Sisalermkiat Phandanouvong

6230080004

Mini-Project Final Report

URBAN SPRAWL DETECTION IN VIENTIANE CAPITAL, LAO PDR SINCE 2000-2020:
USING REMOTE SENSING DATA
The Joint Graduate School of Energy and Environment, King Mongkut's University of
Technology Thonburi
GIS and Remote Sensing (JEE-684)
Sisalermkiat Phandanouvong (Student ID: 62300800004)

ABSTRACT
Mapping of land-use and land-cover (LULC) changes are regularly referred for sustainable development,
planning, monitoring, and management. This present study is an effort to explore and detect the urban
LULC changes in Vientiane Capital, Lao PDR from 2000 to 2020. The urban area of Vientiane has seen
tremendous growth during the last decades. The ENVI 5.2 has been used to detect the urban LULC
changes in Vientiane Capital by using data from Landsat TM-5 and Landsat OLI-8 for satellite images from
2000 to 2020. After preprocessing Vientiane Capital imagery, supervised classification using Support
Vector Machine (SVM) classifier was applied to prepare LULC maps of study area. The generated data
of both change detection techniques from 2000 to 2020 revealed the interesting aspect that the urban area
and agriculture area of Vientiane has been increased by 19.65% and 4.24% respectively. However, the
forest area, bare land has been declined by 17.56% and 5.79% respectively and there was not much change
in the water with a decrease by 0.54%.

Keywords: Land use and land cover, remote sensing and GIS, change detection, Vientiane Capital.

1. Introduction

Lao People’s Democratic Republic (Lao PDR) has a land area of 236,800 square kilometers and the gross
Domestic Product (GDP) growth more than 6% in last five years 2014 to 2019 which is one of the fastest-
growing economies in ASEAN, according to the World Bank [1]. Lao PDR is a sole landlocked country in
both the Greater Mekong Sub-region (GMS) and the Association of South-East Asia Nation (ASEAN),
bordering on China to the north, Vietnam to the east, Cambodia to the south, Thailand to the west and
Myanmar to the north-west, Vientiane is the capital of Lao PDR, also the center of the economic corridors
(CEC) among the ASEAN nations and stimulate the highest income from the international trade and tourist
within the country [2].
Moreover, urban sprawl, a consequence of socio-economic development under certain circumstances, has

1
 Sisalermkiat Phandanouvong
6230080004
increasingly become a major issue facing many metropolitan areas [3]. Urban sprawl is often referred to
as uncontrolled, scattered suburban development that increases traffic problems, reduces local resources,
and destroys open space [4]. It is critically important to properly characterize urban sprawl in order to
develop a comprehensive understanding of the causalities of urbanization processes.
For urban planning, the LULC information is very crucial which includes; vegetation, urban hydrological
modeling, and urban land-use mapping. In general, information about the land-cover is been obtained
from satellites or aerial imagery using pixel-based classification techniques [5]. These techniques have
been used to characterize the spatiotemporal trends of urban sprawl using multi-stage images, providing
a basis for projecting future urbanization processes [6].
Remote sensing has increasingly been used as a source of LULC information at local, regional, and global
scales [7]. To investigate the actual data, analyze the satellite images processing, and created mapping,
the remote sensing, and GIS techniques has been applied to analyze and get the accuracy and precise
outcomes[8].

2. Study area

Vientiane Capital is the Central Economic Corridor (CEC), also the most active city due to the rapid
increase of projects for the future trade and the operating special economic zone in the area [9]. Vientiane
Capital is situated with 17°58’0.01” N latitude and 102°36’0” E longitude. The total area of Vientiane is
3,920 km2 [10]. Also, the total population of Vientiane was 642,000 people in 2015 with a population
density of 209 people per square kilometer and the total population of Vientiane was 683,000 people in
2020, also the urban population of Vientiane is significantly increasing every year.
For this study, satellite imagery has been used to identify the spatial and temporal patterns of urban land
expansion, and in detecting the land cover change in the selected region. Therefore, this current study area
aimed to map the expansion of urbanization in 7 districts of total 9 districts in Vientiane capital included
1)Sisattanak, 2)Xaysetha, 3)Chanthabouly, 4)Sikhottabong, 5)Xaythany, 6)Mayparkngum, and
7)Hadxaifong as shown in Figure 1.
The main objectives of this study as following:
 To study how much area has been used for urban sprawl in Vientiane Capital;
 To examine a methodology to improve the effectiveness of the urban sprawl of Vientiane Capital; and
 To address the objective, Landsat images, geographic information systems, and some secondary data
from urban organizations were deployed in this study.

2
 Sisalermkiat Phandanouvong
6230080004

Lao PDR

Figure 1. Study area in Vientiane Capital, Lao PDR

3. Data used

3.1. Data acquisition

The Landsat TM-5 and Landsat OLI-8 were chosen due to their affordability, availability, and medium
to high spatial resolution. Landsat TM-5 and Landsat OLI-8 data for 2000 and 2020, covering the study
area, were acquired from the United States Geological Survey’s (USGS) GloVis website. Level-one
terrain (L1T) data was provided by USGS in WGS-84 geodetic datum and it was downloaded in geo-tiff
format. Details about the provided data are given in Table 1.

Table 1. Specifications of Landsat TM-5 and Landsat OLI-8 imagery used in this study
No. Landsat Sensor Path/Row Format Date
1 Landsat 5 TM 128/47 Geo TIFF 03/11/2000
2 Landsat 5 TM 128/48 Geo TIFF 03/11/2000
3 Landsat 8 OLI 128/47 Geo TIFF 11/01/2020
4 Landsat 8 OLI 128/48 Geo TIFF 11/01/2020

3
 Sisalermkiat Phandanouvong
6230080004

3.2. Data preprocessing

The standard image preprocessing techniques have been applied for radiometric and atmospheric
correction from two satellite images (2000 and 2020). ENVI 5.2 software has been used for this data
analysis. First of all, the atmospheric correction has been applied using the dark object subtraction method.
After getting started with ENVI 5.2 software, the metadata files from Landsat TM-5 and Landsat OLI-8
data were selected respectively for absolute atmospheric correction: the conversion of DN to top of the
atmospheric correction unit (ToA) to bring all the images at common reference spectral characteristics.
When the radiometric calibration has done with the help of radiometric correction and finished the data
preprocessing, the study area shapefile was used to subset area from Landsat image. The final image after
correction and area subset as shown in Figure.2 and Figure.3 for both Landsat TM-5 and Landsat OLI-8.

Figure 2. Preprocessed imagery of study area in Landsat TM-5 (2000)

Figure 3. Preprocessed imagery of study area in Landsat OLI-8 (2020)

4
 Sisalermkiat Phandanouvong
6230080004

4. Methods
For urban sprawl detection in Vientiane Capital, an integrated approach has been used by following steps;
data acquisition, image preprocessing, image classification, accuracy assessment, change detection, and
producing the final maps. A framework of methodology for this study has been shown in Figure. 4 below.

Remote sensing data

(Landsat 5 and Landsat 8 data)

Data acquisition USGS

(GloVis)

Images processing

Reference data Image classification Data training

No
Accuracy assessment

Yes

Change detection
(Identification of LULC changes from year 2000 to 2020)

Change map

GIS
(The final maps of LULC changes)

Figure 4. Methodology for urban sprawl detection of 7 districts in Vientiane Capital, Lao PDR

5
 Sisalermkiat Phandanouvong
6230080004

5. Results
Supervised Classification
When the preprocessing of data was completed, different classification tools were used to classify the
LULC of the study area. For supervised classification, the region of interest (ROI) was selected for data
training according to the class scheme presented in Table 2.

Table 2. Classification scheme used for study area

Class Type Color Class Description

Urban Red All types of manmade structures: residential, industrial,
commercial and services; transportation and utilities; built-up
Agriculture Yellow gardens, parks and playgrounds, grassland, agricultural, vegetated
lands, agricultural lands, and crop fields
Bare Land Cyan Areas with no vegetation cover, sand, open space, and bare soils
Water Blue River, water reservoirs, lakes, and ponds
Forest Green Trees, natural vegetation

After the ROI selection, SVM was applied respectively to classify the LULC. The results of all the
classifiers for both Landsat TM-5 and Landsat OLI-8 were given in Figure.5 and Figure.6. It has found that
the SVM showed the best results compare to other supervised classification techniques such as Minimum
Distance, Maximum Likelihood and etc. for Landsat TM-5 data and Landsat OLI-8 data respectively.
The class statistics summary of Landsat TM-5 and Landsat OLI-8 data is shown in Table. 3.

Table 3. Summary of class statistics

2000 2020
Class Type Change (%)
Landsat 5 (km2) % Landsat 8 (km2) %
Urban 26.11 1.28 426.907 20.935 19.65
Agriculture 621.56 30.48 707.983 34.718 4.24
Water 71.27 3.50 60.312 2.958 -0.54
Bare Land 562.56 27.59 444.483 21.796 -5.79
Forest 757.75 37.16 399.560 19.594 -17.56

A supervised classification was found to be accurate in the LULC classification. Considering the
existence of complex, overlapping, and inaccessible LULC patterns in the study area, and lower accuracy

6
 Sisalermkiat Phandanouvong
6230080004
level achieved in other methods except for SVM for Landsat TM-5 and Landsat OLI-8 respectively.
For the accuracy assessment of the classified images, the overall accuracy and kappa coefficient were
calculated by using the confusion matrices command. The values of overall accuracy for LULC
classification using the SVM algorithm was 80.31% and 83.69%, with kappa coefficients of 0.75 and
0.8 for both Landsat TM-5 and Landsat OLI-8 in the year 2000 and 2020 respectively.

Figure 5. Final map of study area in Vientiane Capital, Lao PDR (2000)

Figure 6. Final map of study area in Vientiane Capital, Lao PDR (2020)

7
 Sisalermkiat Phandanouvong
6230080004
In addition, the ground truth reference points were collected (chosen through random sampling strategy)
within the study area. Finally, raster files were converted to vector files and were then modified in QGIS
software. The resulting classified classes were edited in QGIS software, according to the previously
mentioned five major land-cover categories (Table 2).

6. Discussion
The success in LULC changes analysis using remote sensing data is reliant on accurate geometric rectification
and radiometric. These pre-processing concerns on the regular present the most challenging points of change
detection studies with precise atmosphere correction and radiometric. Using the remote sensing is quite accurate
and precise data of one date to the other is reached, identified the earth’s surface changes lead to be positively
attributed to land conversion. For this study, the mosaic technique has been used to combine two Landsat images
reflectance produces significant confusion between land-use classes that maintain reflectance characteristics
related to those of land-cover types.
Moreover, the GIS was used to integrated analyses within the spatial database framework and is regularly
compulsory to assign land cover to applicable land-use designations. Land surface attributes are mapped into
the urban sprawl as estimated directly from the remote sensing imagery. Some areas of this study, that
have also found the difficulty to identify the ground surface such as bare land and agriculture area due to
in the summer there is less plantation caused the difficult investigation by own eyes. In this study at the
mapping of LULC on an urban area in Vientiane Capital, utilized Landsat TM-5 and Landsat OLI-8
incorporated with ancillary topographic data as input to do support vector classification classifier to
produce the final map of LULC in Vientiane Capital, Lao PDR.
The LULC changes in Vientiane Capital presented the rapid expansion of urban, also impacts the extents of the
forest and water area of the study region. Due to urban land expansion with a high cost in the last decade caused
the deforestation in the area near the main cities not only Vientiane Capital, but also the other cities in Laos.
Despite the government’s efforts to preserve the forestlands, deforestation is still continuing in response to the
dynamic increase in population and the high demand for using wood. Even though some people pressure-
induced expansion of urban land and weak enforcement of forest protection, is one of the main reasons for the
decline of forestland which obviously sees in the result of this study that the decrease of forest areas by 17.56%.
Most developing countries including Lao PDR, the main cities of Lao PDR rapid increase of urban areas in last
decade because the government opens country for the foreign direct investment (FDI) to stimulate the economic
growth lead the gross domestic product (GDP) growth more than 6% in 2019, which is reasonable to the results
of this study especially the expansion of urban areas in Vientiane Capital by 19.65% from 2000 to 2020.
8
 Sisalermkiat Phandanouvong
6230080004
However, at the present, the government strategies consider more for planting trees to expand the plantation
forests and using more stick rule in the practice to protect the preserve forests as well as promote the country to
be green growth country which people have to truly consider about it more in order to make things come true.
Also, the expansion of cities uses to give an acknowledge to everyone to concern on the environmental impact
with city development to make sure that the development of every country can be able to go with the sustainable
development goals (SDGs) which the United Nations (UN) has been established and all countries around the
world agreed to follow this strategy to make the world sustainable development.

7. Conclusion
The main focus of the current study was to understand and address the dynamics of LULC and urban
sprawl in Vientiane Capital, Lao PDR by analyzing the multi-temporal Landsat data. Using temporal
imagery from Landsat for the years 2000 and 2020, the study area has been classified according to the
LULC type by supervised classification using the Support Vector Machine (SVM) algorithm. By using
the combination of remote sensing and GIS software to get the accurate data, pre-processing, image
processing, post-processing, and produce the final map of LULC, these two are powerful software that
allows users to analyze and edit spatial information to composing and exporting graphical maps.
The accuracy of LULC classification for the method using SVM has been tested and it has been found
that SVM (for Landsat 5 data and Landsat 8 data) classification are more accurate than other supervised
classification techniques such as Parallelepiped, Minimum Distance, Mahalanobis Distance, and
Maximum Likelihood. The results revealed that urban area has been increased by 19.65% during the last
20 years from November 2000 to January 2020. Furthermore, it is also found that forest, bare land, and
water have been decreased by 17.56%, 5.79%, and 0.54% respectively. From obtained class statistics,
we can conclude that land use in Vientiane Capital is rapidly changing from other land use classes to the
urban resulting in a serious threat to sustainable development.

9
 Sisalermkiat Phandanouvong
6230080004

References
[1] Asian Development Bank, “Lao People’s Democratic Republic, 2017 – 2020 — More Inclusive
and Sustainable Economic Growth,” Ctry. Partnersh. Strateg., no. August, pp. 2017–2020, 2017.
[2] S. Phandanouvong and S. Nimsai, “An investigation and evaluation of cross-border truck
transportation between Nongkhai, Thailand - Thanaleng to Vientiane, Lao PDR,” Int. J. Supply
Chain Manag., vol. 7, no. 4, 2018.
[3] P. K. Mishra, A. Rai, and S. C. Rai, “Land use and land cover change detection using geospatial
techniques in the Sikkim Himalaya, India,” Egypt. J. Remote Sens. Sp. Sci., no. xxxx, pp. 1–11,
2019, doi: 10.1016/j.ejrs.2019.02.001.
[4] E. F. Lambin et al., “The causes of land-use and land-cover change: Moving beyond the myths,”
Glob. Environ. Chang., vol. 11, no. 4, pp. 261–269, 2001, doi: 10.1016/S0959-3780(01)00007-3.
[5] M. D. Hossain and D. Chen, “Segmentation for Object-Based Image Analysis (OBIA): A review
of algorithms and challenges from remote sensing perspective,” ISPRS J. Photogramm. Remote
Sens., vol. 150, no. February, pp. 115–134, 2019, doi: 10.1016/j.isprsjprs.2019.02.009.
[6] M. A. Mahboob and I. Atif, “Assessment of Urban Sprawl of Lahore, Punjab, Pakistan Using
Multi-stage Remote Sensing Data,” Sci. Int., vol. 27, no. 6, pp. 6219–6224, 2015.
[7] C. Homer et al., “Conterminous United States land cover change patterns 2001–2016 from the
2016 National Land Cover Database,” ISPRS J. Photogramm. Remote Sens., vol. 162, no.
February, pp. 184–199, 2020, doi: 10.1016/j.isprsjprs.2020.02.019.
[8] R. Pal and P. Pani, “Remote sensing and GIS-based analysis of evolving planform morphology of
the middle-lower part of the Ganga River, India,” Egypt. J. Remote Sens. Sp. Sci., vol. 22, no. 1,
pp. 1–10, 2019, doi: 10.1016/j.ejrs.2018.01.007.
[9] C. Agency and N. E. Co, “Preparatory Survey on Vientiane Logistics Park ( VLP ) Project ( PPP
Infrastructure Project ) In Lao P . D . R . Preparatory Survey on Vientiane Logistics Park ( VLP )
Project ( PPP Infrastructure Project ) In Lao P . D . R .,” 2015.
[10] Public Works and Transport Institute, “Vientiane Capital Urban Development Master Plan,”
Public Work. Transp. Inst., no. March, 2011.

10