
Ulsan, South Korea
Haojie Chen
The Hong Kong Polytechnic University Yong Li
Hung Hom, Kowloon, Hong Kong The Hong Kong University of Science
and Technology
Yang Chen Hong Kong, China
Wuhan University
Wuhan, China Zhiwei Li
The Hong Kong Polytechnic University
Dongjin Cho Hung Hom, Kowloon, Hong Kong
Ulsan National Institute of Science
and Technology Xintao Liu
Ulsan, South Korea The Hong Kong Polytechnic University
Hung Hom, Kowloon, Hong Kong
Hai Guo
The Hong Kong Polytechnic University Cai Meng
Hung Hom, Kowloon, Hong Kong Wuhan University
Wuhan, China
Ce Hou
The Hong Kong University of Science Yamin Qing
and Technology The Hong Kong Polytechnic University
Hong Kong, China Hung Hom, Kowloon, Hong Kong

Jungho Im Justin Hayse Chiwing G. Tang

Ulsan National Institute of Science The Hong Kong Polytechnic University
and Technology Hung Hom, Kowloon, Hong Kong
Ulsan, South Korea
Shuo Wang
Siqi Jia The Hong Kong Polytechnic University
The Hong Kong Polytechnic University Hung Hom, Kowloon, Hong Kong
Hung Hom, Kowloon, Hong Kong
Yu Wang
Xiaoyu Jiao The Hong Kong Polytechnic University
The Hong Kong Polytechnic University Hung Hom, Kowloon, Hong Kong
Hung Hom, Kowloon, Hong Kong

xviii
Contributors xix

Huijuan Xiao Yangzong Zeren

The Hong Kong Polytechnic University The Hong Kong Polytechnic University
Hung Hom, Kowloon, Hong Kong Hung Hom, Kowloon, Hong Kong

Song Xue Fan Zhang

The Hong Kong Polytechnic University Peking University
Hung Hom, Kowloon, Hong Kong Beijing, China

Yuhei Yamamoto Qiming Zheng

Chiba University The Hong Kong Polytechnic University
Chiba-shi, Japan Hung Hom, Kowloon, Hong Kong

Jiang Ying Chengxiang Zhuge

The Hong Kong Polytechnic University The Hong Kong Polytechnic University
Hung Hom, Kowloon, Hong Kong Hung Hom, Kowloon, Hong Kong

Zidong Yu
The Hong Kong Polytechnic University
Hung Hom, Kowloon, Hong Kong
Part I
Artificial Intelligence and
Big Data Analytics
Needs and Requirements
 1 Sensing Urban Physical
Environment with GeoAI
and Street-Level Imagery
Ce Hou, Yong Li, and Fan Zhang

INTRODUCTION
Sensing the urban physical environment is crucial to understanding, managing,
planning, and designing cities. The paramount significance of the urban physical
environment has been repeatedly emphasized by landscape architects, urban
planners, and researchers. The “City Beautiful Movement”, which flourished
from the 1890s to the 1920s, was a dedicated endeavor aimed at enhancing the
quality of the urban living environment and elevating people’s overall quality of life
through the improvement of cities’ physical surroundings. This movement sought to
augment the aesthetic appeal of urban areas, creating a more visually pleasing and
comfortable perception of the city (Wilson, 1989). In Kevin Lynch’s masterpiece
The Image of the City, he provides a comprehensive exploration of how urban
residents perceive and construct their own perception of the physical environment
within the city (Lynch, 1964). The concept of investigating the human-centered
perception of the urban physical environment has exerted a significant influence on
numerous urban planning theorists, including William H. Whyte, Jane Jacobs, and
Jan Gehl (Jacobs, 2016; Whyte et al., 1980; Gehl, 2013). By sensing the physical
environment of a city, urban planners and designers gain more profound insights
into its distinctive attributes and intricacies. This understanding enables them to
devise effective strategies and solutions aimed at bolstering the sustainability,
livability, and overall quality of life for the city’s inhabitants (Zhang et al., 2019).
In the past few decades, various methodologies for perceiving and analyzing
the physical urban environment have been developed, refined, and employed.
Before the flourishing of information technology, traditional work of sensing the
urban physical environment primarily relied on manual observation methods,
such as questionnaires, census data, and geographic measurements (Hoehner
et al., 2005; Galea et al., 2005; Handy et al., 2002). For instance, urban
researchers physically visited cities, meticulously observed buildings, roads,
and traffic patterns, and collected valuable insights from residents through
questionnaires and interviews (Davison and Lawson, 2006; Frank et al., 2005).
Traditional surveying tools and maps were utilized to gather geographical data
pertaining to the city, including topographical features and architectural structures

DOI: 10.1201/9781003244561-2 3
4 Handbook of Geospatial Approaches to Sustainable Cities

(Harley, 1990; DeMers, 2008). These manual-based urban sensing methods are
often labor-intensive, time and resource-consuming, and sometimes difficult to
implement (Fleischhacker et al., 2013).
The advancement of information technology has led to the emergence of urban
sensing methods utilizing remote sensing (RS), global positioning systems (GPS),
and geographic information systems (GIS), which enables deeper sensing of the
urban physical environment (Liang and Weng, 2010; Miller and Small, 2003). RS
technology provides large-scale imagery data with a high spatio-temporal resolu-
tion, facilitating the acquisition of large-scale information on cities, such as city
sprawl, land use, and vegetation cover (Ji et al., 2006; Alqurashi and Kumar, 2013;
Xie et al., 2008). Moreover, a wide range of sensor data enables real-time sensing of
various physical environmental parameters in cities, including climate conditions,
air quality, and urban noise (Dutta et al., 2009). GPS enables the sensing of the
urban physical environment through the collection of travel tracks from city
dwellers (Van der Spek et al., 2009). By leveraging GPS, large volumes of location
data can be gathered and utilized to analyze parameters such as traffic flow and
travel patterns and to assess challenges such as road congestion and traffic safety
(Kamran and Haas, 2007; Zheng et al., 2009). Subsequently, the integration of
multiple data sources within a GIS platform allows for the overlay analysis of
geographic data and the identification of spatial relationships, leading to a deeper
understanding of the characteristics, challenges, and potential solutions related to
the urban physical environment (Jat et al., 2008; Liang and Weng, 2010). The
emergence of multi-source big data and GIS technologies enables sensing the urban
physical environment from various perspectives and scales, extracting valuable
information and patterns that unveil trends and problematic areas within the urban
landscape. In this context, using street-level Imagery (SI) and geospatial artificial
intelligence (GeoAI) are valuable complements, providing fine-grained spatial
visual information and advanced spatial analysis techniques. These advancements
further enhance urban physical sensing methodologies, enabling scholars to uncover
more comprehensive information previously hidden within the city (Zhang 2023a,
2023b, 2023c).
Notably, utilizing SI and GeoAI to conduct urban sensing has demonstrated
remarkable potential in recent years. In contrast to the aforementioned urban
sensing methods, SI effectively supplements urban analysis data sources, particu-
larly from a human perspective (Zhang et al., 2018; Dubey et al., 2016b). Unlike
satellite imagery and aerial imagery, which capture information from vertical or
oblique viewpoints, SIs provide more detailed visual information at the neighbor-
hood level (shown in Figure 1.1), which is aligned with the human perspective of
observing the urban physical environment (Kang et al., 2020; Biljecki and Ito,
2021). Consequently, SIs enable virtual urban audits for governmental purposes and
assist urban planners in investigating people’s objective perceptions of places,
fostering a better understanding of human-land interaction within the city (Kang
et al., 2023). The rapid development of GeoAI technology further amplifies SI’s
potential for information mining (Zhang et al., 2023a, 2023b, 2023c). By leveraging
SIs and GeoAI, we possess the capacity to not merely evaluate the condition of
urban infrastructure by employing semantic segmentation methodologies (Xiao and
Urban Sensing with GeoAI and Street‐Level Imagery 5

FIGURE 1.1 Examples of (a) satellite imagery, (b) aerial imagery, and (c) street view imagery.

Quan, 2009; Rundle et al., 2011; Li et al., 2015b; Gong et al., 2018), but also to
distill more profound insights concealed within the visual data, including safety,
architectural age and style, as well as the urban socio-economic environment
(Zhang et al., 2021; Sun et al., 2022; Suel et al., 2019). The combination of SI and
GeoAI enhances the existing research paradigm established since the inception of
geographic information science and technology, thus broadening the spectrum of
urban physical environment sensing.
Within the scope of this article, we conduct an overview of prevailing studies
centered around the integration of SI and GeoAI for the purpose of sensing the
urban physical environment, concurrently discussing the applications and achieve-
ments within the discipline of urban science. Furthermore, we have addressed the
shortcomings inherent in current research endeavors and have projected potential
avenues for future exploration.

STREET-LEVEL IMAGERY (SI)

SI refers to the photographs taken along street networks, depicting the profile view
of the urban streetscape from a similar view of human vision and describing the
urban physical environment comprehensively (Zhang et al., 2019, 2023a, 2023b,
2023c). To perceive the urban physical environment using SI, it is important to
collect relevant data that aligns with the research area and apply it to the intended
tasks. However, it’s worth noting that different SI data sources may vary in terms of
data quality (Hou and Biljecki, 2022). Additionally, the sampling method employed
for SI can also impact the performance of the model on specific tasks (Zhang et al.,
2023a, 2023b, 2023c). In this section, we will explore various SI data sources and
discuss the quality aspects associated with them.

STREET-LEVEL IMAGERY SOURCES

Based on the providers, SI can be divided into the following two categories:
Commercial SI is provided by commercial companies or organizations and
is typically only accessible through payment (Anguelov et al., 2010). Some
well-known providers of commercial SI include Google Street View,
6 Handbook of Geospatial Approaches to Sustainable Cities

Bing Maps Streetside, Here Street View, Baidu Street View, AutoNavi, and
Tencent Street View.
One of the main advantages of commercial SI is its high data quality.
Commercial providers invest significant resources in data collection, processing,
and maintenance, resulting in consistent image quality during shooting and
processing. As a result, commercial SI often has higher standards for image
quality, resolution, coverage, and update frequency.
In addition, commercial SI typically has extensive global coverage, including in
rural and urban areas, and across multiple countries. This is due to the greater
resources available to commercial providers, which allows them to ensure that
street-view imagery is available in more areas. As such, commercial SI has become
an invaluable resource for applications such as urban planning, transportation,
tourism, and GIS.
Crowdsourcing SI, such as those provided by Mapillary and OpenStreetView,
is a freely accessible resource contributed and maintained by various entities,
including academic institutions, government departments, and community projects
(Juhasz and Hochmair, 2016). These sources of SI are often supplemented by data
collected from self-driving cars and other methods.
However, the quality of crowdsourced SI may vary depending on the
contributor, including aspects such as image resolution, coverage, and update
frequency. As such, the data quality of crowdsourced SI may not be as consistent
or high-precision as that of commercial SI. Additionally, crowdsourced SI has
limited coverage, especially in remote areas (Yan et al., 2020; Goodchild, 2007;
Goodchild and Li, 2012). Despite these limitations, crowdsourced SI remains a
valuable resource for various applications, particularly in areas where commercial
SI is unavailable or insufficient.

STREET-LEVEL IMAGERY ATTRIBUTES

Here, we examine how various attributes of SI impact its use in different
applications and analyses. SI attributes refer to the characteristics and properties
associated with SI. These attributes provide information about image quality,
coverage, timeliness, and metadata availability. By understanding the impact of
these attributes, researchers can assess the reliability, suitability, and limitations
of street view data for the specific application scenario.
Image quality is one such attribute that assesses the visual quality of SI,
including resolution, sharpness, color accuracy, and absence of distortions or
artifacts (Lauko et al., 2020). Poor image quality issues, such as low-light
conditions, blurriness, and variable weather, can impact the accuracy of object
detection and recognition algorithms. For example, computer vision algorithms
may struggle to correctly identify and classify objects in low-light conditions
or blurry images, leading to lower accuracy in tasks like object detection,
tracking, or scene understanding (Gao et al., 2022). Similarly, the presence of
artifacts or color inaccuracies can affect the performance of machine learning
models trained on SI, leading to reduced accuracy in categorizing and segmenting
different objects and regions in the images (Choi et al., 2022).
Urban Sensing with GeoAI and Street‐Level Imagery 7

Coverage is another important attribute that refers to the spatial distribution

of SI. Unbalanced spatial coverage means that some regions or locations have
more comprehensive SI coverage compared to others. This can lead to insufficient
data for assessing and analyzing specific areas, potentially leading to inaccurate
understanding of the existing environment or infrastructure needs. For example,
the significant gap in SI coverage between Asia and Africa can result in inaccurate
assessments of urban greening and potentially mislead policymakers and re-
searchers (Liang et al., 2023; O’Regan et al., 2022).
Timeliness is an attribute that refers to the temporal coverage of SI. It often
exhibits heterogeneity with sporadic updates, resulting in inconsistent coverage
across different locations. This can introduce biases and inconsistencies in
analyses that involve multiple locations within a city or region. Comparisons
or assessments that assume consistent coverage across all areas may be inaccurate
or incomplete, potentially leading to flawed conclusions or misleading interpre-
tations (Yu et al., 2022).
Metadata availability is an attribute that refers to the extent of metadata provided
with SI. Metadata provides useful information about the properties and character-
istics of SI, and its availability can vary across different platforms and data sources.
The lack of complete metadata in SI can introduce uncertainties, limit data
selection, hinder spatial analysis and visualization, impede time-based analysis,
challenge data quality assessment, and restrict data integration and interaction
capabilities (Vo et al., 2023). Commercial providers, such as Google Street View,
often provide comprehensive metadata, while crowdsourced platforms may have
more limited or inconsistent metadata availability.

METHODS FOR SENSING THE URBAN PHYSICAL ENVIRONMENT

Methods of sensing the urban physical environment through SI involve using
computer vision, deep learning, data analysis, and other technical means to perceive
and understand the city’s physical environment. This method has various applica-
tions in city-related tasks, such as urban planning, traffic management, environ-
mental monitoring, and smart cities.
To categorize perception methods based on their tasks in the context of urban
physical environment perception, we can further partition the underlying techniques
based on their learning approaches.

SENSING TASKS
Sensing the urban physical environment with SI can be categorized into the
following tasks (Table 1.1).

Scene Element Extraction

Scene element extraction is the most direct and common way to use street-level
images in representing built environments. Scene elements can be derived either
by object detection DCNN models, such as SSD (Liu et al., 2016) and
 8
TABLE 1.1
Approaches Based on Perception Tasks
Category Work Method
Element extraction Object detection

Handbook of Geospatial Approaches to Sustainable Cities

Object segmentation
Relation detection

Perception Scene category

Perception model

Inference En2End learning

Crime rates
House prices
Human dynamics
Embedding Representation learning
Image deep feature

Generation Scene generation Scene

reconstruction
Urban Sensing with GeoAI and Street‐Level Imagery 9

YOLO (Bochkovskiy et al., 2020), or object segmentation models, e.g., PSPNet

(Zhao et al., 2017), Mask RCNN (He et al., 2017), and HRNet (Wang et al., 2020).
The former models output the detected objects with bounding boxes, so the number
of different objects in a scene can be counted. The latter model predicts the object
categories of every pixel in an image, which can be further calculated as the object
proportions of a scene. Both models provide an approach to measuring a scene with
quantitative metrics.

Scene Perception
Scene perception is based on typical image classification models. Scene perception
models are mainly based on the Place Pulse 2.0 database (Dubey et al., 2016a),
which contains one million human ratings on street-level images along six
dimensions: “safe”, “lively”, “beautiful”, “wealthy”, “boring”, and “depressing”.
The models trained on Place Pulse have been proven to be effective in estimating
these perceptual dimensions of new scenes in most cities.

Scene Inference
Beyond information from visible scenes, scene inference models aim at inferring
the scene information that cannot be directly observed from the visible scenes, such
as crime (Khosla et al., 2014), real estate values (Law et al., 2019), and long-term
human dynamics (Zhang et al., 2019), etc. These kinds of models are trained
through an end-to-end learning process, in which the model automatically learns
efficient features between the initial input image and the final output labels (crime
rate, house price value, etc.), without manually indicating what visual cues matter to
the model.

Scene Embedding
Scene embedding aims to learn and represent a scene with a generic feature, which
is typically a vector of continuous numbers that capture the semantics of a scene. As
a computational representation, this vector makes scenes computable, which can be
used either to measure similarity among scenes or as the input of other machine
learning tasks (Ye et al., 2020; Zhu et al., 2020a). In practice, Zhou et al. (2017)
proposed to use the last activation layer from the trained model by Places dataset as
a generic scene embedding, also termed deep features.

Scene Generation
Scene generation is enabled by a special architecture of DCNN called Generative
Adversarial Nets (GAN) (Goodfellow et al., 2014), which can generate entirely new
predicted images after learning what realistic scenes look like from hundreds of
thousands of real street scenes. GAN creates computer-generated urban scenes
based on user-generated inputs, such as objective characteristics extracted from
images (e.g., buildings, roads, and vehicles) or encoded perceptions (e.g.,
attractiveness, safety, and liveliness of a place). Urban scenes can be also generated
and edited based on some attributes of the built environment (Bau et al., 2020; Zhu
et al., 2020b; Richter et al., 2022). These types of models are useful for scenario
planning and urban design applications which frequently require that participants
10 Handbook of Geospatial Approaches to Sustainable Cities

envision images of cities that don’t exist. Urban designers are also using GAN to
create vivid high-resolution imagery. A pioneering approach looking at this was
introduced by Noyman and Larson (2020), who designed a physical platform that
allows users to generate street scenes by combining a wide range of street elements
based on their preferences, including different land uses, types of roads, density of
buildings, and presence of sidewalks.

APPROACHES BASED ON TECHNIQUES

To gain a comprehensive understanding of the methods used in perceiving the urban
physical environment using SI, we can categorize the models into the following
categories based on their learning approaches (Figure 1.2).

Supervised Learning
Supervised learning methods use labeled training data to train a model. In SI urban
physical environment perception, it is possible to use labeled SI (Ye et al., 2020;
Zhu et al., 2020a), point cloud data (Micusik and Kosecka, 2009), or other
annotated datasets, these methods enable the training of models to perform various

FIGURE 1.2 Approaches based on techniques. The supervised learning model utilizes
known street-level imagery along with its corresponding labels during the training phase. The
learning algorithm generates an inference function, which is eventually used to predict
unseen street-level imagery. Unsupervised learning describes hidden structures from
unlabeled data. In the context of street views, it involves learning to partition or classify
street view images into different groups or categories without any prior labels or annotations.
Semi-supervised models use both labeled and unlabeled data for training. Reinforcement
learning is when an agent interacts with an environment to learn and improve its decision-
making. The agent takes action in the environment and receives feedback through rewards or
penalties. The environment represents the system or context in which the agent operates,
providing states and dynamics for interaction. The agent’s goal is to learn the best actions to
maximize cumulative rewards over time in different environmental states.
Urban Sensing with GeoAI and Street‐Level Imagery 11

tasks, including scene classification, object detection, semantic segmentation, and

3D reconstruction.

Unsupervised Learning
Unsupervised learning, including self-supervised learning, is an important approach
to perceiving the urban physical environment using SI. Unlike supervised learning
that relies on labeled data, unsupervised learning methods utilize unlabeled data for
training models and aim to discover hidden structures and patterns in the data
(Wang and Biljecki, 2022). In the context of SI urban physical environment
perception, unsupervised learning methods can be applied to various tasks:
Clustering: Unsupervised learning algorithms can group similar SI features
together based on their intrinsic properties. Clustering can help identify common
patterns or distinct groups within the data, which can be useful for understanding
the diversity of urban environments or discovering specific types of scenes or
objects (Ke et al., 2022).
Dimensionality reduction: Unsupervised learning techniques such as principal
component analysis (PCA) (Yuan et al., 2023) or autoencoders (Kamal and Bae,
2022) can reduce the dimensionality of the street view data while retaining
important information. This can help visualize and analyze the data in a lower-
dimensional space, facilitate data exploration, and improve the efficiency of
subsequent processing tasks.
Anomaly detection: Unsupervised learning methods can identify anomalous or
outlier SIs that deviate from the expected patterns. This can be valuable for
detecting unusual events, identifying infrastructure issues, or discovering
uncommon urban conditions that require further investigation (Zhao et al., 2022).
Representation learning: Unsupervised learning can also be used to learn
meaningful representations or embeddings from unlabeled street view data. By
training models to capture the underlying structure of the data, these representa-
tions can be leveraged for subsequent tasks such as similarity comparison
(Mohamed et al., 2022), retrieval (Lin et al., 2022), or contrastive learning
(Liu et al., 2023).

Semi-Supervised Learning
Semi-Supervised learning methods play a significant role in SI and physical
environment perception by leveraging labeled and unlabeled training data. These
methods can effectively utilize limited labeled data in conjunction with a large
amount of unlabeled data to enhance model performance and generalization
capabilities (Oliveira et al., 2022; Yang et al., 2022). In the context of SI physical
environment perception, semi-supervised learning methods can be applied in
various ways:
Pseudo-labeling: Initially, a small portion of the SI data is manually labeled
with high-quality annotations. These labeled samples are then used to train a
model. The trained model can generate pseudo-labels for the remaining unlabeled
data based on its predictions. The unlabeled data, along with their pseudo-labels,
are then incorporated into the training process to refine the model iteratively
(Li and Ma, 2022).
12 Handbook of Geospatial Approaches to Sustainable Cities

Consistency regularization: Semi-supervised learning methods often involve the

use of consistency regularization techniques. These techniques aim to ensure that
similar inputs produce similar outputs, even when small perturbations are applied to
the input data. By enforcing consistency across different augmentations or views of
the same street view image, the model can learn robust representations and improve
its ability to generalize to unseen data (Bortoloti et al., 2022).

Reinforcement Learning
Reinforcement learning methods learn the optimal behavioral strategy through
the interaction between the agent and the environment. In SI urban physical
environment perception, reinforcement learning methods can be applied to
intelligent navigation, traffic control, and other scenarios to optimize urban
environment perception and decision-making process (Wang et al., 2022b;
Grabaskas and Wang, 2022).

APPLICATIONS OF URBAN SENSING

According to the depth of urban physical environment sensing, the application can
be divided into three levels hierarchically (Figure 1.3). Observational sensing

FIGURE 1.3 An overview of the application of urban sensing based on GeoAI and SI.
Based on the depth of urban physical environment sensing, the application can be
hierarchically stratified into three levels. Each sensing level offers a distinct perspective
and an understanding of the urban physical environment. The first level, observational
sensing, primarily centers on the physical entities within urban settings. The second level,
urban implicit relationship sensing, delves into the intangible conceptions that extend beyond
mere physical entities. Lastly, human perception sensing is devoted to capturing and
analyzing human responses and perceptions of the environment.
Urban Sensing with GeoAI and Street‐Level Imagery 13

focuses on the physical entities described in the cities; urban implicit relationship
perception aims to capture the more abstract information hidden in the images, such
as urban socio-economic environments and urban culture; and human perception
sensing emphasizes the emotional experience when urban dwellers participate in the
places and the difference between the objective measure and subjective human
perception.

OBSERVATIONAL SENSING: SENSING THE PHYSICAL ENTITY

Observational sensing primarily encompasses the detection of tangible entities
present within the SIs, such as buildings, plants, traffic lights, and other elements
scattered in the city. Given that SIs offer extensive and meticulous data regarding
the urban physical environment, encompassing virtually all places within the city,
GeoAI models demonstrate the capability to effectively acquire information ranging
from individual units to the entirety of the urban landscape through semantic
segmentation, classification, and other feature extraction techniques.

Instance-Level Sensing
The combined approach of SI and GeoAI is proving invaluable for municipal
engineering by facilitating automated monitoring and streamlining of environmental
audits for various aspects of urban infrastructure (Xiao and Quan, 2009; Kang et al.,
2018). This virtual auditing capability enables efficient analysis and evaluation of
diverse environmental factors, such as urban infrastructure and green spaces, among
other critical elements that contribute to the overall built environment.
One of the key advantages of this approach is the ability to supplement
traditional maps with fine-grained details often overlooked. Through semantic
segmentation of street images, elements such as traffic lights (Campbell et al.,
2019), sideroads (Ning et al., 2022), poles (Ao, 2019; Krylov et al., 2018), traffic
signs (Balali et al., 2015), and zebra crossings (Ahmetovic et al., 2015) can be
accurately identified and incorporated into urban data. This supplementation
enhances GIS and drives the development of three-dimensional GIS and digital
twins (Li and Yao, 2020; Lehtola et al., 2022; Pang and Biljecki, 2022).
Additionally, SIs and GeoAI can be employed to monitor the status of urban
infrastructure to assist with construction and transportation in cities. In this context,
SI serves as a readily accessible data source, enabling numerous studies to extract
relevant information regarding building energy consumption using deep learning
models. These models utilize SIs to infer various factors, including building age
(Sun et al., 2021), window-to-wall ratios (Szczesniak et al., 2022), and image-based
energy efficiency estimation (Mayer et al., 2023). In addition, SIs aid in road quality
monitoring, building damage detection (He et al., 2022), and landslide detection
(Voumard et al., 2017). They also enable the observation of pedestrian and vehicle
flow on roads, providing valuable insights for traffic planning (Chen et al., 2020b;
Yin et al., 2015). The strategic utilization of the extensive data aggregated by SIs
considerably enhances the protection and surveillance of urban construction and
transportation systems and fosters the facilitation of sustainable urban development
(Hak et al., 2016).
14 Handbook of Geospatial Approaches to Sustainable Cities

Street-Level Sensing
Streets, as a vital component of urban morphology, serve as the primary interface
for social interactions between urban residents and the built environment (Li and
Ratti, 2019). The layout of streets has a complex impact on the urban climate,
encompassing factors such as airflow patterns, solar radiation variations, and
precipitation distribution within the city. These factors, in turn, have indirect
consequences for human comfort, ambient air quality, and even the broader climate
cycle (Masson et al., 2020; Shi et al., 2016; Carrasco-Hernandez et al., 2015).
The utilization of SI in combination with GeoAI methods enables the modeling
of key indicators of street metrics, such as sky view factor (SVF) (Xia et al., 2021),
solar radiation (Gong et al., 2019), and light pollution (Li et al., 2021). These
indicators serve a wide range of purposes, ranging from microclimate studies to
guiding urban planning decisions. By utilizing these indicators, cities can optimize
their resource allocation and promote more efficient use of resources. For instance,
Deng et al. (2021) employ semantic segmentation of SIs to assess the potential for
collecting street-level solar radiation, thus offering promising prospects for
improving the efficiency of solar radiation utilization. Furthermore, Li et al.
(2019) utilize image segmentation techniques to detect buildings that may block
sunlight, enabling the prediction of locations prone to sun glare on the streets and
facilitating better guidance for road travel. Due to the important role of greenery in
urban street spaces, a large number of articles focus on the use of GeoAI methods to
generate an indicator to measure urban greenery and its visibility in the city (which
is called green view index, GVI) from SIs (Yang et al., 2009; Li et al., 2015b; Cai
et al., 2018). The impact of urban greenery extends to considerations of thermal
comfort, as green spaces can provide shading and cooling effects, improving the
microclimate and mitigating heat-related issues in urban environments (Li et al.,
2018b). Moreover, the inclusion of green spaces is recognized as an essential
component of sustainable urban development, with studies exploring the relation-
ship between greenery and urban growth, livability, and ecological balance (Chen
et al., 2020a).
The combination of SIs and GeoAI can be effectively integrated into spatial
syntax analysis frameworks (which is a quantitative method for perceiving the
interaction of the urban spatial environment) (Abusaada and Elshater, 2021;
Hidalgo and Hernandez, 2001), either by expanding the scope of investigation
from to objective spatial configuration to subjective urban sensing, or by modeling
key parameters (e.g., enclosure, connectivity, accessibility, visibility) within the
context of space syntax (Jiang et al., 2000; van Nes et al., 2021). Other studies
explore the re-assessment of enclosure (Yin and Wang, 2016), employ isovist
analysis (Takizawa and Kinugawa, 2020; Xiang et al., 2021), and investigate the
relationship between human perception of street accessibility (Wang et al., 2022a)
to further understand the relationship between visual perception and spatial syntax.
The utilization of SIs and GeoAI provides valuable insights for street metrics,
empowering informed decision-making in streets such as urban planning, energy
efficiency, and resource management; and integrating SIs and GeoAI into spatial
syntactic analysis enhances our understanding of the interaction between subjective
Urban Sensing with GeoAI and Street‐Level Imagery 15

perception and objective spatial configuration, leading to a more comprehensive

assessment of urban physical environments and the potential for improved urban
design, planning, and user experience.

Neighborhood-Level Sensing
When we focus our perspective on the neighborhood level, the integration of SI and
GeoAI enables us to observe the harmonious atmosphere in neighborhoods and the
interaction between humans and their surroundings. This capability proves highly
beneficial in the context of public health. Extensive research has consistently
demonstrated a significant relationship between the built environment and the
physical and mental well-being of community residents. Communities that boast
well-maintained surroundings and abundant green spaces tend to foster greater
outdoor activities and promote better mental health among their residents (Barton
and Pretty, 2010). In contrast, communities characterized by disorderly environ-
ments and inadequate infrastructure are often associated with the spread of
epidemics and higher rates of psychological depression among local residents
(Wilson and Kelling, 2015; Sampson and Raudenbush, 1999).
Scientists can employ semantic segmentation techniques to identify various
elements within neighborhoods and evaluate their influence on public health
issues (Rzotkiewicz et al., 2018; Kang et al., 2020; Chen et al., 2023). The
analysis of urban greenery is involved across different scales of urban sensing.
Research at the neighborhood level focuses on the relationship between green
spaces and neighborhood engagement in physical exercise, highlighting the
positive influence of greenery in promoting physical activity and healthy
lifestyles (Ki and Lee, 2021). The mental health benefits of urban green spaces
have also been examined, emphasizing their role in reducing stress, improving
mood, and enhancing overall mental well-being (Soga et al., 2021). Furthermore,
the presence of greenery has been studied in relation to enhanced safety
perceptions and reduced crime rates in neighborhoods (Li et al., 2015a).
Additionally, numerous studies have investigated specific and detailed informa-
tion relevant to public health within neighborhoods, such as the distribution of
retail alcohol stores (Less et al., 2015), the presence of no smoking signs around
schools and hospitals (Wilson and Thomson, 2015), and the evaluation of
community accessibility levels (Seekins et al., 2014). Research in this field also
extends to transportation safety, including the analysis of road traffic accidents
and pedestrian injuries (Johnson and Gabler, 2015; Mooney et al., 2016).
Another prominent area of research in neighborhood-level investigations
revolves around public participation in physical exercises. The integration of SIs
and GeoAI provides a human-centered perspective on the built environment,
thereby highlighting the importance of walkability or bikeability for individuals
(Bartzokas-Tsiompras et al., 2020; Nagata et al., 2020; Chow et al., 2014; Zhou
et al., 2019). SIs and GeoAI also play a crucial role in measuring the physical and
mental health of young and elderly individuals in urban settings (Wang et al., 2019;
Li et al., 2018a). Furthermore, they assist in the detection of environmental factors
associated with healthy or unhealthy lifestyles, such as obesity, alcohol abuse, and
16 Handbook of Geospatial Approaches to Sustainable Cities

diabetes, among residents (Clews et al., 2016; Yang et al., 2020; Egli et al., 2019).
SIs provide valuable insights into the community places frequently visited by
younger and older individuals, shedding light on the development of strategies for
building healthier cities and promoting active lifestyles (Kang et al., 2020).

City-Level Sensing
We can also conduct a comprehensive analysis encompassing the entire city due to
the extensive accessibility of SIs within the urban area. By integrating RS imagery
and SIs, GeoAI methodologies are capable of producing multi-perspective
portrayals of an identical urban physical environment, thereby augmenting the
performance of urban sensing (Najmi et al., 2022; Zou and Wang, 2022). City-level
urban sensing proves invaluable for examining places characterized by “unusual”
conditions on a large scale, which is meaningful for improving the overall quality of
life for urban dwellers and reducing socio-economic disparities. By identifying
urban slums (Najmi et al., 2022; Chen et al., 2022; Fan et al., 2023) and areas with
abundant housing (Zou and Wang, 2022), we can detect the physical segregation
between the affluent and the underprivileged in the city as early as possible,
enabling city managers to proactively address this issue. Moreover, this approach
serves the purpose of uncovering overlooked locales within urban settings (Zhang
et al., 2020a). The identification of so-called “ghost cities” also contributes to
understanding inter-city interactions, which is helpful in guiding broader urban
development strategies (Jin et al., 2017).
By leveraging SIs and GeoAI, we are empowered to assess the vulnerability of
cities in the face of natural disasters, such as flooding and hurricanes (Xing et al.,
2023). The intricate link between the spread of infectious diseases and the
community environment is widely recognized (Perdue et al., 2003). Through
comparing SIs at different time stamps, it is also possible to evaluate the urban
regeneration process (Thackway et al., 2023) and guide the process of post-disaster
recovery (Meyer and Hendricks, 2018). The integration of SIs and GeoAI offers a
synergistic approach enabling environmental monitoring to identify potential
factors contributing to disease transmission. Furthermore, this approach facilitates
the inference of potential transmission trajectories based on the spatial distribution
of infectious disease outbreaks (Nguyen et al., 2020; Delicado and Rowland, 2021;
Haddawy et al., 2019).

URBAN IMPLICIT RELATIONSHIP SENSING: SENSING BEYOND THE IMAGE

The information provided by SIs extends far beyond the visual representation itself.
The application of GeoAI enables us to uncover deeper information pertaining to
the intangible characteristics of cities, including their socio-economic environment,
urban function, and cultural aspects.
Numerous studies have demonstrated the link between the physical environment
and the socio-economic conditions of communities. For instance, neighborhoods
characterized by cleanliness and visual unity often indicate higher socio-economic
status, while areas with higher levels of graffiti and disarray tend to be associated
with poorer socio-economic conditions. By combining SI and GeoAI, we can
Urban Sensing with GeoAI and Street‐Level Imagery 17

uncover hidden socio-economic indicators and gain a deeper understanding of the

socio-economic environment of a city, in addition to analyzing the physical
environment depicted in SIs (Arietta et al., 2014). SIs offer a novel perspective
for measuring the socio-economic composition of cities. For example, Gebru et al.
(2017) utilized SIs and GeoAI to measure the demographic composition of the
United States. Additionally, Arietta et al. (2014) proposed a GeoAI method to
extract predictive relationships between the visual appearance of cities captured in
SIs and non-physical attributes such as crime rates, house prices, and population
density. The topic of social inequality is also a significant focus in urban research.
Through the analysis of SIs, researchers have identified various forms of inequality,
including disparities in green space exposure (Wang et al., 2021b), walkability
(Zhou et al., 2019), health (Suel et al., 2019), environment (Wang et al., 2021a), and
racial disparities (Hwang and Sampson, 2014). Furthermore, the integration of SI
and GeoAI enables a reconsideration of phenomena such as social segregation and
gentrification from a new perspective (Ilic et al., 2019; Thackway et al., 2023; Fan
et al., 2023).
The discrimination of urban functions is also one of the primary applications
of SI and GeoAI. Many studies have focused on discriminating urban points of
interest (POIs) and analyzing land use and cover change (LUCC) (Fan et al.,
2022; Fang et al., 2021; Hu et al., 2020; Ye et al., 2021; Cao et al., 2018).
Specifically, SIs provide the ability to capture the changes and details in POI and
land use analysis with their rich visual information (Fang et al., 2021). Moreover,
the combination of Street View and GeoAI enables the exploration of various
social phenomena, including commuting patterns, work-life balance, and crime
(Zhang et al., 2019; Yao et al., 2021; Khorshidi et al., 2021). It has also been
utilized to enhance the accuracy of real estate value estimates by incorporating
neighborhood appearance into the analysis, a previously underexplored factor
(Law et al., 2019; Kang et al., 2021). By utilizing SI and GeoAI, researchers can
gain insights into urban functions, analyze spatial dynamics, and explore various
social phenomena related to urban environments.
Even though only a limited number of articles specifically focus on urban
culture, the integration of SI and GeoAI facilitates the utilization of sensing
technology to extract valuable insights about urban culture. The style of buildings,
for instance, can be easily identified from SIs (Sun et al., 2022; Zhang et al., 2020b),
and landmarks that represent the unique identity of cities can be detected through
the analysis of multiple SIs (Lander et al., 2017). These visual distinctions allow us
to perceive the humanistic accumulation of cities. The diverse civilizations that
have shaped different cities are reflected in their architectural styles and ultimately
captured in the streetscape.
By harnessing SI and GeoAI, researchers can enhance their understanding of the
socio-economic environment of cities, uncover patterns of social inequality, and
investigate phenomena such as social segregation and gentrification, enabling
evidence-based policy-making, urban planning, and interventions to address socio-
economic disparities and promote equitable and inclusive urban environments.
Moreover, the utilization of SI and GeoAI empowers researchers to explore urban
functions, analyze spatial dynamics, and investigate various social phenomena,
18 Handbook of Geospatial Approaches to Sustainable Cities

facilitating informed decision-making in urban planning, land management, and

policy development, thereby promoting efficient resource allocation and the
creation of sustainable and livable cities. The integration of SI and GeoAI also
enables the extraction of valuable insights about urban culture, reflecting the diverse
civilizations that have shaped them.

HUMAN PERCEPTION SENSING: SENSING WITH HUMAN-CENTRIC PERSPECTIVE

Understanding how people perceive streets is crucial for urbanists aiming to create
more livable and thriving urban environments. Streets that are perceived as
comfortable and safe foster emotional attachment among residents, leading to
improved mental and physical well-being (Hidalgo and Hernandez, 2001).
SIs, captured from a human perspective, provide a valuable opportunity for
researchers to model urban perceptions and support human-centered urban planning
(Kang et al., 2020). One notable project in this realm is the “Place Pulse” project
developed by MIT’s Senseable City Lab. This initiative invites individuals to rate
the safety of street images from around the world, resulting in a collection of over
one million street images with associated safety ratings. These data have been used
to train GeoAI models to accurately predict perceptions of safety (Dubey et al.,
2016b; Naik et al., 2014). The outcomes of this study have inspired numerous urban
researchers and facilitated various related investigations (Naik et al., 2017; Zhang
et al., 2018; Li et al., 2015a).
Furthermore, studies have focused on uncovering people’s emotional perception
of community environments and roads, examining the emotions evoked by the built
environment (Zhang et al., 2022; Ye et al., 2019; Xiang et al., 2021). Researchers
have also explored simulating residents’ subjective perceptions of the urban
environment using visual information (Verma et al., 2020).
It is worth noting that the objective perception of urban phenomena may not
always align with the subjective perception. For instance, an environment that appears
safe may not necessarily be safe in reality, and vice versa. By combining SIs with
questionnaires, researchers can compare these differences, known as perception
biases (Zhang et al., 2021). For example, Kang et al. (2023) conducted a study in
Stockholm, examining residents’ perceptions of safety on streets using GeoAI and
questionnaires concurrently. The research explores the discrepancies between
modeling results and residents’ subjective perceptions, utilizing the GeoAI approach.
Apart from the diverse applications described earlier, SIs and GeoAI can be
applied to augmented reality (AR) and virtual reality (VR) to augment human
sensing for the built environment. One such application is the utilization of SIs to
generate virtual environments within AR or VR platforms. This approach enhances
the realism of the virtual environment, thereby improving users’ perception and
experience of the virtual built environment (Kostakos et al., 2019; Johnson, 2019).
This technology can be valuable for architectural and urban design, allowing
designers and planners to visualize and explore different design options and their
potential impact on the built environment.
Through the integration of SIs, GeoAI, and questionnaire-based assessments,
urbanists gain insights into the subjective perceptions of streets and urban
Urban Sensing with GeoAI and Street‐Level Imagery 19

environments. This knowledge helps bridge the gap between objective and
subjective perceptions, informing urban planning decisions and interventions to
create safer, more enjoyable, and people-centric cities.

FUTURE TRENDS
We currently reside in an era where sensing is ubiquitous, with people’s approach to
sensing the world expanding rapidly from the utilization of our natural sense organs to
various smart sensing devices. SI effectively broadens the fine-grained urban sensing
methods at the fundamental level of the urban environment. Simultaneously, we find
ourselves amidst an era dominated by the advancements of artificial intelligence (AI),
wherein diverse AI models are employed to accomplish various objectives with
exceptional performance. This newfound efficacy enables human beings to undertake
tasks previously deemed inconceivable. The remarkable achievements attained by
large-sized deep learning models, such as ChatGPT and Segment Anything Model
(SAM), demonstrate the immense potential of AI in accomplishing multipurpose
tasks and analyzing multi-modal data. These models surpass the capabilities of most
pre-existing models, thereby reshaping the research paradigm of deep learning.
Sensor technology and AI systems, distinguished by SI and GeoAI, continue to offer
the profound potential for advancements in every step of urban studies, including the
detecting, handling, and interpreting of urban challenges.
First, the innovation of SIs and GeoAI can provide a more profound and
comprehensible interpretation of the underlying principles behind urban issues.
Future studies should place greater emphasis on examining the augmentation effect
of SIs through the integration of other geospatial data sources, such as social media
and GPS trajectories. By harnessing a diverse range of multi-modal data, we can
expand our urban sensing capabilities and render previously abstract information in
SIs visible. From a public management perspective, it is paramount to understand
the rationale behind AI-based decision-making rather than solely focusing on the
outcomes. Incorporating interpretable AI techniques enables effective analysis of
the reasoning behind the approaches taken by AI models in addressing urban
problems, as well as understanding the disparities between AI and human policy-
making. Consequently, this fosters policymakers’ awareness of previously over-
looked phenomena.
Furthermore, harnessing the power of SI and GeoAI offers a wide array of
diverse and precise methodologies to enrich urban studies. Artificial Intelligence-
Generated Content (AIGC) empowers us to generate practical solutions tailored to
address specific problems in a variety of scenarios. Moreover, it facilitates the
creation of urban designs that are customized to align with the expectations and
objectives of urban development set forth by governments or policymakers. By
leveraging SI and GeoAI, we can introduce a range of alternative approaches to
urban planning and design processes, thereby yielding more targeted and effective
solutions for the challenges faced in urban development.
Finally, the promising prospect of artificial general intelligence (AGI), which
possesses heightened intelligence and extraordinary information-gathering capabil-
ities, holds the potential to serve as a powerful agent for the autonomous
20 Handbook of Geospatial Approaches to Sustainable Cities

identification and resolution of urban issues by human urban governments. This

further augments the overall efficiency and effectiveness of urban management.
Additionally, AGI can play a crucial role in unveiling the universal laws underlying
complex urban phenomena through data mining and analysis. By leveraging AGI’s
capabilities, we can uncover hidden patterns and causal relationships that contribute
to a deeper understanding of urban dynamics.
Above all, these advancements in SI, and GeoAI hold significant potential to
revolutionize our understanding and approach to urban challenges. By leveraging
these technologies, we can delve deeper into the intricate dynamics of cities,
promote evidence-based decision-making, and foster sustainable and livable urban
environments.

REFERENCES
Abusaada, H. and Elshater, A. (2021). Effect of people on placemaking and affective
atmospheres in city streets. Ain Shams Engineering Journal, 12(3):3389–3403.
Ahmetovic, D., Manduchi, R., Coughlan, J. M., and Mascetti, S. (2015). Zebra crossing
spotter: Automatic population of spatial databases for increased safety of blind
travelers. In Proceedings of the 17th International ACM SIGACCESS Conference
on Computers & Accessibility, pages 251–258.
Alqurashi, A. and Kumar, L. (2013). Investigating the use of remote sensing and GIS
techniques to detect land use and land cover change: A review. Advances in Remote
Sensing, 2: 193–204.
Anguelov, D., Dulong, C., Filip, D., Frueh, C., Lafon, S., Lyon, R., Ogale, A., Vincent, L.,
and Weaver, J. (2010). Google street view: Capturing the world at street level.
Computer, 43(6):32–38.
Ao, Y. (2019). Fully Convolutional Networks for Street Furniture Identification in Panorama
Images. Master’s thesis, University of Twente.
Arietta, S. M., Efros, A. A., Ramamoorthi, R., and Agrawala, M. (2014). City forensics:
Using visual elements to predict non-visual city attributes. IEEE Transactions on
Visualization and Computer Graphics, 20(12):2624–2633.
Balali, V., Ashouri Rad, A., and Golparvar-Fard, M. (2015). Detection, classification, and
mapping of US traffic signs using Google street view images for roadway inventory
management. Visualization in Engineering, 3:1–18.
Barton, J. and Pretty, J. (2010). What is the best dose of nature and green exercise
for improving mental health? a multi-study analysis. Environmental Science &
Technology, 44(10):3947–3955.
Bartzokas-Tsiompras, A., Tampouraki, E. M., and Photis, Y. N. (2020). Is walkability
equally distributed among downtowners? evaluating the pedestrian streetscapes of
eight European capitals using a micro-scale audit approach. City Sustainability and
Regeneration, 4, page 135.
Bau, D., Zhu, J.-Y., Strobelt, H., Lapedriza, A., Zhou, B., and Torralba, A. (2020).
Understanding the role of individual units in a deep neural network. Proceedings of the
National Academy of Sciences, 117(48):30071–30078.
Biljecki, F. and Ito, K. (2021). Street view imagery in urban analytics and GIS: A review.
Landscape and Urban Planning, 215:104217.
Bochkovskiy, A., Wang, C.-Y., and Liao, H.-Y. M. (2020). Yolov4: Optimal speed and
accuracy of object detection. arXiv preprint arXiv:2004.10934.
Bortoloti, F. D., Tavares, J., Rauber, T. W., Ciarelli, P. M., and Botelho, R. C. G. (2022). An
annotated image database of building facades categorized into land uses for object
Urban Sensing with GeoAI and Street‐Level Imagery 21

detection using deep learning: Case study for the city of Vila Velha-ES, Brazil.
Machine Vision and Applications, 33(5):80.
Cai, B. Y., Li, X., Seiferling, I., and Ratti, C. (2018). Treepedia 2.0: applying deep learning
for large-scale quantification of urban tree cover. In 2018 IEEE International Congress
on Big Data, pages 49–56. IEEE.
Campbell, A., Both, A., and Sun, Q. C. (2019). Detecting and mapping traffic signs from
Google street view images using deep learning and GIS. Computers, Environment and
Urban Systems, 77:101350.
Cao, R., Zhu, J., Tu, W., Li, Q., Cao, J., Liu, B., Zhang, Q., and Qiu, G. (2018). Integrating
aerial and street view images for urban land use classification. Remote Sensing,
10(10):1553.
Carrasco-Hernandez, R., Smedley, A. R., and Webb, A. R. (2015). Using urban canyon
geometries obtained from Google street view for atmospheric studies: Potential
applications in the calculation of street level total shortwave irradiances. Energy and
Buildings, 86:340–348.
Chen, B., Feng, Q., Niu, B., Yan, F., Gao, B., Yang, J., Gong, J., and Liu, J. (2022).
Multimodal fusion of satellite and street-view images for urban village classification
based on a dual-branch deep neural network. International Journal of Applied Earth
Observation and Geoinformation, 109:102794.
Chen, J., Chen, L., Li, Y., Zhang, W., and Long, Y. (2023). Measuring physical disorder in
urban street spaces: A large-scale analysis using street view images and deep learning.
Annals of the American Association of Geographers, 113(2):469–487.
Chen, J., Zhou, C., and Li, F. (2020a). Quantifying the green view indicator for assessing
urban greening quality: An analysis based on internet-crawling street view data.
Ecological Indicators, 113:106192.
Chen, L., Lu, Y., Sheng, Q., Ye, Y., Wang, R., and Liu, Y. (2020b). Estimating pedestrian
volume using street view images: A large-scale validation test. Computers,
Environment and Urban Systems, 81:101481.
Choi, K., Lim, W., Chang, B., Jeong, J., Kim, I., Park, C.-R., and Ko, D. W. (2022). An
automatic approach for tree species detection and profile estimation of urban street
trees using deep learning and Google street view images. ISPRS Journal of
Photogrammetry and Remote Sensing, 190:165–180.
Chow, C. K., Corsi, D. J., Lock, K., Madhavan, M., Mackie, P., Li, W., Yi, S., Wang, Y.,
Swaminathan, S., Lopez-Jaramillo, P., et al. (2014). A novel method to evaluate
the community built environment using photographs–environmental profile of a
community health (epoch) photo neighbourhood evaluation tool. PLoS One, 9(11):
e110042.
Clews, C., Brajkovich-Payne, R., Dwight, E., Ahmad Fauzul, A., Burton, M., Carleton, O.,
Cook, J., Deroles, C., Faulkner, R., Furniss, M., et al. (2016). Alcohol in urban
streetscapes: A comparison of the use of Google street view and on-street observation.
BMC Public Health, 16:1–8.
Davison, K. K. and Lawson, C. T. (2006). Do attributes in the physical environment
influence children’s physical activity? A review of the literature. International Journal
of Behavioral Nutrition and Physical Activity, 3(1):1–17.
Delicado, A. and Rowland, J. (2021). Visual representations of science in a pandemic: Covid-
19 in images. Frontiers in Communication, 6:645725.
DeMers, M. N. (2008). Fundamentals of Geographic Information Systems. John Wiley &
Sons.
Deng, M., Yang, W., Chen, C., Wu, Z., Liu, Y., and Xiang, C. (2021). Street-level solar
radiation mapping and patterns profiling using Baidu street view images. Sustainable
Cities and Society, 75:103289.
22 Handbook of Geospatial Approaches to Sustainable Cities

Dubey, A., Naik, N., Parikh, D., Raskar, R., and Hidalgo, C. A. (2016a). Deep learning the
city: Quantifying urban perception at a global scale, In Computer Vision–ECCV 2016:
14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016,
Proceedings, Part I 14 (pp. 196–212).
Dubey, A., Naik, N., Parikh, D., Raskar, R., and Hidalgo, C. A. (2016b). Deep learning the
city: Quantifying urban perception at a global scale. In Computer Vision–ECCV 2016:
14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016,
Proceedings, Part I 14, pages 196–212. Springer.
Dutta, P., Aoki, P. M., Kumar, N., Mainwaring, A., Myers, C., Willett, W., and Woodruff, A.
(2009). Common sense: Participatory urban sensing using a network of handheld air
quality monitors. In Proceedings of the 7th ACM Conference on Embedded Networked
Sensor Systems, pages 349–350.
Egli, V., Zinn, C., Mackay, L., Donnellan, N., Villanueva, K., Mavoa, S., Exeter, D. J.,
Vandevijvere, S., and Smith, M. (2019). Viewing obesogenic advertising in
children’s neighbourhoods using Google street view. Geographical Research, 57(1):
84–97.
Fan, M., Huang, J., and Wang, H. (2022). Dumapper: Towards automatic verification
of largescale pois with street views at Baidu maps. In Proceedings of the 31st
ACM International Conference on Information & Knowledge Management, pages
3063–3071.
Fan, Z., Su, T., Sun, M., Noyman, A., Zhang, F., Pentland, A. S., and Moro, E. (2023).
Diversity beyond density: Experienced social mixing of urban streets. PNAS Nexus,
2(4): PNAS Nexus, 2023, 1–11.
Fang, F., Yu, Y., Li, S., Zuo, Z., Liu, Y., Wan, B., and Luo, Z. (2021). Synthesizing location
semantics from street view images to improve urban land-use classification.
International Journal of Geographical Information Science, 35(9):1802–1825.
Fleischhacker, S. E., Evenson, K. R., Sharkey, J., Pitts, S. B. J., and Rodriguez, D. A. (2013).
Validity of secondary retail food outlet data: A systematic review. American Journal of
Preventive Medicine, 45(4):462–473.
Frank, L. D., Schmid, T. L., Sallis, J. F., Chapman, J., and Saelens, B. E. (2005). Linking
objectively measured physical activity with objectively measured urban form: Findings
from smartraq. American Journal of Preventive Medicine, 28(2):117–125.
Galea, S., Ahern, J., Rudenstine, S., Wallace, Z., and Vlahov, D. (2005). Urban built
environment and depression: A multilevel analysis. Journal of Epidemiology &
Community Health, 59(10):822–827.
Gao, S., Yang, K., Shi, H., Wang, K., and Bai, J. (2022). Review on panoramic imaging and
its applications in scene understanding. IEEE Transactions on Instrumentation and
Measurement, 71:1–34.
Gebru, T., Krause, J., Wang, Y., Chen, D., Deng, J., Aiden, E. L., and Fei-Fei, L. (2017).
Using deep learning and Google street view to estimate the demographic makeup
of neighborhoods across the United States. Proceedings of the National Academy of
Sciences, 114(50):13108–13113.
Gehl, J. (2013). Cities for People. Island Press.
Gong, F.-Y., Zeng, Z.-C., Ng, E., and Norford, L. K. (2019). Spatiotemporal patterns of
streetlevel solar radiation estimated using Google street view in a high-density urban
environment. Building and Environment, 148:547–566.
Gong, F.-Y., Zeng, Z.-C., Zhang, F., Li, X., Ng, E., and Norford, L. K. (2018). Mapping sky,
tree, and building view factors of street canyons in a high-density urban environment.
Building and Environment, 134:155–167.
Goodchild, M. F. (2007). Citizens as sensors: The world of volunteered geography.
GeoJournal, 69(4):211–221.
Urban Sensing with GeoAI and Street‐Level Imagery 23

Goodchild, M. F. and Li, L. (2012). Assuring the quality of volunteered geographic

information. Spatial Statistics, 1:110–120.
Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville,
A., and Bengio, Y. (2014). Generative adversarial nets. In Advances in Neural
Information Processing Systems, pages 2672–2680.
Grabaskas, N. and Wang, Z. (2022). Intrinsic rewards for reinforcement learning within
complex 2d environments. In Intelligent Systems and Applications: Proceedings of
the 2021 Intelligent Systems Conference (IntelliSys) Volume 1, pages 425–437.
Springer.
Haddawy, P., Wettayakorn, P., Nonthaleerak, B., Su Yin, M., Wiratsudakul, A., Schoning, J.,
Laosiritaworn, Y., Balla, K., Euaungkanakul, S., Quengdaeng, P., et al. (2019). Large
scale detailed mapping of dengue vector breeding sites using street view images.
PLoS Neglected Tropical Diseases, 13(7):e0007555.
Hak, T., Janou skovˇ a, S., and Moldan, B. (2016). Sustainable development goals: A need
for relevant indicators. Ecological Indicators, 60:565–573.
Handy, S. L., Boarnet, M. G., Ewing, R., and Killingsworth, R. E. (2002). How the built
environment affects physical activity: Views from urban planning. American Journal
of Preventive Medicine, 23(2):64–73.
Harley, J. B. (1990). Cartography, ethics and social theory. Cartographica: The International
Journal for Geographic Information and Geovisualization, 27(2):1–23.
He, K., Gkioxari, G., Dollar, P., and Girshick, R. (2017). Mask R-CNN. In IEEE
International Conference on Computer Vision, pages 2980–2988. IEEE.
He, Z., Jiang, S., Zhang, J., and Wu, G. (2022). Automatic damage detection using
anchor-free method and unmanned surface vessel. Automation in Construction,
133:104017.
Hidalgo, M. C. and Hernandez, B. (2001). Place attachment: Conceptual and empirical
questions. Journal of Environmental Psychology, 21(3):273–281.
Hoehner, C. M., Ramirez, L. K. B., Elliott, M. B., Handy, S. L., and Brownson, R. C. (2005).
Perceived and objective environmental measures and physical activity among urban
adults. American Journal of Preventive Medicine, 28(2):105–116.
Hou, Y. and Biljecki, F. (2022). A comprehensive framework for evaluating the quality
of street view imagery. International Journal of Applied Earth Observation and
Geoinformation, 115:103094.
Hu, C.-B., Zhang, F., Gong, F.-Y., Ratti, C., and Li, X. (2020). Classification and mapping of
urban canyon geometry using Google street view images and deep multitask learning.
Building and Environment, 167:106424.
Hwang, J. and Sampson, R. J. (2014). Divergent pathways of gentrification: Racial inequality
and the social order of renewal in Chicago neighborhoods. American Sociological
Review, 79(4):726–751.
Ilic, L., Sawada, M., and Zarzelli, A. (2019). Deep mapping gentrification in a large
Canadian city using deep learning and Google street view. PloS One, 14(3):e0212814.
Jacobs, J. (2016). The death and life of great American cities. Vintage.
Jat, M. K., Garg, P. K., and Khare, D. (2008). Monitoring and modelling of urban sprawl
using remote sensing and GIS techniques. International Journal of Applied Earth
Observation and Geoinformation, 10(1):26–43.
Ji, W., Ma, J., Twibell, R. W., and Underhill, K. (2006). Characterizing urban sprawl using
multi-stage remote sensing images and landscape metrics. Computers, Environment
and Urban Systems, 30(6):861–879.
Jiang, B., Claramunt, C., and Klarqvist, B. (2000). Integration of space syntax into gis for
modelling urban spaces. International Journal of Applied Earth Observation and
Geoinformation, 2(3-4):161–171.
24 Handbook of Geospatial Approaches to Sustainable Cities

Jin, X., Long, Y., Sun, W., Lu, Y., Yang, X., and Tang, J. (2017). Evaluating cities’ vitality
and identifying ghost cities in China with emerging geographical data. Cities,
63:98–109.
Johnson, J. (2019). Jumping into the world of virtual and augmented reality. Knowledge
Quest, 47(4):22–27.
Johnson, N. S. and Gabler, H. C. (2015). Injury outcome in crashes with guardrail end
terminals. Traffic Injury Prevention, 16(sup2):S103–S108.
Juhasz, L. and Hochmair, H. H. (2016). User contribution patterns and completeness
evaluation of Mapillary, a crowdsourced street level photo service. Transactions in
GIS, 20(6):925–947.
Kamal, I. M. and Bae, H. (2022). Super-encoder with cooperative autoencoder networks.
Pattern Recognition, 126:108562.
Kamran, S. and Haas, O. (2007). A multilevel traffic incidents detection approach:
Identifying traffic patterns and vehicle behaviours using real-time GPS data. In 2007
IEEE Intelligent Vehicles Symposium, pages 912–917.
Kang, J., Korner, M., Wang, Y., Taubenb¨ ock, H., and Zhu, X. X. (2018). Building instance
classification using street view images. ISPRS Journal of Photogrammetry and Remote
Sensing, 145:44–59.
Kang, Y., Abraham, J., Ceccato, V., Duarte, F., Gao, S., Ljungqvist, L., Zhang, F., Nasman,
P., and Ratti, C. (2023). Assessing differences in safety perceptions using GeoAI and
survey across neighbourhoods in Stockholm, Sweden. Landscape and Urban Planning,
236:104768.
Kang, Y., Zhang, F., Gao, S., Lin, H., and Liu, Y. (2020). A review of urban physical
environment sensing using street view imagery in public health studies. Annals of GIS,
26(3):261–275.
Kang, Y., Zhang, F., Peng, W., Gao, S., Rao, J., Duarte, F., and Ratti, C. (2021).
Understanding house price appreciation using multi-source big geo-data and machine
learning. Land Use Policy, 111:104919.
Ke, T.-W., Hwang, J.-J., Guo, Y., Wang, X., and Yu, S. X. (2022). Unsupervised hierarchical
semantic segmentation with multiview cosegmentation and clustering transformers.
In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern
Recognition, pages 2571–2581.
Khorshidi, S., Carter, J., Mohler, G., and Tita, G. (2021). Explaining crime diversity with
Google street view. Journal of Quantitative Criminology, 37:361–391.
Khosla, A., An An, B., Lim, J. J., and Torralba, A. (2014). Looking beyond the visible scene.
In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,
pages 3710–3717.
Ki, D. and Lee, S. (2021). Analyzing the effects of green view index of neighborhood streets
on walking time using Google street view and deep learning. Landscape and Urban
Planning, 205:103920.
Kostakos, P., Alavesa, P., Oppenlaender, J., and Hosio, S. (2019). Vr ethnography: a pilot
study on the use of virtual reality ’go-along’ interviews in Google street view. In
Proceedings of the 18th International Conference on Mobile and Ubiquitous
Multimedia, pages 1–5.
Krylov, V. A., Kenny, E., and Dahyot, R. (2018). Automatic discovery and geotagging of
objects from street view imagery. Remote Sensing, 10(5):661.
Lander, C., Wiehr, F., Herbig, N., Kruger, A., and L¨ ochtefeld, M. (2017). Inferring
landmarks¨ for pedestrian navigation from mobile eye-tracking data and Google street
view. In Proceedings of the 2017 CHI Conference Extended Abstracts on Human
Factors in Computing Systems, pages 2721–2729.
Lauko, I. G., Honts, A., Beihoff, J., and Rupprecht, S. (2020). Local color and morphological
image feature based vegetation identification and its application to human environment
Urban Sensing with GeoAI and Street‐Level Imagery 25

street view vegetation mapping, or how green is our county? Geo-spatial Information
Science, 23(3):222–236.
Law, S., Paige, B., and Russell, C. (2019). Take a look around: using street view and satellite
images to estimate house prices. ACM Transactions on Intelligent Systems and
Technology, 10(5):1–19.
Lehtola, V. V., Koeva, M., Elberink, S. O., Raposo, P., Virtanen, J.-P., Vahdatikhaki, F.,
and Borsci, S. (2022). Digital twin of a city: Review of technology serving city
needs. International Journal of Applied Earth Observation and Geoinformation,
page 102915.
Less, E. L., McKee, P., Toomey, T., Nelson, T., Erickson, D., Xiong, S., and Jones-Webb, R.
(2015). Matching study areas using Google street view: A new application for an
emerging technology. Evaluation and Program Planning, 53:72–79.
Li, D., Deal, B., Zhou, X., Slavenas, M., and Sullivan, W. C. (2018a). Moving beyond the
neighborhood: Daily exposure to nature and adolescents’ mood. Landscape and Urban
Planning, 173:33–43.
Li, M. and Yao, W. (2020). 3D map system for tree monitoring in Hong Kong using Google
street view imagery and deep learning. ISPRS Annals of the Photogrammetry, Remote
Sensing and Spatial Information Sciences, 3:765–772.
Li, X., Cai, B. Y., Qiu, W., Zhao, J., and Ratti, C. (2019). A novel method for predicting and
mapping the occurrence of sun glare using Google street view. Transportation
Research Part C: Emerging Technologies, 106:132–144.
Li, X., Duarte, F., and Ratti, C. (2021). Analyzing the obstruction effects of obstacles on light
pollution caused by street lighting system in Cambridge, Massachusetts. Environment
and Planning B: Urban Analytics and City Science, 48(2):216–230.
Li, X. and Ratti, C. (2019). Using Google street view for street-level urban form analysis, a
case study in Cambridge, Massachusetts. The Mathematics of Urban Morphology,
5:457–470.
Li, X., Ratti, C., and Seiferling, I. (2018b). Quantifying the shade provision of street trees in
urban landscape: A case study in Boston, USA, using Google street view. Landscape
and Urban Planning, 169:81–91.
Li, X., Zhang, C., and Li, W. (2015a). Does the visibility of greenery increase perceived
safety in urban areas? evidence from the place pulse 1.0 dataset. ISPRS International
Journal of Geo-Information, 4(3):1166–1183.
Li, X., Zhang, C., Li, W., Ricard, R., Meng, Q., and Zhang, W. (2015b). Assessing street-
level urban greenery using Google street view and a modified green view index. Urban
Forestry & Urban Greening, 14(3):675–685.
Li, Z. and Ma, J. (2022). Discussing street tree planning based on pedestrian volume
using machine learning and computer vision. Building and Environment, 219:
109178.
Liang, B. and Weng, Q. (2010). Assessing urban environmental quality change of
Indianapolis, United States, by the remote sensing and GIS integration. IEEE
Journal of Selected Topics in Applied Earth Observations and Remote Sensing,
4(1):43–55.
Liang, C., Serge, A., Zhang, X., Wang, H., and Wang, W. (2023). Assessment of street forest
characteristics in four African cities using Google street view measurement: Potentials
and implications. Environmental Research, 221:115261.
Lin, J., Zheng, Z., Zhong, Z., Luo, Z., Li, S., Yang, Y., and Sebe, N. (2022). Joint
representation learning and keypoint detection for cross-view geo-localization. IEEE
Transactions on Image Processing, 31:3780–3792.
Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C.-Y., and Berg, A. C. (2016).
SSD: Single shot multibox detector. In European Conference on Computer Vision,
pages 21–37.
26 Handbook of Geospatial Approaches to Sustainable Cities

Liu, Y., Zhang, X., Ding, J., Xi, Y., and Li, Y. (2023). Knowledge-infused contrastive
learning for urban imagery-based socioeconomic prediction. In Proceedings of the
ACM Web Conference 2023, pages 4150–4160.
Lynch, K. (1964). The Image of the City. MIT Press.
Masson, V., Lemonsu, A., Hidalgo, J., and Voogt, J. (2020). Urban climates and climate
change. Annual Review of Environment and Resources, 45:411–444.
Mayer, K., Haas, L., Huang, T., Bernabe-Moreno, J., Rajagopal, R., and Fischer, M. (2023).
Estimating building energy efficiency from street view imagery, aerial imagery, and
land surface temperature data. Applied Energy, 333:120542.
Meyer, M. A. and Hendricks, M. D. (2018). Using photography to assess housing damage
and rebuilding progress for disaster recovery planning. Journal of the American
Planning Association, 84(2):127–144.
Micusik, B. and Kosecka, J. (2009). Piecewise planar city 3D modeling from street view
panoramic sequences. In Proceedings of the IEEE Conference on Computer Vision and
Pattern Recognition.
Miller, R. B. and Small, C. (2003). Cities from space: potential applications of remote
sensing in urban environmental research and policy. Environmental Science & Policy,
6(2):129–137.
Mohamed, A., Lee, H.-y., Borgholt, L., Havtorn, J. D., Edin, J., Igel, C., Kirchhoff, K., Li, S.-
W., Livescu, K., Maaløe, L., et al. (2022). Self-supervised speech representation
learning: A review. IEEE Journal of Selected Topics in Signal Processing, 16,
1179–1210.
Mooney, S. J., DiMaggio, C. J., Lovasi, G. S., Neckerman, K. M., Bader, M. D., Teitler, J.
O., Sheehan, D. M., Jack, D. W., and Rundle, A. G. (2016). Use of Google street view
to assess environmental contributions to pedestrian injury. American Journal of Public
Health, 106(3):462–469.
Nagata, S., Nakaya, T., Hanibuchi, T., Amagasa, S., Kikuchi, H., and Inoue, S. (2020).
Objective scoring of streetscape walkability related to leisure walking: Statistical
modeling approach with semantic segmentation of Google street view images. Health
& Place, 66:102428.
Naik, N., Kominers, S. D., Raskar, R., Glaeser, E. L., and Hidalgo, C. A. (2017). Computer
vision uncovers predictors of physical urban change. Proceedings of the National
Academy of Sciences, 114(29):7571–7576.
Naik, N., Philipoom, J., Raskar, R., and Hidalgo, C. (2014). Streetscore-predicting the
perceived safety of one million streetscapes In Proceedings of the IEEE Conference on
Computer Vision and Pattern Recognition Workshops. 779–785.
Najmi, A., Gevaert, C. M., Kohli, D., Kuffer, M., and Pratomo, J. (2022). Integrating remote
sensing and street view imagery for mapping slums. ISPRS International Journal of
GeoInformation, 11(12):631.
Nguyen, Q. C., Huang, Y., Kumar, A., Duan, H., Keralis, J. M., Dwivedi, P.,
Meng, H.-W., Brunisholz, K. D., Jay, J., Javanmardi, M., et al. (2020). Using
164 million Google street view images to derive built environment predictors of
covid-19 cases. International Journal of Environmental Research and Public
Health, 17(17):6359.
Ning, H., Ye, X., Chen, Z., Liu, T., and Cao, T. (2022). Sidewalk extraction using aerial and
street view images. Environment and Planning B: Urban Analytics and City Science,
49(1):7–22.
Noyman, A. and Larson, K. (2020). A deep image of the city: Generative urban-design
visualization. Challenge, 7:30.
Oliveira, A. A. A., Wang, Z., and Hirata, R. (2022). Locating urban trees near electric wires
using Google street view photos: A new dataset and a semi-supervised learning
Urban Sensing with GeoAI and Street‐Level Imagery 27

approach in the wild. In Proceedings of the IEEE/CVF Conference on Computer

Vision and Pattern Recognition, pages 4286–4294.
O’Regan, A. C., Byrne, R., Hellebust, S., and Nyhan, M. M. (2022). Associations between
Google street view-derived urban greenspace metrics and air pollution measured
using a distributed sensor network. Sustainable Cities and Society, 87:104221.
Pang, H. E. and Biljecki, F. (2022). 3D building reconstruction from single street view
images using deep learning. International Journal of Applied Earth Observation and
Geoinformation, 112:102859.
Perdue, W. C., Stone, L. A., and Gostin, L. O. (2003). The built environment and its
relationship to the public’s health: The legal framework. American Journal of Public
Health, 93(9):1390–1394.
Richter, S. R., AlHaija, H. A., and Koltun, V. (2022). Enhancing photorealism enhancement,
45(2):1700–1715.
Rundle, A. G., Bader, M. D., Richards, C. A., Neckerman, K. M., and Teitler, J. O. (2011).
Using Google street view to audit neighborhood environments. American Journal of
Preventive Medicine, 40(1):94–100.
Rzotkiewicz, A., Pearson, A. L., Dougherty, B. V., Shortridge, A., and Wilson, N. (2018).
Systematic review of the use of Google street view in health research: Major
themes, strengths, weaknesses and possibilities for future research. Health & place,
52:240–246.
Sampson, R. J. and Raudenbush, S. W. (1999). Systematic social observation of public
spaces: A new look at disorder in urban neighborhoods. American Journal of
Sociology, 105(3):603–651.
Seekins, T., Rennie, B., and Hammond, J. (2014). Exploring the feasibility of using “google
street view” to assess the accessibility of community environments: developing
definitions and observational protocol for image recognition and classification. In
Environmental Contexts and Disability, volume 8, pages 123–139.
Shi, L., Chu, E., Anguelovski, I., Aylett, A., Debats, J., Goh, K., Schenk, T., Seto, K. C.,
Dodman, D., Roberts, D., et al. (2016). Roadmap towards justice in urban climate
adaptation research. Nature Climate Change, 6(2):131–137.
Soga, M., Evans, M. J., Tsuchiya, K., and Fukano, Y. (2021). A room with a green view: The
importance of nearby nature for mental health during the covid-19 pandemic.
Ecological Applications, 31(2):e2248.
Suel, E., Polak, J. W., Bennett, J. E., and Ezzati, M. (2019). Measuring social, environmental
and health inequalities using deep learning and street imagery. Scientific Reports,
9(1):1–10.
Sun, M., Zhang, F., and Duarte, F. (2021). Automatic building age prediction from street
view images. In 2021 7th IEEE International Conference on Network Intelligence
and Digital Content (IC-NIDC), pages 102–106. IEEE.
Sun, M., Zhang, F., Duarte, F., and Ratti, C. (2022). Understanding architecture age and style
through deep learning. Cities, 128:103787.
Szczesniak, J. T., Ang, Y. Q., Letellier-Duchesne, S., and Reinhart, C. F. (2022). A method
for using street view imagery to auto-extract window-to-wall ratios and its relevance
for urbanlevel daylighting and energy simulations. Building and Environment, 207:
108108.
Takizawa, A. and Kinugawa, H. (2020). Deep learning model to reconstruct 3d cityscapes by
generating depth maps from omnidirectional images and its application to visual
preference prediction. Design Science, 6:e28.
Thackway, W., Ng, M., Lee, C.-L., and Pettit, C. (2023). Implementing a deep-learning
model using Google street view to combine social and physical indicators of
gentrification. Computers, Environment and Urban Systems, 102:101970.
28 Handbook of Geospatial Approaches to Sustainable Cities

Van der Spek, S., Van Schaick, J., De Bois, P., and De Haan, R. (2009). Sensing human
activity: GPS tracking. Sensors, 9(4):3033–3055.
van Nes, A., Yamu, C., van Nes, A., and Yamu, C. (2021). Space syntax applied in urban
practice. Introduction to Space Syntax in Urban Studies, pages 213–237.
Verma, D., Jana, A., and Ramamritham, K. (2020). Predicting human perception of the urban
environment in a spatiotemporal urban setting using locally acquired street view
images and audio clips. Building and Environment, 186:107340.
Vo, A. V., Bertolotto, M., Ofterdinger, U., and Laefer, D. F. (2023). In search of basement
indicators from street view imagery data: An investigation of data sources and analysis
strategies. KI-Künstliche Intelligenz, pages 1–13.
Voumard, J., Abellan, A., Nicolet, P., Penna, I., Chanut, M.-A., Derron, M.-H., and Jaboyedo
ff, M. (2017). Using street view imagery for 3-d survey of rock slope failures. Natural
Hazards and Earth System Sciences, 17(12):2093–2107.
Wang, B., Xu, T., Gao, H., Ta, N., Chai, Y., and Wu, J. (2021a). Can daily mobility alleviate
green inequality from living and working environments? Landscape and Urban
Planning, 214:104179.
Wang, J. and Biljecki, F. (2022). Unsupervised machine learning in urban studies: A
Systematic Review of Applications. Cities, 129:103925.
Wang, J., Sun, K., Cheng, T., Jiang, B., Deng, C., Zhao, Y., Liu, D., Mu, Y., Tan, M., Wang,
X., et al. (2020). Deep high-resolution representation learning for visual recognition.
IEEE Transactions on Pattern Analysis and Machine Intelligence, 43:3349–3364.
Wang, L., Han, X., He, J., and Jung, T. (2022a). Measuring residents’ perceptions of city
streets to inform better street planning through deep learning and space syntax.
ISPRS Journal of Photogrammetry and Remote Sensing, 190:215–230.
Wang, R., Feng, Z., Pearce, J., Yao, Y., Li, X., and Liu, Y. (2021b). The distribution of
greenspace quantity and quality and their association with neighbourhood socio-
economic conditions in Guangzhou, China: A new approach using deep learning
method and street view images. Sustainable Cities and Society, 66:102664.
Wang, R., Lu, Y., Zhang, J., Liu, P., Yao, Y., and Liu, Y. (2019). The relationship between
visual enclosure for neighbourhood street walkability and elders’ mental health in
China: Using street view images. Journal of Transport & Health, 13:90–102.
Wang, Y., Zeng, Z., Li, Q., and Deng, Y. (2022b). A complete reinforcement-learning-based
framework for urban-safety perception. ISPRS International Journal of Geo-
Information, 11(9):465.
Whyte, W. H. et al. (1980). The Social Life of Small Urban Spaces. Conservation Foundation
Washington, DC.
Wilson, J. Q. and Kelling, G. L. (2015). Broken windows. Routledge.
Wilson, N. and Thomson, G. (2015). Suboptimal smokefree signage at some hospitals: field
observations and the use of google street view. The New Zealand Medical Journal,
128(1415):56.
Wilson, W. H. (1989). The City Beautiful Movement. Johns Hopkins University Press
Baltimore.
Xia, Y., Yabuki, N., and Fukuda, T. (2021). Sky view factor estimation from street view
images based on semantic segmentation. Urban Climate, 40:100999.
Xiang, L., Cai, M., Ren, C., and Ng, E. (2021). Modeling pedestrian emotion in high-density
cities using visual exposure and machine learning: Tracking real-time physiology and
psychology in Hong Kong. Building and Environment, 205:108273.
Xiao, J. and Quan, L. (2009). Multiple view semantic segmentation for street view images.
In 2009 IEEE 12th International Conference on Computer Vision, pages 686–693.
IEEE.
Xie, Y., Sha, Z., and Yu, M. (2008). Remote sensing imagery in vegetation mapping: a
review. Journal of Plant Ecology, 1(1):9–23.
Urban Sensing with GeoAI and Street‐Level Imagery 29

Xing, Z., Yang, S., Zan, X., Dong, X., Yao, Y., Liu, Z., and Zhang, X. (2023).
Flood vulnerability assessment of urban buildings based on integrating high-
resolution remote sensing and street view images. Sustainable Cities and Society,
92:104467.
Yan, Y., Feng, C.-C., Huang, W., Fan, H., Wang, Y.-C., and Zipf, A. (2020). Volunteered
geographic information research in the first decade: A narrative review of selected
journal articles in GIScience. International Journal of Geographical Information
Science, 34(9):1765–1791.
Yang, J., Zhao, L., Mcbride, J., and Gong, P. (2009). Can you see green? assessing the
visibility of urban forests in cities. Landscape and Urban Planning, 91(2):97–104.
Yang, X., Song, Z., King, I., and Xu, Z. (2022). A survey on deep semi-supervised learning.
IEEE Transactions on Knowledge and Data Engineering, 35:8934–8954.
Yang, Y., Lu, Y., Yang, L., Gou, Z., and Zhang, X. (2020). Urban greenery, active school
transport, and body weight among Hong Kong children. Travel Behaviour and Society,
20:104–113.
Yao, Y., Zhang, J., Qian, C., Wang, Y., Ren, S., Yuan, Z., and Guan, Q. (2021). Delineating
urban job-housing patterns at a parcel scale with street view imagery. International
Journal of Geographical Information Science, 35(10):1927–1950.
Ye, C., Zhang, F., Mu, L., Gao, Y., and Liu, Y. (2020). Urban function recognition by
integrating social media and street-level imagery. Environment and Planning B: Urban
Analytics and City Science.
Ye, C., Zhang, F., Mu, L., Gao, Y., and Liu, Y. (2021). Urban function recognition by
integrating social media and street-level imagery. Environment and Planning B: Urban
Analytics and City Science, 48(6):1430–1444.
Ye, Y., Zeng, W., Shen, Q., Zhang, X., and Lu, Y. (2019). The visual quality of streets: A
human-centred continuous measurement based on machine learning algorithms and
street view images. Environment and Planning B: Urban Analytics and City Science,
46(8):1439– 1457.
Yin, L., Cheng, Q., Wang, Z., and Shao, Z. (2015). ‘Big data’ for pedestrian volume:
Exploring the use of Google street view images for pedestrian counts. Applied
Geography, 63:337–345.
Yin, L. and Wang, Z. (2016). Measuring visual enclosure for street walkability: Using
machine learning algorithms and Google street view imagery. Applied Geography,
76:147–153.
Yu, X., Her, Y., Huo, W., Chen, G., and Qi, W. (2022). Spatio-temporal monitoring of
urban street-side vegetation greenery using Baidu street view images. Urban Forestry
& Urban Greening, 73:127617.
Yuan, Y., Wang, R., Niu, T., and Liu, Y. (2023). Using street view images and
a geographical detector to understand how street-level built environment is
associated with urban poverty: A case study in Guangzhou. Applied Geography,
156:102980.
Zhang, A., Song, L., and Zhang, F. (2022). Perception of pleasure in the urban running
environment with street view images and running routes. Journal of Geographical
Sciences, 32(12):2624–2640.
Zhang, F., Fan, Z., Kang, Y., Hu, Y., and Ratti, C. (2021). “Perception bias”: Deciphering a
mismatch between urban crime and perception of safety. Landscape and Urban
Planning, 207:104003.
Zhang, F., Mirandaa, A. S., Duarte, F., Vale, L., Hack, G., Liu, Y., Batty, M., and Ratti, C.
(2023a). Urban visual intelligence: Studying cities with AI and street-level imagery.
Annals of the American Association of Geographers, 114(2):1–18.
Zhang, F., Wegner, J. D., Yang, B., and Liu, Y. (2023b). Street-level imagery analytics and
applications. ISPRS Journal of Photogrammetry and Remote Sensing, 199:195–196.
Random documents with unrelated
content Scribd suggests to you:
—Adieu, mon chéri ... Et à toi, si tu veux, quand tu voudras!...
Adieu!
La porte claqua au loin. C'était fini.
Mareuil alluma une cigarette et descendit sans se presser. Il se
sentait un peu ahuri et tamponnait machinalement ses paupières
brûlantes.
Sous la voûte, Mme Honoré la concierge courut après lui:
—Est-ce que Monsieur vient demain? Dois-je préparer du feu?
Il n'eut pas l'humilité de dire non, et d'un ton sec il répondit:
—Je ne sais pas, préparez toujours!
Puis il s'en alla rapidement, car Mme Honoré fixait, avec une
persistance curieuse, ses yeux gonflés d'avoir pleuré.
 IV
Le soir même, chez les Brévannes, il y avait dîner intime, toute la
bande conviée.
Dès son arrivée, Mareuil prit Brévannes à part et lui conta l'entrevue
de l'après-midi.
Le journaliste souriait, approuvait par instants:
—Bravo! Très bien!
Il conclut:
—Eh bien, tout cela me semble parfait ... Je suppose que
maintenant vous allez vous tenir tranquille ...
Henriette, que ce colloque intriguait et qui avait entendu les derniers
mots, jugea bon d'intervenir:
—Quoi donc?... Qu'est-ce qui est parfait?
Et s'adressant à Mareuil, avec un air de compassion:
—Encore des ennuis, mon pauvre Soif-d'Amour?
Elle se tut, ayant remarqué que Brévannes la guettait d'un œil
courroucé.
—Ma jeune enfant, lui dit-il, je vous dispense, à l'avenir, d'appeler
Mareuil de ce nom ridicule ... Un jour il se fâchera, et je vous jure
bien que ce n'est pas moi qui lui donnerai tort!...
Henriette, interdite par cette réprimande, si contraire aux usages
constants de la maison, essaya, par contenance, de la puérilité:
—Est-ce qu'on m'avait dit, à moi?... Faut pas me gronder, moi!
La tentative échoua misérablement.
—Voyons, tu sais bien que j'ai horreur de ces manières-là, bougonna
Brévannes ... Occupe-toi donc plutôt du dîner!...
Les autres s'étaient à peine dérangés durant cette courte altercation.
Labernerie, qui avait hargneusement levé la tête de dessus son
journal, reprit sa lecture. Le Grand-Cob referma l'œil qu'il s'était
efforcé d'entr'ouvrir, malgré la somnolence qui lui restait d'une nuit
passée, la veille, dans les cabarets des Halles. Charleval paraissait
emporté dans des conceptions ... Ce Mareuil, encore! Ses affaires de
cœur! Sa dame du monde! Et ils songeaient qu'il était bien gentil,
Mareuil, mais tout de même un peu raseur, quelquefois, avec ses
histoires de femmes!

Le lendemain matin, Gilbert, qui, sur les exhortations de Brévannes,

s'était docilement grisé au dîner, se réveilla presque à l'aube, après
un sommeil lourd et sans rêves.
Il glissa à bas de son lit et s'approcha de la fenêtre. L'avenue de
Villiers était toute solitaire, toute blanche. On entendait, dans le
silence du matin, le bruissement rythmé du balai d'un balayeur qui
frôlait les trottoirs. Le ciel était bleu foncé, un ciel profond de
premier printemps. Mareuil se dit que la journée serait sans doute
belle et tiède, puis aussitôt il pensa: «Je n'ai plus d'amie!»
C'était la même pensée qui l'avait obsédé à son départ de la rue
Fortuny, qui l'avait hanté, la soirée précédente, jusque dans le
trouble de la griserie, qui revenait dès le réveil: «Plus d'amie!»—et il
se répétait ces mots mélancoliques comme pour s'accoutumer à leur
sens étrange.
Il aurait voulu avoir été l'amant de Mme Hardouin, mais il y avait
bien, bien longtemps—l'avoir été à la manière des hommes
grisonnants dont on dit: «Il a été l'amant de Mme Un tel ...»—mais
d'un ton historique, d'une voix qui indique des années et des années
écoulées depuis.
Il réfléchissait:
«Et il viendra, ce moment-là, comme ils viennent tous, l'un après
l'autre, en mystère, à pas de loup ... Comme est venu celui de la
rupture, ce moment que j'attendais si fiévreusement depuis un mois
et qui est arrivé pourtant, qui est accompli, qui est fini maintenant ...
J'ai déjà vu cela en chemin de fer ... Le train n'avançait pas, avait du
retard ... Et tout à coup, j'étais en gare, dans la rue, en voiture, chez
moi, arrivé, installé, sans comprendre comment ... Est-ce curieux,
ces jours, ces heures, ces minutes, ce temps qui passe dans l'ombre,
en silence!...»
On lui apportait son thé. Il mangea d'assez bon appétit. Puis il fit sa
toilette, s'attardant en des flâneries calculées, s'arrêtant pour
rêvasser, parcourir un article de journal, choisir un vêtement, une
cravate; et, comme il lui restait une heure à employer avant le
déjeuner, il voulut relire d'anciennes lettres de Jack.
Il les retira d'un tiroir où elles étaient serrées, empilées au point que
la serrure fermait avec peine; et il les jeta par poignées sur sa table.
Il contemplait tous ces papiers blancs, gris et mauves, épais comme
du carton ou ténus et plissés comme de la dentelle, d'où montait
une vapeur douce de parfum mourant, de roses desséchées.
«Y en a-t-il, des saletés, là-dedans, et des canailleries!»
Il commença à lire, au hasard, élucidant les pensées que cachaient
ces mots enchevêtrés, cette encre noire ou violette, ces lignes
crayonnées en hâte,—comparant, confrontant, fouillant activement,
dans ce tas de mensonges, comme un chiffonnier picorant dans un
monceau d'ordures; et quand il avait découvert la preuve d'une
imposture, établi l'évidence d'une contradiction, il ressentait une joie
sauvage, une satisfaction méprisante. Il quittait sa chaise, marchait
quelques instants, la tête basse, puis venait se remettre à sa lecture.
On frappa à la porte.
—Monsieur est servi.
Il repoussa les papiers dans le tiroir, les refoula à coups de poing, et
descendit déjeuner.
Mais lorsqu'il fut remonté dans son atelier et qu'il vit, presque devant
lui, la grande étendue de temps, blanche et vide, qui se déroulait
jusqu'au dîner, il ne put se défendre d'un intime mouvement de
détresse:
«Qu'est-ce que je vais faire d'ici là? Qu'est-ce que je vais faire?»
Pas de lettre à attendre! Pas de lettre à écrire! Il aurait souhaité
d'être encore plus jeune de deux jours, au temps maudit où il
s'inquiétait, où il souffrait, mais où il luttait, au moins, où il se
démenait, où il faisait quelque chose. Et il éprouvait ce terrible
frisson de regret et d'ennui qui tourmente souvent les vieux officiers
démissionnaires quand sonne dans leur oisiveté l'heure connue du
rapport, de la botte ou de la manœuvre.
Il s'allongea dans un fauteuil, fumant coup sur coup, et pour
s'étourdir, des cigarettes qui allaient s'éparpiller en petits tas
jaunâtres, à demi brûlées, sur le dallage noir de la cheminée.
«Oui, je n'ai pas le choix; il s'agit de changer ma vie, d'inventer une
distraction, de m'organiser autrement ...»
Mais, à peine les combinaisons formées, elles s'écroulaient, et, sur
leurs décombres, sur leurs chiffres en ruines, voletait l'image de Jack
—de Mme Hardouin s'acheminant vers le logis d'un monsieur à vague
moustache, ou montant son escalier, tout aimable, prête à s'offrir.
Mareuil se leva, dégoûté:
—Eh bien, cela va être gai! murmura-t-il. Cela va être frais, si c'est
tous les jours ainsi!
Il saisit une large feuille de papier dans un cartonnier placé près de
la fenêtre, et, s'asseyant en bonne lumière, il se mit à dessiner une
tête de femme, au caprice du crayon, sans autre idée nette que de
s'occuper, de fuir ce néant du rien-à-faire.
Mais, soudain, d'un geste furieux, il brisa contre le papier la pointe
de son crayon qui vint frapper la vitre avec un bruit argentin.
Non! Il en avait assez, à la fin, de tracer des petites lignes, d'écraser
du noir sur du blanc, de s'appliquer aux reliefs, de tenir compte de la
perspective, de s'acharner à ce nez, à cette bouche, à ces yeux de
femme—à ces attraits en mine de plomb, qui n'étaient rien, rien du
tout, il le savait bien, lui Mareuil, auprès d'un vrai visage de femme
tendu de peau vivante et parfumée.
Alors, il ouvrit un journal, et, du regard, courut à la recherche d'un
écho, d'une annonce qui lui fournirait le moyen de terminer cette
infernale après-midi.
Il lut que, ce jour-là, avait lieu la dernière réunion du Concours
Hippique; et en même temps, il revit l'entassement de dames
élégantes parmi les uniformes clairs et les fanfares de chasse, toute
la brillante assemblée qui devait s'agiter, sous le soleil printanier,
dans la grande nef sonore.
«C'est cela!... Je vais y aller», pensa-t-il.
Puis en rangeant ses objets de dessin, il s'avoua: «Si je trouvais là-
bas, ce qu'il me faut!... C'est peu probable ... Mais qu'est-ce que je
risque?... Et, dès le lendemain ... hé! ce ne serait pas ordinaire!»
Il s'habilla avec soin, comme autrefois aux jours des rendez-vous, et,
vers quatre heures, il franchissait le tourniquet du Palais de
l'Industrie.

Il hésitait de quel côté il monterait, quand une voix railleuse le héla:

—Comment? Vous ici?
Il se retourna et aperçut, face à la porte, le Grand Cob, les jambes
écartées, les mains balançant, derrière le dos, un parapluie—l'œil
inspecteur et aigu du boulevardier auquel rien d'un défilé parisien
n'échappera:
—Vous ici? répéta le Grand Cob, en dégageant mollement une de ses
mains gantées de rouge, et la tendant à Mareuil ... Mais je croyais
que vous ne vous montriez jamais dans ces endroits, que l'on ne
vous rencontrait nulle part ... Cela ne va donc plus?
Et de la pomme de son parapluie, il se cognait le thorax à gauche, à
la place où il supposait qu'on avait ce qu'on appelle un cœur.
Mareuil rectifia:
—Dites que cela va mieux ...
Puis, pour couper court aux questions:
—Faisons-nous un tour?
—Comment donc! déclara le Grand Cob flatté de voir Gilbert se
départir de sa froideur coutumière ... Tenez, je vais vous mener à la
Butte ... Nous avons aujourd'hui quelques numéros de luxe.
Et, prenant le bras de Mareuil, il l'entraîna vers la tribune réservée
aux demoiselles.
Elles étaient reléguées là, non par la pudibonderie des règlements,
mais de leur plein gré, pour la commodité des causeries libres et des
affaires à traiter, comme en une sorte de Bourse d'amour—toutes les
courtisanes de Paris, les illustres et les obscures, celles qui ne
vivaient que de leur beauté, celles qui avaient réussi par la gaieté
seule, la bonne humeur, et d'autres, petites femmes de petits
théâtres, dont on ne savait si c'était au lit ou à la scène qu'elles
avaient gagné leur clientèle et leurs falbalas pimpants. Elles se
serraient dans les étroites rangées de gradins, la plupart tournant le
dos à l'immense rectangle jaune de la piste, tandis que, juchés sur
un degré supérieur, ou du haut du couloir longeant le mur de la
tribune, des hommes leur parlaient, penchés sur elles comme pour
les humer—des hommes souriant d'un sourire camarade ou lubrique,
des hommes de toute catégorie: vieillards au regard indécis et
glouton, officiers à la taille pincée, aux cuisses disparues en des
culottes éclatantes et boursouflées, clubmen réputés, portant à leurs
vastes cravates des épingles sauvages, marques de leurs goûts
hippiques et formées de deux longues dents de cheval accolées.
Mais au passage de Mareuil et de Gendrey, les conversations
particulières cessaient. Les têtes pâles ou peintes de ces dames,
leurs chapeaux fleuris frémissaient, ainsi qu'un parterre sous une
brise; et elles se poussaient du coude, murmurant: «Le Grand Cob!
le Grand Cob!»
Gendrey semblait insensible à ces signes de déférence. Depuis
longtemps, il ne saluait même plus ses féales, se contentant de leur
grimacer de l'œil, de la bouche, du nez, des bonjours familiers et
paternels, comme en a pour les employés de son ministère un vieil
huissier inamovible, au courant du personnel et des traditions.
Il nommait maintenant les femmes à Mareuil, joignant à sa
nomenclature des commentaires sur les mœurs du lieu, tirant, à
mesure, la philosophie de ce qu'ils voyaient:
—Voilà Thérèse Nivolas ... Suzette de Luz ... Claire de Kerjeu ...
Réussie, cette blonde, hein?... Ninette Rabastens ... Paula Mériel ...
Et cette petite, basse sur jambes, à droite, en rouge, avec une figure
de garçon ... C'est Angèle de Cérans ... Cela a débuté, il n'y a pas six
mois, et cela a déjà hôtel avenue d'Iéna, victoria, cheval de selle ...
Et ce n'est pas fini! Elle ira loin, cette enfant, c'est moi qui vous le
dis!...
Il avait des inclinaisons du buste, des façons de se rejeter en arrière,
de dessiner de la main, dans le vide, des tailles, des poitrines, des
croupes, comme un maquignon désintéressé promenant un amateur
à travers la foire aux chevaux. On se doutait qu'il eût aimé faire
reconnaître à Mareuil la délicatesse des attaches, la solidité des
chairs, l'élasticité des muscles, lui faire toucher, palper ces membres
bien pris et sans tares dont il répondait.
Il continua:
—C'est le dernier jour. On n'a que le temps, vous comprenez.
Pendant toute la semaine, on a préparé les villégiatures, les petits
collages d'été ... Aujourd'hui, ce n'est plus l'heure de débattre ... Il
faut prendre ses arrangements, conclure ... La saison marche. Nous
n'avons plus avant juillet que les courses, pour rencontres sérieuses.
Mais là, les hommes jouent. Ils sont inabordables. Ils sont comme
fous ... Ils ne se connaissent plus ... Ah! elles le savent bien, les
petites ... Ainsi, regardez Claire de Kerjeu ... un caractère impossible
... On l'a baptisée la Fée-Colère ... Eh bien, voyez-la donc en ce
moment avec le jeune Châtel, le fils du grand épicier ... A-t-elle l'air
assez bonne fille, assez bon enfant!...
Mareuil ne l'écoutait plus. Instinctivement et comme attiré, il s'était
rapproché d'une jeune femme vêtue d'une correcte robe de drap
bleu sombre et à qui des cheveux bouffant en éventail, au-dessus du
front large et haut, donnaient un certain aspect de ressemblance
avec Mme Hardouin.
La jeune personne paraissait très excitée, et Mareuil entendit une
grosse voix enrouée qui sortait de sa bouche, sinueuse comme celle
de Jack, et qui disait:
—Alors, ma chère, figure-toi que cette espèce de sale voyou ...
Mareuil se recula, navré de la déception, sans la moindre curiosité
au sujet de ce que s'était permis l'espèce de sale voyou mentionné.
—Au revoir, je vais jeter un coup d'œil dans les autres tribunes, fit-il
en serrant la main de Gendrey.
Le Grand Cob parut choqué de ce départ comme d'un insuccès
personnel.
—Bon! Bon! Ainsi il n'y a rien ici qui vous plaise? Peste, mon petit,
vous êtes difficile!... Allez donc voir dans le bâtiment à côté. C'est
bien mieux ... Ah! c'est joli! C'est tout neuf! Elles sont là quatre ou
cinq avec une réputation de beauté qu'on leur continue depuis vingt-
cinq ans comme une rente viagère ... La belle madame Fourneau, la
belle madame de Bleize, n'est-ce pas? On en parlait déjà quand
j'étais haut comme cela ... Est-ce que ce sont toujours les mêmes?...
Mareuil lui serra de nouveau la main et s'esquiva, ne voulant pas
engager une discussion sociale sur ces rivalités de classes, ni relever
tout ce qu'avaient d'injustes et de superficiel les appréciations du
Grand Cob.
Il se fraya péniblement sa route à travers la foule qui, chaque
minute, devenait plus dense, plus résistante, n'avançait que par
brèves secousses, suivies de long moments d'arrêt.
Il se réjouissait secrètement de l'indifférence avec laquelle il avait
examiné toutes les charmantes amies du Grand Cob.
Devant leurs jolis corps accessibles, il n'avait eu aucun désir, aucune
tentation. Il n'avait pas dérogé, il demeurait le cœur dédaigneux des
amours brutales qu'il était la veille encore, lors de cette dramatique
scène d'adieux que personne de toutes ces personnes ne savait,
dont personne ne le soupçonnait le héros.
Mais, lorsqu'il eut achevé le tour des autres tribunes,
progressivement son orgueil l'abandonna.
Ici c'étaient, comme là-bas, des dames tournant le dos à la piste et
debout, sur lesquelles se penchaient des messieurs en uniforme ou à
épingles en dents de cheval. Ici, comme là-bas, les femmes étaient
suspendues des deux mains, ainsi qu'à un frêle mât de cocagne, au
manche de leur ombrelle, dont la pointe plissait la toile rouge des
banquettes; et elles gardaient, dans cette posture implorante, les
yeux levés vers les yeux baissés des messieurs. Ici, comme là-bas,
on paraissait pressé de préparer des petits collages d'été, de prendre
des arrangements, de conclure.
Et Mareuil avait des accès de jalousie impersonnelle, rétrospective, à
voir toutes ces créatures, toutes ces Jack, en train certainement de
faire à d'autres,—à d'autres amants absents, ingénus et dévoués—ce
qu'on lui avait fait à lui, pendant deux années.
Il dévisageait insolemment les clubmen en redingote, les cavaliers
en habit rouge, les officiers bleu-de-ciel ou noirs songeant, à
chacun: «Si c'était lui!»—énervé à l'idée de frôler peut-être, sans le
savoir, un de ses rivaux d'hier, un de ceux qui avaient tenu dans
leurs bras Mme Hardouin, tandis qu'il agonisait à l'attendre.
Plusieurs étaient célèbres pour leur élégance, leur agilité en selle,
leurs succès mondains et féminins. Les plus humbles bourgeoises les
connaissaient par leurs noms, les suivaient longtemps du regard
comme des acteurs populaires. Et au milieu de ces gars solides, à la
figure vaniteuse et rude, ou bien qui souriaient d'un sourire de
maître, en se chuchotant, par-dessus l'épaule, des réflexions
comiques, Mareuil avait presque honte de s'être donné la peine
d'aimer, de s'être tant courbé à supplier, à souffrir. Il se sentait
ignorant, naïf et faible comme un petit potache cerné par une cohue
de «grands».
Il murmura:
—Tous ces gens me répugnent! Allons-nous-en!
Et il marcha vers la sortie, saluant, d'un air hâtif d'homme qui s'en
va, les têtes qui le saluaient sur la route.
—Bigre! fit-il à un tournant ... La mère Lepassereau!
Il voulut se dissimuler derrière un monsieur qui le précédait, affecter
de n'avoir rien aperçu. Mais trop tard! Leurs regards s'étaient joints,
heurtés, et maintenant ceux de ladite mère Lepassereau ne le
lâchaient plus, dardaient contre lui des lueurs à la fois avenantes et
de menace. Il eut peur de sembler impoli et, résigné, il s'approcha
de la grosse dame.
Mme Lepassereau feignit hypocritement la surprise.
—Tiens, monsieur Mareuil!
Puis elle multiplia les questions au sujet de madame sa mère, une si
agréable femme; de monsieur son oncle, dont le château était voisin
de celui des Lepassereau, en Normandie, et au sujet aussi de son
travail à lui, M. Mareuil, qu'on lui avait dit peindre de si jolies choses,
mais là, sans compliments.
Elle l'appelait, à certains instants, Gilbert, s'excusant de sa
familiarité, l'expliquant par ce fait qu'elle l'avait connu tout petit,
qu'elle ne pouvait s'habituer à ce qu'il fût un homme, un vrai
homme; et elle s'inquiétait s'il viendrait cette année, au château de
monsieur son oncle, à Monneville, où, depuis six ans bientôt, on
n'avait pas eu le plaisir de sa visite.
—D'ailleurs, à Paris, vous ne sortez guère davantage ... Et même soit
dit sans reproches, nous vous avons bien regretté chez nous, cet
hiver ... C'étaient des réunions tout intimes, où je crois que vous ne
vous seriez pas ennuyé ...
Mareuil, qui guettait l'attaque, riposta:
—Moi, j'en suis sûr, madame ... Ça été une coïncidence malheureuse
... Par hasard, ces deux fois-là j'étais pris ... Mais j'espère ...
Mme Lepassereau l'interrompit:
—Vous ne reconnaissez pas Germaine, n'est-ce pas?
Et d'un ton bienveillant:
—Oh! cela ne m'étonne pas ... En six ans, on change!... C'est bien
naturel ... Et toi, Germaine, tu ne reconnais pas M. Mareuil, toi non
plus?
Mlle Germaine Lepassereau se retourna et, saluant d'un salut grave
et réservé:
—Non, je n'aurais pas reconnu Monsieur.
Elle avait une allure assez gracieuse, des yeux clairs et larges, des
cheveux châtain pâli, frisés en triangle sur le front; mais sans qu'elle
fût déplaisante, il n'y avait rien de troublant dans sa petite figure
froide, lisse et propre de jeune miss bien savonnée.
Elle reprit, après un moment:
—Maintenant, je me rappelle, je me rappelle parfaitement.
Une fanfare sonnait, annonçant la fin d'un parcours et se mêlant au
grondement lointain des derniers applaudissements, tandis qu'un
nouveau cavalier, un hussard stoppait devant la tribune du jury.
—Regarde, maman, dit Germaine sans laisser le temps à Mareuil de
trouver la phrase courtoise qu'il cherchait ... Regarde! Voilà M. de
Saint-Lys ... C'est la troisième fois qu'il monte aujourd'hui.
Mareuil saisit l'occasion de réparer son silence:
—Cela vous intéresse, ce concours, Mademoiselle?
Mlle Lepassereau eut une moue ironique:
—Heu?... Cela m'intéresse autant que d'aller faire des visites ou que
d'aller à la Sorbonne ...
—Ah! vous fréquentez la Sorbonne? Et vous vous y ennuyez,
Mademoiselle?
Il souriait le plus sympathiquement qu'il pouvait.
Elle leva les yeux pour voir s'il se moquait; puis subitement, son
regard sembla se voiler de défiance, se refermer sur ce qu'elle
pensait, et elle répondit d'un ton bref comme un tour de clef:
—Je n'ai pas dit cela!
Elle s'était tournée vers la piste et affectait de s'occuper de la course
de M. de Saint-Lys, marquant au crayon, sur son programme, les
fautes commises, tapant le sol du bout de son en-cas, lorsque la
haie ou la barrière avaient été convenament franchies.
Gilbert, posté derrière, détaillait hostilement son buste enserré d'un
long covercoat jaune, ses cheveux trop tirés sur la nuque où nul
frison ne dépassait, toute sa netteté frigide de novice inexperte, et il
éprouvait pour elle des sentiments de pédant d'amour, le mépris du
savant pour l'illettré qui lui a manqué.
Plus tard, quand elle aurait accompli ses preuves, oui, il eût compris
qu'elle le reçût dédaigneusement, de cet air de majesté hautaine
que donnent parfois aux femmes la conscience de leur mystérieuse
valeur, le souvenir récent de ce qu'elles peuvent, avec leur corps!
Mais, aujourd'hui, qu'elle l'accueillît ainsi, lui, Gilbert, l'ancien amant,
l'ancien adversaire d'une gaillarde telle que Mme Hardouin, qu'elle fît
sa contractée, cette petite Lepassereau qui ignorait tout de la vie,
qui n'avait d'autre mérite que sa virginité fade, non, c'était pénible,
c'était sévère!
Et pendant que M. de Saint-Lys terminait ses bonds, Mareuil
parcourait du regard les dames proches, pour en découvrir une de
laquelle il eût subi, sans récriminer, les intolérables façons de cette
petite glaçon de Lepassereau. Tout autour de lui, tout au loin, se
dressaient des têtes familières, têtes d'hommes, de femmes, de
jeunes filles, inaperçues depuis deux ans et dont il déchiffrait, peu à
peu, les traits changés par le temps. Après cet exil de sa liaison, il se
faisait l'effet d'un voyageur revenant à Paris après un long voyage. Il
reconnaissait un nez, une bouche, une attitude, puis le nom fuyait.
Ou bien, il hésitait, croyait s'être trompé. Soudain, il s'inclina avec
vivacité vers Mme Lepassereau qui se passionnait pour la course de
M. de Saint-Lys, et demanda:
—Madame!... Madame!... Est-ce que cette dame blonde, à droite,
au-dessous de vous, au second banc, est-ce que ce n'est pas Mme
Lozières?
—Où cela?
—Là, à droite, au deuxième rang, un chapeau à coques de velours
grenat ...
Il désignait une jeune femme blonde, au visage ovale et pâle, aux
sourcils noirs très épais, et dont les cheveux ondulés recouvraient à-
demi l'oreille.
—Où cela?... Où cela?... répétait Mme Lepassereau dont l'attention,
loin de travailler vers la droite, restait captivée, à gauche, par la
double haie qu'atteignait, à cet instant, l'infatigable Saint-Lys.
Enfin, elle répondit:
—Oui, oui, c'est Mme Lozières ... C'est elle!
Mareuil pensait:
«Eh bien! en voilà une à qui le mariage a fait du bien!»
Il se rappelait sa silhouette maigre et informe de jeune fille, d'enfant
même, à l'époque où il jouait avec elle, dans le verger vert de
Monneville, sous les pommiers tordus; et il l'examinait tout charmé
de son épanouissement nouveau, de sa beauté grandie, lui trouvant
un curieux air petit fifre de la Révolution, petit tambour Bara, avec
ses cheveux dorés sur l'oreille.
Mais une rumeur triomphale, une immense explosion de bravos
venaient d'éclater, saluant le hussard qui avait sauté d'un superbe
saut la rivière du centre. Une musique militaire entonna une marche
d'opérette. La réunion était close.
La piste, rapidement, se remplit de spectateurs et de spectatrices
accourant pour assister à la distribution des récompenses. Près de la
rivière, des groupes se formaient, joyeux et bavards, comme à la
sortie des grandes administrations, la journée de travail finie. Des
gigolos, en longue redingote, mesuraient, d'un œil effaré, la largeur
de l'obstacle. Les femmes, sans interrompre leur causerie,
s'entrejugeaient furtivement, le front impitoyable, notant les erreurs
de mode ou les inventions habiles dans les toilettes qui circulaient,—
et des figures s'avançaient, interrogatives et plus indicatrices que
des doigts, des figures priant qu'on leur nommât ce monsieur avec
une barbiche rousse ou cette dame en vert, là-bas.
Au milieu de la foule, Mareuil, qui avait profité de la presse du
départ pour semer Mme Lepassereau, se promenait, l'aspect
indifférent et ennuyé, quoiqu'en réalité tout à la préoccupation de
revoir Mme Lozières, de vérifier si elle était de près la jolie personne
qu'elle lui avait semblé de loin.
Elle devait avoir actuellement deux ans de moins que lui, quelque
chose comme dans les vingt-huit ans; et on l'avait mariée, sept ans
avant, avec un haut fonctionnaire des finances, un receveur, un
trésorier-payeur,—Mareuil ne se souvenait plus au juste,—un
républicain de vieille date, dont le choix avait fait scandale, lors du
mariage, à Monneville, et dans toute la société conservatrice des
environs.
Un mariage qui s'expliquait pourtant, étant donné l'ingénuité docile
de la jeune fille et les ambitions de sa famille, dont plusieurs
membres déjà, dans la diplomatie ou l'armée, s'étaient un peu ralliés
au gouvernement. Mais on avait mis longtemps, néanmoins, à le
pardonner aux Brégy, à oublier leur fâcheux manquement à la bonne
cause en péril ...
Gilbert évoquait ces souvenirs: «Lucie de Brégy!... Lucie Lozières!
Est-ce bizarre, cette rencontre!»
Tout à coup, il remarqua les coques rouges de son chapeau, et
lentement il se dirigea de son côté, de façon à ne pas être masqué
par les dames avec qui elle causait. Arrivé devant elle, il salua; puis,
après quelques pas, s'étant retourné pour la contempler de nouveau,
il aperçut le regard de Mme Lozières fixé vers lui, qui se dégageait
vite, comme par honte d'avoir été surpris.
Alors, il rôda de groupe en groupe, s'arrêtant à des poignées de
main, à des conversations inutiles, retenu par le désir de s'approcher
encore de la jeune femme, de renouer connaissance, si possible,
sans but précis, pour voir, et parce qu'il en avait envie, tout
simplement, en somme.
Mais elle avait changé de place, était partie peut-être. Il ne put la
retrouver.
«C'est stupide, pensait-il, j'aurais dû lui parler ... Quelle gaffe!»
Il chercha partout, traîna autour des sauteurs primés, faillit se faire
écraser par les lauréats qui se rendaient aux écuries en trottinant, et
finalement, à bout de patience, il gagna la sortie.
La bousculade était grande, et, par moments, la masse serrée des
partants se serrait davantage pour laisser passage à des chevaux
qu'un lad emmenait dehors.
—Hep! hep!
On se rangeait, on s'étouffait, on se collait les uns aux autres, en
rentrant les pieds. Des dames protestaient. D'autres souriaient à
leurs voisins inconnus ou bien montraient une mine revêche aux
privautés éventuelles. Une voix, derrière Mareuil, une voix claire et
gaie prononça:
—Bonjour, monsieur Gilbert!
Il tourna la tête et vit Mme Lozières qui, le coude au corps, lui
tendait, avec difficulté, la main. Il balbutia, un peu décontenancé par
cette apostrophe inespérée:
—Bonjour, Madame!... Vous allez bien?... M. Lozières va bien?...
Quelle foule, n'est-ce pas?
—Très bien, je vous remercie ... et Madame votre mère?
Il reprit:
—Je vous croyais à Bourges?
—Oh! non ... Nous avons quitté Bourges il y a trois mois ... Je suis
tout à fait Parisienne, maintenant ... M. Lozières a été nommé au
ministère, à l'administration centrale ...
—Ah! c'est une bonne chose! fit par courtoisie Mareuil. Vous devez
être très contents!
—Oui, nous sommes assez contents ... Oh! oh!... Prenez garde!
Une poussée l'avait jetée contre lui, et il sentit une odeur fine et
forte, cet heureux alliage du parfum et d'elles-mêmes qu'exhalent
certaines femmes toujours.
Elle murmura:
—Je vous demande pardon ... Je ne vous ai pas fait mal?
—Du tout, du tout!... Mais ce service est bien tristement organisé ...
Ils arrivaient sur le seuil.
Une large nappe de lumière jaillit devant eux, et ils aperçurent des
laquais en livrée et l'œil inquiet, à la recherche des maîtres, une
triple rangée de badauds qui se penchaient avidement dans l'espoir
de ces femmes, et ces femmes, et ces femmes qui sortaient.
Mme Lozières regarda à droite, à gauche—et d'un ton contrarié:
—Comme c'est ennuyeux!... J'ai perdu mes amies, les dames avec
qui j'étais tout à l'heure!
Elle se haussait sur la pointe des pieds:
—Non!... non, elles ne sont pas là ... Elles seront parties sans
m'attendre!
—Alors? fit-il brièvement.
—Alors, je vais vous dire au revoir et rentrer chez moi ... Je reçois le
samedi ... Mais je n'ai plus que deux jours à votre disposition,
puisque samedi en huit mes réceptions finissent ... N'oubliez pas, 9,
rue Galilée, près de l'avenue Kléber ...
Il retint un peu sa main et, d'une voix respectueusement caressante:
—Est-ce qu'il vous déplairait que je vous accompagne?
—Quand cela?... Maintenant?... Ce n'est guère votre chemin, il me
semble ...
Elle s'arrêta, inspecta vivement les alentours, et reprit:
—Enfin, si vous voulez bien vous déranger de votre route, j'aurais
mauvaise grâce à refuser.
Ils quittèrent le trottoir, traversèrent la chaussée, puis, fendant la
foule entassée sur le refuge voisin et qui les guignait déjà, comme
si, de longue date, ils eussent eu ensemble toutes sortes d'intimités,
ils remontèrent, au pas de promenade, les Champs-Elysées, sans
mot dire, malicieusement satisfaits du petit pacte audacieux qu'ils
venaient de conclure.
Mme Lozières rompit, la première, le silence:
—Vous travaillez beaucoup? J'ai vu de vous des choses délicieuses à
l'Exposition du Blanc et Noir, l'an dernier, et aussi au Grand-Art ...
Mareuil se défendit modestement.
—Non, je vous assure, continua Mme Lozières, j'aime énormément ce
que vous faites ... Vous avez une façon de poser les personnages et
surtout des teintes d'une délicatesse!...
Elle lui parlait familièrement, tout de suite revenue au ton de
camaraderie de jadis. Elle citait ce qu'elle préférait parmi ses pastels,
ses esquisses formulant même des critiques, s'embarrassant parfois,
manquant des termes exacts, mais lâchant moins de niaiseries qu'il
n'en eût pu craindre.
Mareuil déclara en raillant:
—Vous avez du goût ... Permettez-moi de vous le dire ...
Elle remercia d'une inclinaison de tête, et ils allèrent de nouveau
quelque temps en silence.
Mareuil observait sournoisement Mme Lozières de son regard expert
de peintre et d'amant sagace.
Elle lui paraissait aussi séduisante que là-bas, à l'Hippique, gardait
cet air petit fifre qui, dès l'abord, lui avait plu. Et, quoique d'une
taille plutôt grande, elle avait de la grâce, elle marchait bien, de ce
pas aisé, harmonieux, solide, qui distingue les femmes adroites et
sûres de leur corps.
Il l'évalua encore d'un coup d'œil sommaire:
«Elle est bien, cette petite!... Quel dommage que je ne l'aime pas,
que j'aie l'esprit ailleurs!...»
Mais Mme Lozières sentait probablement sur elle la pesanteur des
regards de Mareuil, car, comme pour mettre fin à ce gênant examen,
elle demanda:
—Savez-vous comment s'appelle la personne qui passe dans cette
victoria bleue?...
C'était une des demoiselles de la Butte, une grosse brune avec un
chapeau empanaché de blanc. Mareuil répondit sèchement:
—Elle se nomme Mériel ou Nivolas ... je ne vous garantis rien ... Du
reste, si vous voulez des renseignements sur ces dames, vous vous
adressez fort mal ... je n'en fréquente aucune.
Mme Lozières parut aguichée par l'expression mauvaise des paroles
de Mareuil:
—Tiens, pourquoi?... Il y en a cependant de bien jolies?
—Je ne dis pas ... Mais, je n'aime pas cette façon d'aimer ...
Elle s'exclama d'une voix incrédule:
—Vous voulez donc de l'amour, de l'amour vrai?
—Peut-être!
—C'est surprenant! Je n'aurais pas cru cela de vous!... Je ne sais
comment ... Est-ce parce que je vous vois encore petit garçon,
turbulent et méprisant pour les filles, comme vous disiez?... Mais je
n'aurais jamais supposé que vous fussiez un amoureux, un
sentimental....
Mareuil répliqua:
—C'est pourtant comme cela!
Puis, après un silence:
—Et même, savez-vous de quel nom on m'a surnommé chez un de
mes amis?
—Je n'ai pas idée ...
—On m'a surnommé Soif-d'Amour!
Elle eut un petit rire cordial.
—Soif-d'Amour! Oh! c'est drôle, c'est très drôle, Soif-d'Amour.
Evidemment, cela prouve en votre faveur ... Vous allez me trouver
bien indiscrète ... Est-ce qu'il y a longtemps que vous êtes ainsi?
—Je l'ai été ...
—Ah! vous ne l'êtes plus?...
—Plus pour le moment ...
Elle ne répliqua pas, la figure égayée d'un mince sourire, la tête
baissée vers le bitume grisâtre, mais la pensée visiblement tendue
vers d'autres choses, vers tout ce que Mareuil ne lui avait pas révélé
et qu'elle eût bien voulu connaître.
Il réfléchissait: «Pourquoi est-ce que je lui confie ces histoires?... Je
ne tiens pas à me faire admirer, et j'ai l'air de me poser en être
extraordinaire ... C'est imbécile!»
Le temps avait fraîchi. La nuit tombait. Une rafale de vent s'éleva,
vint coller la souple robe de Mme Lozières contre son corps.
«Elle n'est décidément pas mal, songeait Mareuil, et cela vaudrait
toujours mieux qu'une Mériel, qu'une Kerjeu, qu'une Nivolas ...»
Elle l'interrompit dans ces parallèles:
—Vous autorisez une question?
—Tant qu'il vous plaira!
—Eh! bien, voilà! dit-elle, comme livrant la conclusion de ses
raisonnements secrets ... Voilà! Pensez-vous qu'un jeune homme de
votre âge puisse aimer d'une manière absolue ... Vous m'entendrez à
demi ... C'est très difficile à expliquer ... Pensez-vous qu'un jeune
homme peut être fidèle, complètement fidèle, même quand il aime,
au milieu des entraînements, des tentations, des occasions?...
—Certainement! fit Mareuil.
Elle reprit:
—Oh! ce doit être bien rare ... Ainsi, par exemple, ceux que nous
avons rencontrés tantôt au Concours, tous ces officiers, ces hommes
de sport, pensez-vous ...
Mareuil s'exclama rageusement:
—Ceux-là! ceux-là!... Ah! mais non!... Et puis, je préfère que vous ne
m'en parliez pas, des jeunes gens ... Je les déteste!
—Vous les détestez?
—Oui, je les déteste, surtout lorsque je les vois en foule, en masse,
réunis ensemble, comme ils étaient aujourd'hui ... Tenez, il m'est
arrivé de me sauver d'un salon de cercle, d'une salle de théâtre,
d'une salle d'armes, parce qu'il y avait là trop d'hommes, parce que
cela m'indignait de rêver à tout ce qu'ils étaient capables de faire
avec leurs yeux, leurs moustaches, leur vigueur, leurs corps alertes
... Ce sont des voleurs de cœurs, des voleurs de femmes ... Je les
déteste!...
Elle lui lança un regard ému, un de ces instinctifs regards de
gratitude comme elles en ont toutes quand on dit, à leur propos, des
paroles même un peu absurdes, mais qui sont plus que des mots de
désir brutal ou d'offres libertines.
Mareuil continuait:
—Non, je ne comprends pas qu'on trahisse ... Je trouve ça puéril ...
Il serait si simple de quitter quand on cesse d'aimer ... Pour ma part,
je n'ai jamais trompé ... et je suis persuadé, je crois bien que je ne
tromperai jamais ... oui, je le crois bien ...
Il avait proféré cela comme une leçon récitée, sans conviction dans
la pensée, et il demeurait stupéfait de sentir en lui ce subit
désaccord, répétant:
—Oui, réellement, je le crois!
—C'est très bien! fit Mme Lozières ... Et si vous ne redoutez pas de
vous ennuyer, vous viendrez en recauser chez moi, samedi prochain
...
Il s'écria d'un air de regret:
—Comment! vous rentrez?
—Oui, je rentre ... D'abord, ce vent me donne la migraine ... Et puis
il est très tard ...
Mareuil prit, en plaisantant, un ton de mélodrame:
—Ecoutez!... Suivez-moi et je vous raconterai ma vie!
Mme Lozières souriait:
—Mais je ne vous la demande pas!
—Alors, je vous raconterai autre chose, tout ce que vous voudrez ...
Il est à peine six heures et demie ... Vous avez largement le temps!
Et comme elle hésitait, il feignit de l'amertume:
—Quand je pense qu'à Monneville, c'était toujours vous qui me
relanciez, qui me suppliiez de jouer ...
Elle répondit:
—Nous ne sommes plus à Monneville ... Enfin, je ne veux pas que
vous vous imaginiez que je me venge ... Entendu!... Nous
marcherons encore un peu, mais quelques pas, seulement!
Ils se remirent en route, traversèrent la place de l'Etoile et
s'engagèrent dans l'avenue de la Grande-Armée.
Le crépuscule devenait plus noir, et une à une, les lanternes des
réverbères s'allumaient. Lorsqu'ils passaient auprès d'elles, Mareuil
s'amusait de voir son ombre et l'ombre de Lucie s'allongeant, côte à
côte, comme celles de deux amants flâneurs. Parfois, à une
bourrasque plus violente, elle détournait la tête de son côté et elle
échangeait avec lui un sourire amical, un sourire complice.
—Non, je n'ai jamais, jamais trahi! disait Gilbert.
Il avait repris ses théories sur l'amour, sur ce qu'on doit à une
femme aimée, sur le dégoût que nécessairement on éprouve pour
les autres, qui ont le tort de ne pas être elle—décrivant à Mme
Lozières tout son trésor de sentimentalité, inventoriant toutes ses
richesses de cœur, les faisant scintiller, miroiter, valoir comme un
bijoutier ses bijoux devant une cliente à gagner.
Il avait conscience d'être sincère maintenant, et l'indécision bizarre
de tout à l'heure ne l'oppressait plus. Il parlait de son irréprochable
fidélité ainsi qu'un brave de son courage, avec cette assurance
naturelle qui nous vient d'un passé glorieux, d'actions d'éclat
incontestables. Chacune de ses phrases se rapportait à des hauts
faits qu'il eût pu prouver, à l'un des épisodes de la terrible campagne
de deux ans qu'il avait menée contre Jack. Il s'écriait à tout instant:
«Quant à moi ...» ou bien «Il m'est souvent arrivé» ou encore:
«Sans me citer comme exemple ...»; et sous ces formules
personnelles qui lui échappaient, Mme Lozières devinait une histoire
vraie, un drame touchant, peut-être.
Elle n'interrogeait plus, le laissait disserter, craignant de prononcer
des questions significatives, de trop montrer son intérêt, sa
sympathie.
Ils étaient près de la Porte-Maillot. Elle vit la grille, et d'un ton
impératif:
—Il faut que je rentre ...
—Il le faut?...
—N'insistez pas!... Cela me désobligerait.
Ils revinrent en arrière. Mme Lozières hâtait le pas, se plaignant
d'être en retard, très fâchée, très mécontente.
Mareuil, pour la calmer, consultait sa montre, donnait de fausses
heures auxquelles elle refusait de croire.
A l'angle de l'avenue Kléber, elle s'arrêta:
—Si vous voulez, j'aime mieux que nous nous séparions ici!
Puis, Mareuil, serrant un peu fort sa main, elle la retira
brusquement:
—A bientôt, j'espère!
Elle n'indiquait plus de jour, cette fois. Il répliqua de même:
—A bientôt!... Je suis bien heureux de vous avoir revue!
Elle le regarda d'un regard franchement tendre, presque prometteur
et elle ajouta:
—A bientôt alors!... Je me sauve!
Elle avait disparu au coin d'une rue.
Mareuil songea:
«Elle est agréable ... Mais quels yeux bêtes en partant!... Quels
regards sales elles peuvent vous avoir!... C'est écœurant!...»
Il monta dans un fiacre et se fit conduire chez lui. Auprès du parc
Monceau, il distingua, malgré la nuit, un endroit où jadis, avec Jack,
il s'était promené, aux premiers jours de leur liaison. Il poussa un
soupir, et resta quelque temps avec une vague mélancolie.
 V
—On peut entrer?
—Entrez! dit Gilbert, en reconnaissant la voix de Brévannes ... Vous?
Un samedi?... Je croyais que c'était votre jour d'article?
Le journaliste s'assit et alluma un cigare:
—Oui, j'ai eu fini plus tôt que de coutume ... Dites donc, qu'est-ce
que vous faites, aujourd'hui?
—Pourquoi cela?
—Enfin, êtes-vous libre?
—Libre!... Libre!... Ça dépend! dit Mareuil, qui s'était précisément
promis d'aller, dans l'après-midi, chez Mme Lozières ... Ça dépend un
peu de l'heure et aussi de ce que vous avez à m'offrir ...
—Voilà! fit Brévannes. Il y a tantôt répétition générale à l'Odéon ...
La jeune enfant est souffrante et ne peut venir ... Voulez-vous la
remplacer? On raconte que la pièce est intéressante ...
Mareuil se recueillit. Cette proposition ne le séduisait guère, mais
une soudaine envie le poussait à profiter du prétexte pour retarder
cette visite de combat que, la veille encore, il hésitait à rendre, à
livrer.
Il pensa avec soulagement: «Il ne serait peut-être pas mauvais de la
faire attendre ...»—puis, tout haut:
—C'est convenu! Cela me va!
—Alors prenez votre chapeau ... Nous mangerons de l'autre côté de
l'eau ... Ne lambinons pas!... Le rideau est à une heure.
Lorsqu'ils furent en voiture, après quelques instants de silence,
Brévannes demanda:
—A propos, et la santé du cœur?
—Mon Dieu, dit Mareuil ... je n'ai pas trop à me plaindre ... Je suis
même étonné de souffrir si peu ... A tel point, que je fais des
expériences, que je me mets à l'épreuve pour voir si c'est bien fini, si
ce n'est pas un engourdissement passager ... Je me représente la
dame en train d'accomplir des infamies, des débauches
extraordinaires ... Rien! Cela me laisse calme, cela augmente
seulement un petit peu la répulsion que j'ai pour elle ... Je lui crie un
gros mot, une injure ... Mes lèvres se soulèvent de dégoût ... Et c'est
tout!... Quant à ces messieurs ...
—Quels messieurs?
—Vous savez bien, le boursier, l'avocat, la bande, enfin ...
—Eh bien?
—Eh bien! je ne leur en veux presque plus ... Je n'ai plus pour eux
qu'une sorte de curiosité ... Oui, tenez, je voudrais bien voir leurs
têtes, leurs têtes de cocus ... Car ils l'ont été, comme moi, les
imbéciles, comme tout le monde le sera avec la dame!... Je
commence à la comprendre, cette personne, je commence à la
connaître!...
Brévannes approuva:
—Votre état d'esprit me semble, en effet, assez bon ... Et alors, vous
ne l'aimez plus?...
—On le dirait ... C'est, d'ailleurs, ce qui me confond, mon vieux ...
Car, pourquoi est-ce que, du jour au lendemain, je ne l'ai plus
aimée? Parce que j'ai su qu'elle me trompait? Je l'ai cru d'abord.
Mais, ensuite, j'ai réfléchi ... Ce n'est pas la raison ... J'aurais pu
savoir et pardonner, continuer de l'aimer ... J'aurais pu désirer la
revoir, accepter qu'elle revînt comme elle m'en priait ... Pas du tout,
j'ai refusé ...
—Et maintenant, vous refuseriez encore?... Vous refuseriez si elle
vous écrivait, si elle vous donnait pour demain ou après-demain un
petit rendez-vous amical?
—Je refuserais, parce que je n'ai plus envie d'elle ... Et puis, pas de
danger qu'elle m'écrive ... Elle est sans doute enchantée que ce soit
terminé, d'avoir sa tranquillité, sa liberté ... Elle ne viendra pas me
chercher, allez ... Mais vous, Brévannes, vous expliquez-vous cela,
que je ne l'aime plus?...
Brévannes tiraillait sa longue moustache blonde.
—Moi?... Moi? Je ne me suis jamais engagé à vous fournir des
éclaircissements psychologiques sur votre cas ... Comment!... Vous
ne souffrez plus, vous avez la chance de ne plus souffrir et vous
demandez, en sus, des explications!... Vous en voulez trop, mon
jeune ami! Vous êtes insatiable!
Mareuil suivait son idée:
—Par contre, si je ne souffre pas, je m'ennuie ferme!... J'ai essayé
de travailler ces jours-ci. Pas moyen! Est-ce affaire de tempérament
ou d'habitude? mais je sens que je ne suis capable que d'aimer, que
je ne suis bon qu'à cela, qu'il n'y a que cela qui m'intéresse ... Je
suis devenu homme de sentiment, comme d'autres sont hommes de
cheval, hommes de finances, hommes d'études ...
Et il répéta, ravi de sa trouvaille:
—Oui, je suis un homme de sentiment!... Saisissez-vous?
Brévannes répliqua d'une voix narquoise:
—Si je saisis?... Passez-moi donc une allumette ... Si je saisis?...
Comme un huissier!... Mais on n'est pas homme de sentiment tout
seul. Vous oubliez, mon brave monsieur, qu'on ne l'est généralement
qu'à deux ... c'est le minimum!
Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.

More than just a book-buying platform, we strive to be a bridge

connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.

Join us on a journey of knowledge exploration, passion nurturing, and

personal growth every day!

ebookbell.com