International Journal of

Geo-Information

Article
Geospatial Analysis in Web Browsers—Comparison Study on
WebGIS Process-Based Applications
Rostislav Netek * , Tereza Pohankova, Oldrich Bittner and Daniel Urban

Department of Geoinformatics, Palacký University Olomouc, 77900 Olomouc, Czech Republic;

[email protected] (T.P.); [email protected] (O.B.); [email protected] (D.U.)
* Correspondence: [email protected]; Tel.: +420-585-63-45-84

Abstract: With the rapid development of internet technologies in recent years, the shift from the
desktop to the web platform can be seen within geospatial analysis. While analytical tools, such as
buffer or clip, are routinely used in desktop environments, WebGIS deals with geographic information,
including geospatial analysis, within the online environment. The main aim of this paper is to perform
a comparison and evaluation of vector-oriented online geoprocessing tools in a WebGIS environment,
supported by the development of a custom solution for geospatial analysis. The application called
GeOnline is developed and tested as a case study to demonstrate the availability of spatial analysis
tools within the web browser. It implements the specialized geospatial library Turf.js, which allows
using non-trivial geospatial analysis, such as intersect, clip or calculate centroids. It handles client-
side processes. Both a functionality comparison and performance testing are carried out, while the
paper primarily focuses on data-driven (data-based) analysis and not only on visual-driven (visual-
based) analysis. The comparative study evaluates five geospatial tools (ArcGIS Online, GISCloud,
CARTO, FOURSQUARE, GeOnline) and summarizes the solutions from different aspects, including
the number of supported operations. Finally, performance tests on GeOnline separately and among
alternative solutions are performed. While ArcGIS Online is considered the most comprehensive
solution on the market, GeOnline performs well compared to alternative solutions.

Keywords: web service; spatial analysis; geospatial; comparison study

Citation: Netek, R.; Pohankova, T.;
Bittner, O.; Urban, D. Geospatial
Analysis in Web
Browsers—Comparison Study on
1. Introduction
WebGIS Process-Based Applications.
ISPRS Int. J. Geo-Inf. 2023, 12, 374. Geographic information systems (GIS) have undergone a turbulent evolution over
https://doi.org/10.3390/ijgi12090374 the last two decades. They were literally revolutionized by the rise of web maps in
the new millennium, which started the routine processing and visualization of spatial
Academic Editors: Wolfgang Kainz
data in the form of interactive outputs and tools to more dynamic and accessible web-
and Christophe Claramunt
based solutions [1]. WebGIS platforms have emerged as powerful tools, which allow
Received: 10 May 2023 users to perform complex geospatial analyses and collaborate remotely through web
Revised: 31 July 2023 browsers. However, despite the advancements in web-based geospatial technologies, one
Accepted: 4 September 2023 significant challenge remains: the integration of various (raster, sensor, etc.) data analyses
Published: 7 September 2023 within WebGIS environments. With the enormous development of internet technologies in
recent years (canvas, data storages, responsive design, geolocation, etc.), more advanced,
mostly vector-oriented WebGIS-based solutions are also becoming increasingly popular [2].
Han [3] also highlights the “interaction with users and connections to a wider audience
Copyright: © 2023 by the authors.
and its advanced data integration capabilities”. These allow not only the publication but
Licensee MDPI, Basel, Switzerland.
also the analysis and processing of spatial data directly in the World Wide Web (WWW)
This article is an open access article
environment. This is a paradigm shift in GIS [4], where the entire infrastructure is being
distributed under the terms and
moved from desktop- (GIS) to web-based solutions (WebGIS).
conditions of the Creative Commons
Attribution (CC BY) license (https://
Tools supporting spatial analysis are taking spatial data processing to a new level.
creativecommons.org/licenses/by/
They bring a higher level of interactivity. Analytical data processing can be performed
4.0/).
in real time or based on individual user parameters (e.g., buffer distance, hot spot limit,

ISPRS Int. J. Geo-Inf. 2023, 12, 374. https://doi.org/10.3390/ijgi12090374 https://www.mdpi.com/journal/ijgi

ISPRS Int. J. Geo-Inf. 2023, 12, 374 2 of 18

attribute field selection, etc.) [5]. Geographic and geoinformatic analyses are now common
practices in the desktop environment, and they are used routinely. Similarly, web mapping
applications have become so established that they are now part of the daily habits of millions
of users [6] on a global, national and regional level within numerous settings [7–10]. The
next step reflecting the evolution of WebGIS (such as the combination of GIS within the
web platform) is the implementation of spatial features and advanced analysis directly
into online map libraries [11]. Cloud-computing-based applications are already following
this trend [12]. Applications such as ArcGIS Online already offer more or less analytical
functions in their environment, i.e., tools and functionalities, which were, until recently,
the prerogative of desktop-based solutions only. The list of simple operations, such as
line/area measurement, buffer or heat map generation, is being extended with non-trivial
geoprocessing operations, e.g., intersect, clip or calculate centroids. This functionality thus
takes the use and deployment of GIS tools to the next level for both laymen and experts [13].
Turf.js [14] is a geospatial web processing service within web-based GIS environments.
One of the limitations of Turf.js is its lack of support for raster data analysis. As a result,
tasks such as remote sensing, satellite imagery analysis and terrain modeling, which rely
heavily on raster data, cannot be performed using Turf.js. Despite this limitation, Turf.js
remains a valuable tool for vector-oriented geospatial tasks and contributes significantly to
the advancement of web-based geospatial analysis.
Despite the challenge of raster data analysis, the adoption of WebGIS has seen remark-
able progress. Several web-based geospatial applications have emerged, showcasing the
potential of online geospatial analysis—for instance, “52◦ North WPS” [15], which is an
open-source implementation of the Open Geospatial Consortium (OGC) standard. This
implementation provides a framework for developing web-based geospatial processing
services and allows users to execute geospatial processes over the web. “EnviroMonitor”
empowers environmental scientists to monitor real-time sensor data, analyze environ-
mental trends and make informed decisions using a web-based interface. Such examples
demonstrate the practicality and value of WebGIS in addressing diverse geospatial chal-
lenges.
The main contribution of this paper is the performance of a comparison and evaluation
of online geoprocessing tools in a WebGIS environment, focusing on vector data operations
supported by the development of a custom solution for geospatial analysis. Compared to
alternative comparative studies, such as those by Yue et al. [13] or Chen et al. [16], this study
is aimed toward end-user-oriented solutions. Therefore, the paper has two sub-goals. The
primary sub-goal is to perform an analysis and evaluation of online libraries supporting
geoprocessing tools in WebGIS, specifically focusing on ArcGIS Online, GISCloud, CARTO,
FOURSQUARE and GeOnline. A comparison of alternative solutions on the market is made
in terms of the functionality offered. In addition, testing is also conducted to measure the
speed and performance of individual geospatial operations. Based on previous comparison,
the secondary goal is to implement a custom solution for geospatial analysis through a
specialized geospatial library Turf.js. The mapping application through a user-friendly
interface will allow the performance of spatial operations, analysis and data manipulation.
Additionally, the development of the GeOnline application demonstrates the availability of
specialized geospatial libraries such as Turf.js for non-trivial geospatial analysis in a web
browser.

2. Geospatial Analysis in the Context of WebGIS

In general, the term WebGIS is accepted as a strategy for accessing and dealing with
geographic information through the internet environment [17]. WebGIS is covered in detail
in a number of publications, and it is a widely accepted term [18]. According to Dorman [19],
WebGIS is “a software and hardware configuration that enables the sharing of maps,
spatial data, and geographic operations using common web communication protocols
such as HTTP and WebSockets”. Meanwhile, Čepický [20] defines web mapping as “the
design, implementation, creation and delivery of maps using the WWW”; for WebGIS,
 2. Geospatial Analysis in the Context of WebGIS
In general, the term WebGIS is accepted as a strategy for accessing and dealing with
geographic information through the internet environment [17]. WebGIS is covered in de-
tail in a number of publications, and it is a widely accepted term [18]. According to Dor-
man [19], WebGIS is “a software and hardware configuration that enables the sharing of
ISPRS Int. J. Geo-Inf. 2023, 12, 374 3 of 18
maps, spatial data, and geographic operations using common web communication proto-
cols such as HTTP and WebSockets”. Meanwhile, Čepický [20] defines web mapping as
“the design, implementation, creation and delivery of maps using the WWW”; for
he extendshethe
WebGIS, definition
extends to includetoan
the definition extra analysis
include and processing
an extra analysis phase (“the
and processing phasedesign,
(“the
implementation, creation,creation,
design, implementation, delivery,delivery,
analysis and processing
analysis of maps using
and processing of maps theusing
WWW”). the
Sack
WWW”).[21], in a chapter
Sack [21], in dedicated to web mapping,
a chapter dedicated comes up comes
to web mapping, with a up“swoopy
with a diagram”
“swoopy
dedicated
diagram”to the categorization
dedicated of web maps.
to the categorization ofBased on a trio
web maps. of defined
Based parameters
on a trio of defined(purpose
param-
of
eters (purpose of the map, interactivity, visual hierarchy), he divides webfour
the map, interactivity, visual hierarchy), he divides web maps into mapscategories:
into four
WebGIS,
categories:Web Geovisualization,
WebGIS, Reference Web
Web Geovisualization, Map, Web
Reference Thematic
Map, Web Map Web
Thematic (see Figure
Map (see 1).
In this view,
Figure WebGIS
1). In this view,isWebGIS
characterized as highlyasinteractive
is characterized with a low
highly interactive withdegree of visual
a low degree of
hierarchy and withand
visual hierarchy an emphasis on exploring
with an emphasis implications,
on exploring not just as anot
implications, tooljust
for as
presentation
a tool for
purposes.
presentation purposes.

Figure1.1.Categorization
Figure Categorizationof
ofweb
web maps
maps by
by Sack
Sack [21].
[21].

WebGIS
WebGISemphasizes
emphasizesthe theprocessing,
processing,storage
storageandandanalysis
analysis ofof
spatial
spatialdata [22],
data including
[22], includ-
their algorithms.
ing their It is undoubtedly
algorithms. It is undoubtedly related to thetofield
related the of webofcartography,
field and the
web cartography, and topics
the
overlap and intersect to a large extent. In fact, “WebGIS”, “web cartography”
topics overlap and intersect to a large extent. In fact, “WebGIS”, “web cartography” and and “web
mapping”
“web mapping”are oftenareused
ofteninused
a similar
in a sense,
similaralthough they cannot
sense, although they be considered
cannot synony-
be considered
mous. The exact
synonymous. boundary
The between between
exact boundary the topicstheis difficult
topics istodifficult
define; to
web cartography
define; web cartog-uses
web maps
raphy usesasweb a presentation medium [23]medium
maps as a presentation and focuses
[23] onandvisualization, while WebGIS
focuses on visualization, has
while
recently
WebGISacquired
has recentlymore and more
acquired moredataandand analytical
more data and functions.
analytical The implementation
functions. The imple- of
new visualization
mentation of newmethods and analyses
visualization methodsisand theanalyses
latest trend in the
is the direction
latest trend inoftheGISdirection
into the
future
of GIS[24].
into Veenendaal
the future [24].et al. [25] add that
Veenendaal “WebGIS
et al. [25] addisthatthe“WebGIS
most commonis the form
most of online
common
GIS”; according to Hojati [26], WebGIS is a subset of GIS.
form of online GIS”; according to Hojati [26], WebGIS is a subset of GIS.
Netek
Netek[27]
[27]operates
operateswithwith the
the term
term WebGIS
WebGIS as as aa combination
combination of of the
the separate concepts
separate concepts
of
ofWorld
WorldWideWideWeb WebandandGISGIS(WebGIS
(WebGIS== WWW WWW ++ GIS).
GIS). From
From a a technological perspective,
technological perspective,
the
theconcept
conceptcombines
combines state-of-the-art
state-of-the-art methods
methods in in the
the context
context ofof geographic
geographic information
information
technologies
technologies(GIT),(GIT),principally
principally based on the
based on methods
the methods of service-oriented architecture
of service-oriented (SOA),
architecture
cloud computing and the internet as a platform. GIS affects all processes
(SOA), cloud computing and the internet as a platform. GIS affects all processes of the of the map output
“production
map outputline”—from
“productiondata collection
line”—from andcollection
data storage to data
and analysis
storage and analysis
to data visualization.
and vis-In
that case, the entire workflow is managed within a desktop environment.
ualization. In that case, the entire workflow is managed within a desktop environment. In In the case of
WebGIS, the workflow is managed within a web environment.
the case of WebGIS, the workflow is managed within a web environment.
2.1. Implementation of Cloud Computing
2.1. Implementation of Cloud Computing
Terminologically, an analogy can also be drawn with the concept of “CloudGIS”
Terminologically, an analogy can also be drawn with the concept of “CloudGIS” or
or “GIS cloud” [28]. While WebGIS generally allows the use of cloud storage as one
“GIS cloud” [28]. While WebGIS generally allows the use of cloud storage as one of several
of several alternatives (the application can be accessed from its own server, localhost,
alternatives (the application can be accessed from its own server, localhost, commercial
commercial webhosting, etc.), cloudGIS is oriented exclusively toward being hosted in a
webhosting, etc.), cloudGIS is oriented exclusively toward being hosted in a cloud envi-
cloud environment. Therefore, cloudGIS could be considered a subset of WebGIS. A typical
ronment. Therefore, cloudGIS could be considered a subset of WebGIS. A typical feature
feature is the authenticated access, allowing, among other things, data and layer archiving
is the authenticated access, allowing, among other things, data and layer archiving in ad-
in addition to map application creation, usually under a freemium business model.
dition to map application creation, usually under a freemium business model.
Cloud computing in general is a model of using computing technology available
Cloud computing in general is a model of using computing technology available
through the internet. Mell and Grance [29] define cloud as “a computing model for enabling
through the internet. Mell and Grance [29] define cloud as “a computing model for ena-
ubiquitous, convenient, and on-demand network access to a shared pool of configurable
computing resources (e.g., servers, storage, applications, and services) that can be rapidly
provisioned and released with minimal management effort or service provider interaction”.
The cloud architecture provides services or programs stored (on extraneous servers) in an
internet environment, and the user benefits from the scalability of applications. Users can
access their data or programs through a web browser anywhere in the world, but in fact,
the computing resources lie outside the reach of the actual structure—“somewhere in the
clouds”. In general, cloud computing can be defined as the access to data or computing
power in an internet environment [30].
ISPRS Int. J. Geo-Inf. 2023, 12, 374 4 of 18

The principles of cloud computing build on each other. Lawyer [31] defines five pillars
of cloud computing:
• Internet technologies—services available entirely over the internet;
• Service principle—the consumer and provider needs are separated by a clearly defined
interface, which can be described as a service;
• Measurement and pay-per-use—usage is tracked based on defined metrics, which
then enable charging for it;
• Scalability and elasticity—computing resources can be scaled up/down operationally
to respond to current needs;
• Resource sharing—realizes the economies of scale and maximizes resource efficiency.
From an architectural perspective, cloud computing can be approached at three lev-
els [32]. Infrastructure as a service (IaaS) provides the greatest flexibility. It offers storage
and computing resources to deploy any software. The user can install and configure their
own applications, renting hardware (analogous to a dedicated server). The IaaS is suitable
in cases where the user wants to use a cloud solution but cannot or does not want to
maintain their own technological facilities. It allows the user to set the computing power
exactly according to the requirements. Platform as a service (PaaS) extends the IaaS with
the ability to manage the platform on which the application runs (similar to web hosting),
often supplemented with pre-installed tools. The provider delivers the platform and takes
care of running the technology and hardware. The disadvantage (but also an advantage,
on the other hand) is the lack of hardware management. Hierarchically, the highest level
of cloud computing is the software as a service (SaaS) model [33]. The customer rents a
specific application designed for specific purposes or software from the provider and pays
only for the time of real use. All tools mentioned in this paper are available under the SaaS
cloud computing model.
For deploying WebGIS solutions in the cloud, there are services specialized directly
in the area of spatial data processing according to the SaaS model. The customer leases a
specific application or software from the provider and pays only for the actual use. Specific
examples are ArcGIS Online, CARTO, Mapbox, GIS Cloud, MapTiler and others. These
services allow the visualization, analysis and sharing of spatial data. De facto, these are GIS
clients available online in an internet environment. From the point of view of this study,
the principle of Service Oriented Architecture (SOA) is fundamental.

2.2. Implementation of Services

The principle of SOA is the provision of services to other components through a
communication protocol over a network; in practice, these are standardized services,
which allow working with data or tools “remotely”, without local access to them [18].
It is a general approach, a generic concept for service composition independent of the
implementation and platform. Users can share data, maps, compositions, tools or entire
applications without the need for intervention or direct ownership of the raw data. The
SOA model is based on the interaction of two parties: the service provider and the service
consumer (user/client/consumer).
While in a conventional GIS in a desktop environment, the power of the computer
itself is used to process operations, in a WebGIS solution, it is the services, which are used
for processing. A specific implementation of SOA in the GIS domain consists of web map
services. In general, these are services providing data or operations where the user accesses
the required information through the internet. These are actually located on a remote server.
Data services for visualization are already routinely used. However, the general principle of
web services can be applied to a broader spectrum—metadata, cataloging services, transfor-
mation, sensor networks, ontology, styling or operations (processes). The Open Geospatial
Consortium (OGC) is an expert group “committed to improving access to geospatial, or
location information” [34]. An overview of the most used services in the field of web
cartography is given in Table 1. The most used types of services are map or object services,
enabling the visualization of raster or vector geodata. The Web Map Service (WMS) [35]
ISPRS Int. J. Geo-Inf. 2023, 12, 374 5 of 18

generates a data image based on client request (the required operation, bounding box
defining the extent and scale, format file, coordinate reference system, version, active layers,
image dimensions, style), which then sends the map back to the user as a raster image [36].
This raster image is a portrayal of the data, which may be of raster or vector origin in
the database. Every time there is a movement, however minimal, in the map field, a new
query is sent to the server, and a new output is generated. WMS publishing requires an
infrastructure and a suitable technical solution for data publishing (GeoServer, MapServer,
ArcGIS for server). On the other hand, consuming the service is trivial. It is supported by
most desktop GIS software. However, a simple web browser can also be used for visual-
ization; a specific WMS entry with all parameters, returning the desired image, then looks
like a URL http://geoportal.cuzk.cz/WMS_ORTOFOTO_PUB/WMService.aspx?service=
WMS&request=GetMap&version=1.3.0&layers=GR_ORTFOTORGB&format=image/jpeg&
width=800&height=600&bbox=50.5,14.0,50.8,14.3&crs=epsg:4326&styles= (accessed on 5
September 2023).

Table 1. Most commonly used OGC services [34] in the field of WebGIS.

Abbrv. Full Name Description

WMS Web Map Service Map service (raster)
WMTS Web Map Tiled Service Map service (raster)
WFS Web Feature Service Object service (vector)
WFS-T Transactional Web Feature Service Object service (vector)
WPS Web Processing Service Processing service
WCS Web Coverage Service Multi-dimensional coverage data
Extraction, processing and analysis of
WCPS Web Coverage Processing Service
multi-dimensional coverages
Well-Known Text Representation of
WKT CRS Coordinate reference systems
Coordinate Reference Systems
SLD Styled Layer Descriptor Encoding standard, extends WMS
SE Symbology Encoding Symbology encoding

2.3. Web Processing Service and Turf Library

Process services, analogous to the above examples of web services, allow working
with geographic data in the web interface and performing spatial analyses (e.g., buffer,
clip, filtering), network analyses or spatial data editing. A specific implementation of a
processing service is the Web Processing Service (WPS). The WPS is an OGC standard [37]
and includes three operations, which a client can request from a server. These operations
are
• GetCapabilities—This operation provides the client with metadata regarding the
specific operations implemented offered by the server. It provides a general description
of the operations and their names.
• DescribeProcess—It gives the client detailed information regarding the processes,
including the required inputs, their allowable formats and possible outputs.
• Execute—This operation allows the client to execute an implemented operation, which
takes the specified parameters and returns a result layer.
According to Quinn [38], the WPS specification does not specify any kinds of spatial
data processing operations. There are hundreds of potential operations, which can be
included. Specific processes, inputs and outputs are defined by each individual WPS
implementation. When a WPS is requested for the first time, the GetCapabilities operation
then returns a list of available processes.
Implementing spatial operations via services follows a strictly established workflow
specified by the OGC [39]. For example, if the user wants to create a buffer zone over the
input layer, the following steps are executed: The requests for metadata are called through
the GetCapabilities operation. → The response is that the server supports the buffer
operation, including service metadata and metadata describing the available processes.
ISPRS Int. J. Geo-Inf. 2023, 12, 374 6 of 18

→ The user requests the DescribeProcess. → The answer to this query specifies the ability
to set the size of the wrapper zone, including the units. → Finally, the user launches buffer
processing through the Execute operation, which returns the newly created buffer zone
layer over the input layer.
An example of a concrete WPS implementation is PyWPS [40], Zoo Open WPS Platform
or WPS extension by GeoServer. Each operation is defined by GetCapabilities, and the
requests are based on HTTP GET or POST request. Using these requests, the standard
further defines
• Encoding the request to execute a process;
• Encoding the response from the executed process;
• Inserting data and metadata into the inputs and outputs of the executed process;
• Reference data inputs and outputs accessible on the web;
• Support for long-term processes;
• Process status information;
• Information regarding errors, which occurred during the process;
• Storage of process outputs.
With the gradual transition of applications to the cloud, it can be concluded that WPS
has enormous potential. Process web services can add value to web maps in terms of
the ability to perform spatial operations, analyses and to work with data, not just simple
visualization as before. Web map applications integrating process services can partially
replace the traditional approach through desktop software. Finally, WPS adds value to
the field of WebGIS, but it is only a declaration and description of background processes;
it does not define the user environment, cartographic symbology [41] or user interaction.
Therefore, there is a need to use client-side solutions.
Currently, there are only a few really usable implementations of geospatial analysis
within the WebGIS area, but one library provides a variety of prepared and ready-made
operations, such as buffer, dissolve, intersect, union, transform, etc.: Turf.js [14]. It is a
JavaScript library distributed under the open Massachusetts Institute of Technology (MIT)
license (free to use). According to Quinn [38], “many spatial processing operations are quite
common and there is a great desire from GIS web developers to invoke these without the
overhead of a WPS. The Turf.js has gained popularity for allowing geoprocessing directly
on GeoJSON vectors in a JavaScript environment”. It is built as a modular collection
of 140 modules for geoprocessing tools as a web-based universal alternative for spatial
analysis processes [42].
There is a crucial technical difference between Turf.js as a specific implementation and
other WPS as a general scheme. Since JavaScript is computed on the client side, the client
machine runs the processing code, not the server. Therefore, WPS provides server-based
spatial processes, while Turf.js is based on client-based operations. That strategy allows the
creation of platform-independent and no-installation-process-required solutions because
only a web browser is required. It was developed to directly support the most used open-
source mapping libraries, such as Leaflet and Mapbox. The implementation of Turf.js into
any custom JavaScript application is elementary and simple [43]. It requires Turf core script,
typically located in the <head> element of the webpage, which defines the general variable
“turf” and then defines the required operation.
The typical GIS operation for demonstration used routinely on both desktop and web
solutions is, e.g., buffer. It calculates the wrapping zone around the input features within
a given distance (radius). A run of the buffer command via Turf.js requires the definition
of the global Turf.js variable, the operation name (buffer), two obligatory variables (input
features, radius) and an optional variable (unit). Then, the code in the JavaScript notation
of the command is

var buffered = turf.buffer(point, 500, {units: ‘miles’});

 ISPRS Int. J. Geo-Inf. 2023, 12, 374 7 of 18

3. Methodology—Case Study Development

GeOnline is the authors’ case study, developed with a focus on the implementation of
non-trivial geospatial analysis within the web browser. The motivation of the project is to
demonstrate the availability of spatial analysis tools and to provide the end user with a
unique environment for spatial operations. The main aim of the development process is to
implement, test and compare a list of spatial analyses available on the market of WebGIS
solutions (see below for a list of supported operations). Alternative tools are analyzed
to identify suitable analytical operations, specify their pros and cons, and compare them.
GeOnline thus provides a solution, which implements the features expected in alternative
solutions but at the same time adds additional value to the new tools.
The core is based on the Turf.js library and enables the running of spatial operations,
analyses and data processing. The abbreviation GeOnline stands for a combination of
“geospatial” and “online”. The application is available in English; it is free to use and
available online at URL https://gis.upol.cz/geonline/ (accessed on 5 September 2023).
The solution combines the traditional and widely used combination of Hypertext Markup
Language (HTML), Cascading Style Sheets (CSS) and JavaScript (JS).
The interface of the application is divided into a map field, which occupies the main
part of the display, and a sidebar, which can be used to operate the application (see Figure 2).
The sidebar is divided into three tabs. The first About tab serves as a short introduction to
the application and contains instructions on how to use it. The Application tab is the main
part of the sidebar. It houses the file upload, layer list and spatial operations options. The
last tab, Legend, is used to display the individual cartographic features for the available
layers. The OpenAboutApp, OpenContent and OpenLegend functions are used to switch
between the tabs. Since the Turf.js library only allows performing spatial operations and is
not intended for data visualization, it was necessary to choose a suitable mapping solution.
The Mapbox GL JS [44] library was chosen from a number of open-source libraries. The
main reason is its comprehensiveness, speed, wide selection of plugins and extensive
support, including elaborate documentation. A factor taken into account was also the
Mapbox
ISPRS Int. J. Geo-Inf. 2023, 12, x FOR PEER REVIEWteam’s collaboration with Turf.js; the tutorials for linking Turf.js to the Mapbox GL
8 of 19
JS library are available on the official Mapbox website [44].

Figure
Figure2.2.GeOnline
GeOnlineinterface.
interface.

The
Thefinal
finallist
listofofoperations
operationswas
waschosen
chosentotobebeasasdiverse
diverseasaspossible
possibleandandwith
withadditional
additional
value
valueseen
seenininalternative
alternativesolutions.
solutions.Fourteen
Fourteenspatial
spatialoperations
operationsinintotal,
total,divided
dividedinto
intothree
three
categories,
categories,were
wereimplemented:
implemented:
•• Measurement
Measurement(2):
(2):Calculate
Calculatearea;
area;Calculate
Calculatelength;
length;
• Analysis and transformation (9): Buffer; Envelope; Get centroid coordinates; Line to
polygon; Points within polygon; Polygon to points; Simplify polygon; Create TIN;
Voronoi polygons;
• Random (3): Create random points; Create random lines; Create random polygons.
In the sidebar, under the Select operation heading, a scrolling menu was created,
ISPRS Int. J. Geo-Inf. 2023, 12, 374 8 of 18

• Analysis and transformation (9): Buffer; Envelope; Get centroid coordinates; Line to
polygon; Points within polygon; Polygon to points; Simplify polygon; Create TIN;
Voronoi polygons;
• Random (3): Create random points; Create random lines; Create random polygons.
In the sidebar, under the Select operation heading, a scrolling menu was created,
where all the implemented operations are located. Selecting one of them will expand a
form with additional attributes and basic symbology editing options. Each operation can
only be performed on a specific data type (point, line, polygon), and the suitability of each
layer is ensured in the same way as for file upload—in this case, through the SuitableLayer
function. In the case of the Points within polygon operation, two layers are input, and the
SecondSuitableLayer function is executed with the SuitableLayer function. The result is
the ListOfSuitableLayers array, which is output to the drop-down list below the selected
operation. After selecting the operation, the layer and the attributes, the Execute function
is launched when the SUBMIT button is clicked. It uses the getElementById() method
to provide all input parameters and selects a function from the Turf.js library to execute
the spatial operation (Figure 3). The resulting layer is stored in the toggleableLayersId
field—and thus, in the list of layers in the sidebar—and is displayed in the map field. A link
is created to download the new layer, and a cartographic feature is written on the legend.
Along with the newly created layer, a download link is created under the title List
of layers. The LinkToDownload function is launched after the operation from the Turf.js
library and takes two arguments: the layer with the data and its name. Using the en-
codeURIComponent(JSON.stringify()) function, the data from the layer are stored in the
data variable, and a link is created using the document.createElement() method, which is
placed
ISPRS Int. J. Geo-Inf. 2023, 12, x FOR PEER in the download icon of the layer. The parameters of color, width, line connection
REVIEW 9 of 19
type and line termination are implemented for the line geometries.

Figure 3. Source code ensuring the “Line to polygon” operation.

Figure 3. Source code ensuring the “Line to polygon” operation.

4. Results—Comparison Study
The main objective of this paper is to make a comparison of alternative solutions
providing spatial analytical operations through a web-based solution. Inspired by studies
focused on performance analysis by El-Sayed et al. [45] or by Alfaqih [28], the authors
attempted to compile a comprehensive review of the studied issue in two phases. In the
first phase, testing and comparison of the tools available on the market in terms of the
available functionality were carried out. In the second step, performance testing of the
ISPRS Int. J. Geo-Inf. 2023, 12, 374 9 of 18

4. Results—Comparison Study
The main objective of this paper is to make a comparison of alternative solutions
providing spatial analytical operations through a web-based solution. Inspired by studies
focused on performance analysis by El-Sayed et al. [45] or by Alfaqih [28], the authors
attempted to compile a comprehensive review of the studied issue in two phases. In the
first phase, testing and comparison of the tools available on the market in terms of the
available functionality were carried out. In the second step, performance testing of the
selected solutions was performed.
For the comparative study, the tools available on the market, which support the
functionality of web process services and their visualization in a cloud environment, were
included as ready-made solutions: ArcGIS Online, CARTO, Foursquare Studio, GISCloud
and GeOnline (in alphabetical order). Therefore, the study did not include solutions in the
form of APIs (MapBox GL JS, Google Maps API, Esri API, etc.). In addition to the above
products, there are a number of alternative products available on the market (e.g., MapBox,
Maptiler Cloud, Kepler, etc.). Overlapping these products is the MapBox API, which partly
uses the Turf.js libraries.
This comparison study was based on comprehensive research of the issues and so-
lutions available on the market. Two basic groups of geospatial processes were defined
for the purpose of this article: visual-driven and data-driven analyses (see Table 2). The
first group comprises visual-driven geospatial processes. They only involve changing the
visualization over the original input data; however, there is no backward spatial operation
over the original data. Examples of visual-driven visualizations are the heat map, grid,
hexbin, clustering. In this case, the input data remain the same, while the computation
is only necessary to change the expression method; a typical example is changing the
visualization of a point layer into the form of a heat map. Visualizations in the form of
heat maps or grids can today be considered as standard visualization methods, which
are available through a number of visualization tools (ArcGIS Online, CARTO, Maptiler,
Mapbox, Google Maps API, Leaflet and many others), but not all of them support analysis
tools. The issue of heat maps is discussed in detail in Netek et al. [46].
The second group comprises data-driven computed (true) geospatial analyses. This is
a set of methods based on spatial operations of input data, where the data are computed and
then transformed, usually into a new data layer. Typical examples include the operations
of buffer, calculate centroids, spatial join, intersect, merge, aggregate and others.
A specific group of spatial operations includes geocoding (forward and reverse) and
geolocation operations. However, these operations are not based on spatial analysis; for
this reason, they are not the focus of this study.
This paper primarily focuses on data-driven (data-based) analysis, available via a
cloud computing environment.

Table 2. Geospatial analysis categorization.

Visual-Driven Data-Driven (Data-Based)

Name
(Visual-Based) Analysis Analysis
Type False geospatial analysis True geospatial analysis
Computation for Visualization purposes only Spatial data distribution
Change of Visualization method Input data
Operation examples Heat map, grid, hexbin Buffer, centroids . . .

4.1. Functionality Comparison

Several geoinformatic aspects, such as the number of operations, input formats, at-
tribute or geometry editing, were selected for comparison (see Table 3). The most important
aspect is the number of spatial operations offered by the applications. First, a comparison
of quantitative characteristics and aspects was made. The highest number of operations
(29) is offered by ArcGIS Online; the GeOnline application supports 14 spatial operations,
comparable to CARTO, which offers 15 operations, while GIScloud offers only 4.
ISPRS Int. J. Geo-Inf. 2023, 12, 374 10 of 18

Table 3. Comparison of alternative solutions.

ArcGIS GIS
CARTO FOURSQUARE GeOnline
Online Cloud
Number of
29 4 15 3 14
operations
Analysis processing client side client side client side client side client side
SQL queries no no yes yes no
different different different different
Input formats GeoJSON
formats formats formats formats
Interactive zoom yes yes yes yes yes
Geocoding yes yes yes yes yes
Geolocation no yes no yes yes
Export layers yes yes yes yes yes
yes (for
Attribute editing yes databases yes yes no
only)
Geometry editing yes yes yes no no
Search yes yes yes no no
yes yes (nu-
Scale no no no
(graphical) merical)
User accounts yes yes yes yes no

ArcGIS Online [47] is a cloud-based and most comprehensive solution for these
options [48]. It is distributed as a SaaS for publishing, viewing and sharing data, interactive
maps and applications in Esri’s web environment. Data can be injected into ArcGIS Online
in a variety of ways: both web services (WMS, WFS, WMTS, WPS) and local files (Shapefile
(SHP), Geographic JavaScript Object Notation (GeoJSON), comma-separated values (CSV)
or GPS Exchange Format (GPX), etc.) are supported. It offers the largest number of spatial
operations, and apart from the absence of geolocation or numerical scaling, it has great
support for input formats and offers advanced cartographic visualization capabilities. A
great advantage is the possible integration with the Esri ecosystem (ArcGIS Pro, ArcGIS
Server). For each operation performed, the user has points deducted, which must be
purchased. Almost all web browsers are supported, even in mobile versions.
Supported ArcGIS Online operations (29 in total, divided into six categories):
• Summarize data (5): Aggregate Points; Join Features; Summarize Nearby; Summarize
Within; Summarize Center and Dispersion;
• Find locations (8): Find Existing Locations; Derive New Locations; Find Similar Loca-
tions; Choose Best Facilities; Create Viewshed; Create Watersheds; Trace Downstream; Find
Centroids;
• Data enrichment (1): Enrich Layer;
• Analyze patterns (5): Calculate Density; Find Hot Spots; Find Outliers; Find Point Clusters;
Interpolate Points;
• Use proximity (5): Create Buffers; Create Drive-Time Areas; Find Nearest; Plan Routes;
Connect Origins to Destinations;
• Manage data (5): Extract Data; Dissolve Boundaries; Generate Tessellations; Merge Layers;
Overlay Layers.
GISCloud [49] is one of the first web-based GIS, whose functionality is stored in the
cloud. It supports a range of vector formats (SHP, GPX, KML, SQLite, etc.), raster formats
(TIF, JPG, PNG, JP2, etc.) and web services (WMS, WFS, WMTS, TMS). It is a commercial
software. GISCloud offers the smallest selection of spatial operations, only four. Moreover,
two of them can only be performed with the input layer in the PostGIS database. The same
limitation applies to editing attributes of individual features. In addition to WebGIS itself,
GISCloud also provides APIs (REST and JavaScript), which allows data manipulation and
UI customization. The user can also host custom applications on GISCloud.
Supported GISCloud operations (four in total): Buffer; Heat Map; Near; Area.
ISPRS Int. J. Geo-Inf. 2023, 12, 374 11 of 18

CARTO (formerly CARTODB) [50] is a cloud computing platform for spatial data
analysis and distribution distributed as SaaS, which is distributed as a free software built
on PostGIS and PostreSQL. It is primarily intended for users who do not have professional
experience in GIS. CARTO supports all basic formats (SHP, GeoJSON, CSV, KML, GPX)
or geodatabases. It offers several operations (geocoding, filtering, envelope zones, trends,
point selection). CARTO provides the second highest number of spatial operations. It
offers a large variety of input formats, and apart from the absence of scaling and geolo-
cation, it offers all the basic functions. CARTO tries to support and implement general
standards for spatial data, such as the wide use of GeoJSON or vector tiles. It works in a
framework-agnostic way, so the implementation in user web applications does not depend
on a specific framework; however, use of the React library is recommended by CARTO
developers (the CARTO for the React framework is ready for this case). In addition to
JavaScript implementation, CARTO also supports Python implementation using its own
CARTOframes library. CARTOframes is intended to serve as a complement or replacement
for the pandas or jupyternotebooks libraries.
Supported CARTO operations (15 in total): Create second layer; Geocoding; Add columns
from two datasets; Intersection and Aggregation; Create path buffers/distances; Create centers of
ISPRS Int. J. Geo-Inf. 2023, 12, x FOR PEER REVIEWCreate polygons from points; Filter by column value; Select points in polygon;12Create
geometries; of 19

lines from points; Subsample percent rows; Calculate point clusters; Remove outliers and clusters;
Predict trends and volatility; Find nearest.
The Foursquare Studio (formerly Unfolded) Unfolded) [51] is one of the new WebGIS tools. It
supports data in
supports data in CSV,
CSV, JSON,
JSON, GeoJSON,
GeoJSON, Esri Esri SHP,
SHP, ZIP,
ZIP, Excel,
Excel, KML
KML and
and other
other formats.
formats. ItIt
uses
uses aa custom
custom HexTiles
HexTiles tiling
tiling system
system to to display
display data,
data, which
which should
should speed
speed up
up on-the-fly
on-the-fly
data
data analysis
analysis and
and also
also support
support performance
performance efficiency
efficiency in
in bigdata
bigdata processing.
processing. It
It includes
includes
two spatial operations in total: Cluster and Outlier Analysis (hot spots
two spatial operations in total: Cluster and Outlier Analysis (hot spots and cold spots;and cold spots;
Figure
Figure 4) and Suitability Analysis (finding potentially suitable locations). There are also
4) and Suitability Analysis (finding potentially suitable locations). There are also
point,
point, arc
arc and
and grid
grid visualization
visualization tools.
tools. The
The application
application does
does not
not yet
yet support
support standalone
standalone
tools; only Spatial
tools; only SpatialJoin
Joincan
canbe becreated,
created,orordata
data can
can bebe filtered
filtered byby spatial
spatial andand temporal
temporal af-
affiliation. Foursquare Studio can also be used in Grafana and Google
filiation. Foursquare Studio can also be used in Grafana and Google Data Studio to visu- Data Studio to
visualize spatial data using
alize spatial data using SQL. SQL.
Supported
Supported Foursquare
Foursquare Studio
Studio operations
operations (three
(three in
in total):
total): Outlier
Outlier Analysis;
Analysis; Suitability
Suitability
Analysis; Geocoding.
Analysis; Geocoding.

Figure 4. Cluster and Outlier Analysis in Foursquare Studio.

Figure 4. Cluster and Outlier Analysis in Foursquare Studio.

GeOnline is a lighter version of the above solutions. It offers 14 spatial operations;

the biggest drawback is the limited number of input formats and the lack of editing attrib-
utes and geometry. The biggest advantage of this application is its simplicity and speed.
The application is the only one, which does not offer the option of user accounts. This can
ISPRS Int. J. Geo-Inf. 2023, 12, 374 12 of 18

GeOnline is a lighter version of the above solutions. It offers 14 spatial operations; the
biggest drawback is the limited number of input formats and the lack of editing attributes
and geometry. The biggest advantage of this application is its simplicity and speed. The
application is the only one, which does not offer the option of user accounts. This can be
considered both an advantage (simplicity and accessibility) and a disadvantage (lack of
data saving into local storage and history). One of the biggest weaknesses of the GeOnline
application is the range of input formats; the user has the option to upload data only in the
GeoJSON format. Other applications offer a wider portfolio of raster and vector formats.
In this aspect, GeOnline cannot compete with alternative solutions from other companies.
All applications offer basic map elements (legend, zoom, export). In addition to GeOnline,
other web applications also offer attribute and geometry editing. However, the GISCloud
platform only offers attribute editing for layers, which were uploaded via the database.
Additionally, unlike other products, GeOnline does not offer an underlying map selection
or concept search. On the other hand, GeOnline does offer a geolocation option.
Supported GeOnline operations (14 in total): Area calculation; Length calculation;
Buffer; Envelope; Get center coordinates; Line to polygon; Points in polygon; Polygon to points;
Polygon simplification; TIN creation; Voronoi polygons; Random point creation; Random line
creation; Random polygon creation.

4.2. Performance Testing

In addition to the conventional methods available via user interface, we also have
web-based tools for using geospatial functions and processing spatial data. The OGC has
created WPS, a service, which communicates using XML, for this purpose. WPS can also be
used within a client machine, for example in QGIS. Similarly, some web APIs allow the use
of this service (OpenLayers).
Two stages of performance tests were performed. First, the performance test of
GeOnline, in two exemplary operations (create random points and buffer), was performed.
The aim of this testing stage was to identify the performance limits and calibrate the
performance of the GeOnline application. Then, a comparison test of five alternatives,
including GeOnline, was performed to compare the performance results among alternative
solutions. DevTools (Developer Console) is an integral tool in the current web browsers
used to proceed, record and calculate the results of both stages (Figure 5) [52]. Google
Chrome, the most used web browser [53] with Chromium rendering engine, was used for
this study to follow the same rendering as the majority of the market. Ten measurements in
total were performed for each test; the minimum and maximum results were excluded; and
the arithmetic mean was calculated as the final value. All performance tests were always
carried out with the maximum map field zoom-out to ensure the results’ comparability.
When the map field is zoomed in on a certain area, the generated layer is usually rendered
earlier/faster because of the smaller data extent. All measurements were carried out under
the same conditions on the same computer with the following specifications:
• Processor: Intel(R) Core(TM) i5-3570 CPU @ 3.40 GHz,
• RAM: 16.0 GB RAM,
• Apacer AS350 512 GB (SSD),
• Graphics Card: Intel HD Graphics,
• Wired internet connection: 930 Mbps download,
• Browser: Google Chrome, version 111.0.5563.147.
First, the performance test of GeOnline was performed (see Figure 6). The aim was
to identify performance limits. Two example operations were selected (create random
points and buffer) and tested among several datasets with different parameters (number
of generated points) (see Tables 4 and 5). The complexity of buffer operations depends on
the distance parameter; therefore, 100 m radius was set for all tests. The buffer operation
passed within all datasets. Creating random points failed with 1,000,000 or more generated
points. Moreover, source code optimization should reduce the waiting time for the full
loading parameter to approx. 25 s (for 250k points) or approx. 58 s (for 500k points).
 out under the same conditions on the same computer with the following specifications:
• Processor: Intel(R) Core(TM) i5-3570 CPU @ 3.40 GHz,
• RAM: 16.0 GB RAM,
• Apacer AS350 512 GB (SSD),
ISPRS Int. J. Geo-Inf. 2023, 12, 374
• Graphics Card: Intel HD Graphics, 13 of 18
• Wired internet connection: 930 Mbps download,
ISPRS Int. J. Geo-Inf. 2023, 12, x FOR PEER REVIEW 14 of 19
• Browser: Google Chrome, version 111.0.5563.147.

First, the performance test of GeOnline was performed (see Figure 6). The aim was to
identify performance limits. Two example operations were selected (create random points
and buffer) and tested among several datasets with different parameters (number of gen-
erated points) (see Tables 4 and 5). The complexity of buffer operations depends on the
distance parameter; therefore, 100 m radius was set for all tests. The buffer operation
passed within all datasets. Creating random points failed with 1,000,000 or more gener-
ated points. Moreover, source code optimization should reduce the waiting time for the
full loading parameter to approx. 25 s (for 250k points) or approx. 58 s (for 500k points).

Table 4. GeOnline test results—Create random points.

Generated
10,000 50,000 100,000 250,000 500,000 1,000,000 2,000,000
Points
Loading
2461 4787 9041 25,449 57,605 N/A N/A
Time (ms)

Table 5. GeOnline test results—Buffer (100 m radius).

Generated
1000 2000 5000 10,000 25,000 50,000 100,000
Points
Loading
1233 1679 2511 3818 8063 17,068 33,311
Time (ms)
Figure 5. DevTools—an integral tool in web browsers used for performance testing evaluation.
Figure 5. DevTools—an integral tool in web browsers used for performance testing evaluation.

Figure 6. Creating a buffer in GeOnline and measuring indicator values.

Figure 6. Creating a buffer in GeOnline and measuring indicator values.

TablePrevious performance
4. GeOnline tests gave
test results—Create us important
random points. feedback for the next development,
but on the other hand, we wanted to compare the performance of GeOnline among alter-
Generated
native solutions. In order to compare the alternative solutions, the performance
10,000 50,000 100,000 250,000 500,000 1,000,000 of2,000,000
selected
Points implemented in them was tested. These operations were selected for inclusion
operations
in asLoading
few solutions2461
as possible:
4787 9041 25,449 57,605 N/A N/A
Time (ms)
• Buffer creation (used twice with different numbers of input points),
• Polygon centroid calculation (not in GISCloud),
•TableCluster
5. GeOnline
and test results—Buffer
outlier (100inmGeOnline
analysis (not radius). and GISCloud),
• Heat map creation (not in GeOnline and GISCloud).
Generated Points 1000 2000 5000 10,000 25,000 50,000 100,000
The following indicators were monitored as part of performance testing:
Loading Time (ms) 1233 1679 2511 3818 8063 17,068 33,311
• Full time loading—the total time from the start of the operation to the loading of the
output data,
ISPRS Int. J. Geo-Inf. 2023, 12, 374 14 of 18

Previous performance tests gave us important feedback for the next development, but
on the other hand, we wanted to compare the performance of GeOnline among alternative
solutions. In order to compare the alternative solutions, the performance of selected
operations implemented in them was tested. These operations were selected for inclusion
in as few solutions as possible:
• Buffer creation (used twice with different numbers of input points),
• Polygon centroid calculation (not in GISCloud),
• Cluster and outlier analysis (not in GeOnline and GISCloud),
• Heat map creation (not in GeOnline and GISCloud).
The following indicators were monitored as part of performance testing:
• Full time loading—the total time from the start of the operation to the loading of the
output data,
• Process time/execution—total time of calculation of the given operation (computa-
tional process),
• Number of requests—number of requests to the server,
• Transferred over network—volume of data transferred over the network,
• Resources loaded by the page—volume of data loaded by the page without compres-
sion and caching.
In the buffer creation (see Tables 6 and 7), polygon centroid (see Table 8) and cluster
analysis (see Table 9) comparison test, the best (lowest) values were measured using
GeOnline. The calculations of the output display operations were performed in fast times,
with only one query sent and a small volume of data transferred and retrieved. On the
contrary, the highest values in terms of loading and computation time were achieved when
measuring these operations on the ArcGIS Online platform. The reason for these higher
values is the optimization of the platform for high traffic and performance in combination
with the multitude of tools, applications and data available.

Table 6. Comparison of test results—Buffer (100 m radius, 100 input points).

ArcGIS
GISCloud CARTO FOURSQUARE GeOnline
Online
Full time loading (s) 19.78 9.03 6.13 1.97 0.05
Process/service
5.31 1.06 0.35 0.56 0.05
time (s)
Number of requests 46 36 4749 6 1
Transferred over
116 kB 39.55 kB 26.1 kB 5.8 kB 779 B
network
Resources loaded by
106 kB 33.45 kB 144 MB 5.7 kB 495 B
the page

Table 7. Comparison of test results—Buffer (50 km radius, 50 000 input points).

ArcGIS
GISCloud CARTO FOURSQUARE GeOnline
Online
Full time
591.52 22.57 50.5 6.24 8.07
loading (s)
Process/execution
41.95 16.63 s 39.8 3.24 6.89
time (s)
Number of
225 27 5382 6 1
requests
Transferred over
10.1 MB 2.6 MB 65.2 MB 7.8 kB 780 B
network
Resources loaded
10 MB 10.4 MB 228 MB 7.7 kB 495 B
by the page
ISPRS Int. J. Geo-Inf. 2023, 12, 374 15 of 18

Table 8. Comparison of test results—Centroid of the polygon (one input polygon).

ArcGIS
GISCloud CARTO FOURSQUARE GeOnline
Online
Full time loading (s) 19.39 - 6.7 2.05 0.06
Process/service
5.08 - 2.16 0.64 0.06
time (s)
Number of requests 90 - 5030 6 1
Transferred over
1.9 MB - 16.8 kB 6.6 kB 780 B
network
Resources loaded by
1.9 MB - 152 MB 6.5 kB 495 B
the page

Table 9. Comparison of test results—Cluster and outlier analysis.

ArcGIS
GISCloud CARTO FOURSQUARE GeOnline
Online
Full time loading (s) 37.5 - 7.9 1.89 -
Process/service
5.03 - 2.64 0.47 -
time (s)
Number of requests 40 - 4610 6 -
Transferred over
92.7 kB - 18.7 kB 10.4 kB -
network
Resources loaded by
83.5 kB - 140 MB 10.3 kB -
the page

The most queries to complete the operation were always required while using the
CARTO application. This application was also measured to have the highest values of data
volume loaded by the web page (application) during the measurement. However, in terms
of time measurements, average results were obtained compared to the other applications.
Only in the case of the use of heat map (see Table 10) tools were the results different.
On the ArcGIS Online platform, the heat map was created fastest by sending only one
request; the volume of data transferred and retrieved was also low. On the other hand, the
heat map took the longest to load with FOURSQUARE. The measurement values of other
operations for this application were comparable to GeOnline. The heat map creation tool
was also included in the GISCloud application, but the values were not measured in this
application due to the long computation time in the online environment.

Table 10. Comparison of test results—Heat map (distance 500 m, 100 input points).

ArcGIS
GISCloud CARTO FOURSQUARE GeOnline
Online
Full time
3 - 1860 3375 -
loading (ms)
Process/service
3 - 152 421 -
time (ms)
Number of requests 1 - 1395 7 -
Transferred over
0 kB - 2.6 kB 6.2 kB -
network
Resources loaded by
227 kB - 42.5 MB 6 kB -
the page

The measurement values of the speed of calculation of simple operations are in the
order of units of seconds. In the case of a larger volume of data in the input layer or a more
computationally intensive operation, the processing time and subsequent rendering of the
data are longer. Overall, the number of queries sent to the server and the volume of data
ISPRS Int. J. Geo-Inf. 2023, 12, 374 16 of 18

transferred over the network and retrieved by the page varied depending on the operation
and, more importantly, the application used.

5. Conclusions
Online geoprocessing tools in the WebGIS environment are becoming increasingly
important in providing spatial analysis capabilities to users. While web cartography
focuses more on visualization, WebGIS enables users to perform analytical operations.
The development of GeOnline demonstrates the availability of specialized geospatial
libraries such as Turf.js for non-trivial geospatial analysis in a web browser. By utilizing
the specialized geospatial library Turf.js, GeOnline successfully implemented non-trivial
vector operations, such as intersect, clip and calculate centroids. The paper emphasized
the importance of data-driven analysis in addition to visual-driven analysis, making it a
valuable contribution to the field of geospatial research.
Five geospatial tools, namely ArcGIS Online, GISCloud, CARTO, FOURSQUARE
and GeOnline, were thoroughly compared and evaluated, considering various aspects,
including the range of supported operations. While ArcGIS Online was identified as the
most comprehensive solution currently available on the market, GeOnline performed com-
mendably and demonstrated competitive performance compared to alternative solutions.
Overall, the study shows that there are a variety of cloud-based solutions available
for WebGIS, each with its strengths and weaknesses. The performance tests conducted
demonstrate that GeOnline performs well compared to alternative solutions. The study
provides a useful guide for users and developers interested in selecting and evaluating
online geoprocessing tools for WebGIS applications.
It is worth noting that the field of web-based geospatial analysis is constantly evolving,
and new tools and technologies may emerge after the completion of this study. Thus,
further research and updates to this comparative analysis will be necessary to remain
current with the latest developments in the domain.
In terms of the future, we suggest several options for future research, such as imple-
menting tools for (near) real-time data handling; integrating machine-learning algorithms;
cross-platform compatibility; integration with IoT networks; or integrating measures of
data security and user privacy.
During the study, several limitations were brought up. The research was limited
to only four other solutions, which could have an effect on the results. The focus was
specifically on vector data operation. For broader analysis scope, another comparison
should be performed for other data formats.

Author Contributions: Conceptualization, Rostislav Netek; methodology, Daniel Urban, Rostislav

Netek; case study development, Daniel Urban; comparison study, Oldrich Bittner, Tereza Pohankova;
resources, Daniel Urban, Rostislav Netek; writing—original draft preparation, Rostislav Netek,
Tereza Pohankova; writing—review and editing, Tereza Pohankova; supervision, Rostislav Netek.
All authors have read and agreed to the published version of the manuscript.
Funding: This paper was supported by the project “Analysis, modelling, and visualization of spatial
phenomena by geoinformation technologies II” (IGA_PrF_2023_017) funded by the Internal Grant
Agency of Palacký University Olomouc.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Dobson, J.E. The GIS revolution in science and society. In Geography and Technology; Springer: Dordrecht, Netherlands, 2004;
pp. 573–587.
2. Jing, T.; Baiping, Z.; Li, W. Design and Implementation of WebGIS Based on Distributed Components. In Proceedings of the 2008
International Conference on Computer Science and Software Engineering, Washington, DC, USA, 12–14 December 2008; IEEE:
New York, NY, USA, 2008; Volume 2, pp. 131–134.
ISPRS Int. J. Geo-Inf. 2023, 12, 374 17 of 18

3. Han, R. Web GIS in development: From research and teaching perspectives. In GIScience Teaching and Learning Perspectives;
Springer: Cham, Switzerland, 2019; pp. 103–122.
4. Manfren, M.; Caputo, P.; Costa, G. Paradigm shift in urban energy systems through distributed generation: Methods and models.
Appl. Energy 2011, 88, 1032–1048. [CrossRef]
5. Congiu, E.; Desogus, G.; Frau, C.; Gatto, G.; Pili, S. Web-Based Management of Public Buildings: A Workflow Based on Integration
of BIM and IoT Sensors with a Web–GIS Portal. Buildings 2023, 13, 1327. [CrossRef]
6. Brown, M.; Sharples, S.; Harding, J.; Parker, C.; Bearman, N.; Maguire, M.; Forrest, D.; Haklay, M.; Jackson, M. Usability of
geographic information: Current challenges and future directions. Appl. Ergon. 2013, 44, 855–865. [CrossRef] [PubMed]
7. Duarte, L.; Teodoro, A.C.; Santos, P.; Rodrigues de Almeida, C.; Cardoso-Fernandes, J.; Flores, D. An Interactive WebGIS
Integrating Environmental Susceptibility Mapping in a Self-Burning Waste Pile Using a Multi-Criteria Decision Analysis
Approach. Geosciences 2022, 12, 352. [CrossRef]
8. Sánchez-Aparicio, L.J.; Masciotta, M.-G.; García-Alvarez, J.; Ramos, L.F.; Oliveira, D.V.; Martín-Jiménez, J.A.; González-Aguilera,
D.; Monteiro, P. Web-GIS approach to preventive conservation of heritage buildings. Autom. Constr. 2020, 118, 103304. [CrossRef]
9. Li, Y.; Li, S.; Peng, J. A WebGIS-based public opinion information system for urban planning. ISPRS Int. J. Geo-Inf. 2017, 6, 360.
10. Balla, D.; Zichar, M.; Kiss, E.; Szabó, G.; Mester, T. Possibilities for Assessment and Geovisualization of Spatial and Temporal
Water Quality Data Using a WebGIS Application. ISPRS Int. J. Geo-Inf. 2022, 11, 108. [CrossRef]
11. Farkas, G. Applicability of open-source web mapping libraries for building massive Web GIS clients. J. Geogr. Syst. 2017, 19,
273–295. [CrossRef]
12. Bhat, M.A.; Shah, R.M.; Ahmad, B. Cloud Computing: A solution to Geographical Information Systems(GIS). Int. J. Comput. Sci.
Eng. 2011, 3, 594–600.
13. Yue, P.; Zhou, H.; Gong, J.; Hu, L. Geoprocessing in Cloud Computing platforms—A comparative analysis. Int. J. Digit. Earth
2012, 6, 404–425. [CrossRef]
14. Turf.js. online. Available online: https://turfjs.org/ (accessed on 4 April 2023).
15. Zhao, Y.Y.; Liu, X.F.; Mao, J.H.; Yang, X.L.; Wang, H.R. 52◦ North WPS and its Application in Fire Emergency Response. Adv.
Mater. Res. 2013, 760, 1748–1752. [CrossRef]
16. Chen, H.; Zhu, D.; Wu, W.; Tong, X.; Liu, D. Research on application of cloud computing techniques in WebGIS system. In
Proceedings of the IEEE Conference Anthology, Hunan, China, 1–8 January 2013; pp. 1–5.
17. Quinn, S. Web GIS. In The Geographic Information Science & Technology Body of Knowledge (1st Quarter 2018 Edition); Wilson, J.P., Ed.;
University Consortium for Geographic Information Science UCGIS: Ithaca, NY, USA, 2018. [CrossRef]
18. Netek, R. Interconnection of Rich Internet Application and Cloud Computing for Web Map Solutions. In Proceedings of the 13th
SGEM GeoConference on Informatics, Geoinformatics and Remote Sensing, Albena, Bulgaria, 16–22 June 2013; ISBN 978-954-
91818-9-0. Available online: www.sgem.org (accessed on 17 March 2023).
19. Dorman, M. Introduction to Web Mapping; Chapman and Hall, CRC Press: Boca Raton, FL, USA, 2020; p. 347. ISBN 9780367861186.
20. Čepický, J. WebGIS. ČVUT Praha. 2008. Available online: http://geo.fsv.cvut.cz/wiki/images/1/1c/Cepickywebgis.pdf
(accessed on 23 March 2023).
21. Sack, C. Web Mapping. In The Geographic Information Science & Technology Body of Knowledge (4th Quarter 2017 Edition); Wilson, J.P.,
Ed.; University Consortium for Geographic Information Science UCGIS: Ithaca, NY, USA, 2017. [CrossRef]
22. Wang, Y.; Wang, S.; Zhou, D. Retrieving and indexing spatial data in the cloud computing environment. In Cloud Computing;
Springer: Berlin/Heidelberg, Germany, 2009; pp. 322–331.
23. Kraak, M.J.; Brown, A. (Eds.) Web Cartography: Developments and Prospects; Taylor & Francis Group: London, UK, 2001; p. 213.
ISBN 9780748408696.
24. Kong, N.; Zhang, T.; Stonebraker, I. Evaluation of web GIS functionality in academic libraries. Appl. Geogr. 2015, 60, 288–293.
[CrossRef]
25. Veenendaal, B.; Brovelli, M.A.; Li, S. Review of web mapping: Eras, trends and directions. ISPRS Int. J. Geo-Inf. 2017, 6, 317.
[CrossRef]
26. Hojati, M. What is the Difference between Web GIS and Internet GIS? 2014. Available online: https://www.gislounge.com/
difference-web-gis-internet-gis/ (accessed on 23 March 2023).
27. Netek, R.; Balun, M. WebGIS solution for crisis management support–case study of Olomouc municipality. In Proceedings of the
Computational Science and Its Applications–ICCSA 2014: 14th International Conference, Guimarães, Portugal, 30 June–3 July
2014.
28. Alfaqih, T.M.; Hassan, M.M. GIS Cloud: Integration between cloud things and geographic information systems (GIS) opportunities
and challenges. Int. J. Comput. Sci. Eng. 2016, 3, 360–365.
29. Mell, P.T.; Grance, T. The NIST definition of cloud computing. Natl. Inst. Stand. Technol. 2009, 53, 50.
30. Huang, Q. Spatial Cloud Computing. In The Geographic Information Science & Technology Body of Knowledge (2nd Quarter 2020
Edition); Wilson, J.P., Ed.; University Consortium for Geographic Information Science UCGIS: Ithaca, NY, USA, 2020. [CrossRef]
31. Lawyer, D. The Five Key Pillars that Support Effective Cloud Architecture. IT Brief. 2022. Available online: https://itbrief.com.
au/story/the-five-key-pillars-that-support-effective-cloud-architecture (accessed on 17 March 2023).
32. Bhardwaj, S.; Jain, L.; Jain, S. Cloud computing: A study of infrastructure as a service (IAAS). Int. J. Eng. Inf. Technol. 2010, 2,
60–63.
ISPRS Int. J. Geo-Inf. 2023, 12, 374 18 of 18

33. Cusumano, M. Cloud computing and SaaS as new computing platforms. Commun. ACM 2010, 53, 27–29. [CrossRef]
34. Open Geospatial Consorcium. Standards. 2023. Available online: https://www.ogc.org/standards/ (accessed on 29 March 2023).
35. Open Geospatial Consortium. Web Map Service (WMS) 1.3.0. 2019. Available online: https://www.opengeospatial.org/
standards/wms (accessed on 31 July 2023).
36. Michaelis, C.D.; Ames, D.P. Web Feature Service (WFS) and Web Map Service (WMS). In Encyclopedia of GIS; Springer: Boston,
MA, USA, 2008.
37. Open Geospatial Consortium. Web Processing Service (WPS) Version 2.0. 2015. Available online: https://www.opengeospatial.
org/standards/wps (accessed on 31 July 2023).
38. Quinn, S. WPS, Turf.js John A. Dutton e-Education Institute, College of Earth and Mineral Sciences, The Pennsylvania State
University, and Spatial Data Processing on the Web. 2022. Available online: https://www.e-education.psu.edu/geog585/node/
778 (accessed on 17 March 2023).
39. Open Geospatial Consortium. OGC®WPS 2.0 Interface Standard. 2015. Available online: https://docs.ogc.org/is/14-065/14-06
5r0.html (accessed on 31 July 2023).
40. Čepický, J.; de Sousa, L.M. New implementation of OGC Web Processing Service in Python programming language. PyWPS-4
and issues we are facing with processing of large raster data using OGC WPS. In The International Archives of Photogrammetry;
Copernicus Publ.: Göttingen, Germany, 2016.
41. Dobesova, Z.; Vavra, A.; Netek, R. Cartographic aspects of creation of plans for botanical garden and conservatories. Int.
Multidiscip. Sci. GeoConfrence SGEM 2013, 1, 653. [CrossRef]
42. Onyıl, H.I.; Yılmaz, M. Realization of Web Based with Turf. JS Library of Spatial Analysis in Open Source Softwares. In
Proceedings of the 1st Intercontinental Geoinformation Days, Mersin, Turkey, 25–26 November 2020; Volume 1, pp. 52–55.
43. Piyathamrongchai, K. Extended Web Direction Service to Avoid Obstacle on Road using Turf.js. In Proceedings of the International
Symposium on Geoinformatics for Spatial Infrastructure Development in Earth and Allied Sciences (GIS-IDEAS), Can Tho,
Vietnam, 22–25 November 2018.
44. Mapbox GL JS. Available online: https://www.mapbox.com/mapbox-gljs (accessed on 28 February 2023).
45. El-Sayed, L.S.; Abdul-Kader, H.M.; El-Sayed, S.M. Performance Analysis of Spatial Indexing in the Cloud. Int. J. Comput. Appl.
2015, 118, 1–4.
46. Netek, R.; Brus, J.; Tomecka, O. Performance Testing on Marker Clustering and Heatmap Visualization Techniques: A Comparative
Study on JavaScript Mapping Libraries. ISPRS Int. J. Geo. Inf. 2019, 8, 348. [CrossRef]
47. Esri. ArcGIS—A Mapping Platform for Your Organization. 2023. Available online: https://www.arcgis.com/ (accessed on 31
July 2023).
48. Kne, L. ArcGIS Online. In The Geographic Information Science & Technology Body of Knowledge (2nd Quarter 2022 Edition); Wilson, J.P.,
Ed.; University Consortium for Geographic Information Science UCGIS: Ithaca, NY, USA, 2022. [CrossRef]
49. GIS Cloud. GIS Cloud: Online GIS Mapping for Operations Management. 2023. Available online: https://www.giscloud.com/
(accessed on 31 July 2023).
50. CARTO. CARTO|Location Intelligence & GIS for Cloud Natives. 2023. Available online: https://carto.com/ (accessed on 31 July
2023).
51. Foursquare Studio. Explore the World with Foursquare Studio. Available online: https://studio.foursquare.com/ (accessed on 31
July 2023).
52. Frisbie, M. Devtools Pages. In Building Browser Extensions: Create Modern Extensions for Chrome, Safari, Firefox, and Edge; Apress:
Berkeley, CA, USA, 2022; pp. 269–299.
53. Statcounter. Browser Market Share Worldwide. 2023. Available online: https://gs.statcounter.com/browser-market-share
(accessed on 4 April 2023).

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.