See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/384697810

Modeling the Land use Land cover changes and its impacts on agricultural land
area by using GIS and RS at Fitche town, Central Ethiopia Modeling the Land use
Land cover changes and...

Preprint · October 2024

DOI: 10.21203/rs.3.rs-5201138/v1

CITATIONS READS

0 46

2 authors, including:

Addisu Bekele
Bule Hora University, Ethiopia
3 PUBLICATIONS 10 CITATIONS

SEE PROFILE

All content following this page was uploaded by Addisu Bekele on 08 October 2024.

The user has requested enhancement of the downloaded file.

Modeling the Land use Land cover changes and its
impacts on agricultural land area by using GIS and
RS at Fitche town, Central Ethiopia
Dereje Ketema
Bule Hora University
Addisu Bekele

Bule Hora University

Research Article

Keywords: Fitche town, LULC Change, CA-ANN Multi-layer perception, MOLUSCE Plugin, QGIS, ENVI.

Posted Date: October 7th, 2024

DOI: https://doi.org/10.21203/rs.3.rs-5201138/v1

License:   This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License

Additional Declarations: The authors declare potential competing interests as follows: The authors
declare no con ict of interest with any individual, group or organization.
Modeling the Land use Land cover changes and its impacts on agricultural
land area by using GIS and RS at Fitche town, Central Ethiopia

Dereje Ketema 1 Addisu Bekele 2

1. Department of Surveying Engineering, College of Engineering and Technology, Bule

Hora University. P.O. Box: 144, Email: [email protected]

2. Department of Geology, College of Natural and Computational sciences, Bule Hora

University. P.O. Box: 144, Email: [email protected]

Abstract
The rapid expansion of urban area rose with population growth, movements and demands have a substantial impact
on land use land cover change (LULCC). The study aims to Modeling and analyze LULC changes and its impacts
on the surrounding Agricultural farmlands in Fitche town central Ethiopia during the periods of 2002 to 2022. The
study employed satellite images of Landsat TM for 2002, Landsat OLI datasets for 2013 and 2022. Support Vector
Machine Algorithm of Supervised Classification has been applied to generate LULC maps of 2002, 2013 and 2022
years by using ENVI software and the accuracy of classified land use land cover maps were checked by confusion
matrix to derive overall accuracy and results were above the minimum and acceptable threshold level. The performed
of modelling Land use land cover (LULC) dynamics have been assessed by Cellular Automata (CA) in Modules of
Land Use Change Evaluation (MOLUSCE) Plugins using QGIS and Land Change Modeler (LCM) model to quantify
the land use transitions between land cover classes, to identify gain and losses of each class categories in relation to
other land cover classes and their spatial trend. Finally, Cellular Automata has been run to model LULC changes in
the study area and to predict future land use changes. The parameter, such as Digital Elevation model (DEM), slope,
aspect and distance from the road are used as spatial variable maps in the processes of learning in ANN-Multi layer
perception to predict their influences on LULC between 2002 and 2022. The actual and projected LULC maps for
2022 indicate a good level of accuracy, with an overall Kappa value of 0.83 and with a percentage of the correctness
88.8 %. Artificial neural networks (ANN)-Multi-layer perception model is then used to forecast changes in LULC
for the years 2050. Generally, the results of this study have shown that there was an increased expansion of built up
areas in the last 22 years from 10.43% in 2002 to 16.92 % in 2013 and to 25.76% in 2022 and agricultural land was
inclined from 66.40% in 2002 to 54.93% in 2013 and decreased to 36.41% in 2022. The simulated result indicate that
built-up area covers 29.76% in 2030, 32.31% in 2040 and 35.74% in 2050 and farmlands was expected to be 29.74%
in 2030, 26.56% in 2040 and 22.01% in 2050 of the total area coverage. The findings assist farmers and policy-
makers in developing optimal land use plans and better management techniques for the long-term development of
natural resources.

Keywords: Fitche town; LULC Change; CA-ANN Multi-layer perception; MOLUSCE Plugin; QGIS; ENVI.
Introduction
Urbanization is worldwide phenomena that emphasize the process of both vertical and horizontal physical
expansion of urban area (Ibido,2020). It is a global trend manifested by rapidly shifting human population
concentrations, land usage, and land cover (Ibido, 2020). Land use and land cover (LULC) patterns changed,
industrialization developed, and social transformation happened during the urbanization processes (Seyam et
al.,2023).LULC changes in response to population growth, anthropogenic-environmental interactions or
environmental degradation, climate change, and other socioeconomic issues (Seyam et al.,2023).Many
discussions have focused on population growth and land demand for settlement, economic and industrial
expansion, and the impact on the environment (Angel,2023). Despite its modest area coverage relative to the
earth's surface, dynamic urban growth processes, notably the expansion of urban population in a broader extent
and urbanized region, have a substantial impact on human and natural environments at all geographic scales
(Terfa et al.,2019).The LULC scenario of a region is the outcome of physical and natural characteristics of the
earth's surface that are used and altered by humans across time and space (Girma et al., 2022).Today, LULC
alterations play a vital part in the study and analysis of global changing scenarios since the data available on
such changes is critical for providing critical input to decision-making of ecological management and
environmental planning for the future (Hassan et al.,2016). The spatiotemporal monitoring of land-use dynamics
is also required to notice the changing demands of an expanding population and gain a better understanding of
the link between natural occurrences and human activities (Ibido, 2020).Land-use changes greatly contribute to
land degradation, land fragmentation, and biodiversity loss (Berihun et al., 2019).Various scholars have studied
changes in land-use patterns to determine the factors on a large scale (Stillwell,2021).The fastest urban
population increase and subsequent urbanization are the primary causes of surface fragmentation of the natural
landscape (Haregeweyn et al.,2012; Berihun, 2019).Urbanization resulted in the city's sub-urban and peripheral
areas expanding (Leulsegged et al.,2012).Continuous urbanization has encroached on valuable agricultural land,
forestland, natural vegetation, and wetlands, among other things (Ibido,2020).The majority of urban area has
been conquered by impermeable surfaces, which have detrimental effects on the ecosystem, hydrological
processes, biodiversity, and micro-climate the ecological habitat water quality, and air quality (Misra et al.,2018).
Land cover change refers to a change in some continuous characteristics of the land, such as vegetation type,
soil conditions, and so on, whereas land-use change refers to a change in how a specific area of land is used or
managed by humans (Patel et al.,2019).This entails the modification of natural land use as a result of
urbanization (Weith et al.,2021).It is worth noting that this change is responsible for a variety of local and global
implications, including biodiversity loss and the consequences for human health, as well as the loss of habitat
and ecosystem services (Patel et al.,2019).It is mainly driven by urban growth and is particularly important now
for developing and underdeveloped countries(Cohen, 2004). The primary reasons of urbanization include
population growth, migration from other areas, industries, the economy, and closeness to resources and basic
services (Sunil Sankhala, 2014).The built-up area is widely regarded as the most important criterion for
quantifying urbanization (Sudhira et al.,2004).The exceptional growth of many urban agglomerations in many
developing countries is the result of a threefold structural change process: the transition away from agricultural
employment, high overall population growth, and increasing urbanization rates(Sunil Sankhala, 2014). The
Urbanization in Sub-Saharan Africa is changing conventional livelihood patterns and displacing agricultural
land usage in many locations (Hashidu et al.,2019). The rate of population growth in both urban and rural areas
is not proportional to the amount of available land. Land is fixed in nature and thus does not increase with
increasing population growth. Expansion of cities has an impact on the areas surrounding them (i.e.,the suburbs)
by changing the natural resource base and converting vegetative land cover to new uses, posing a threat to the
environment and the livelihoods of residents (Hashidu et al., 2019).The Ethiopia has 120 million people, making
it the second most populated country in Africa (Bank, 2021).It is one of the world's least urbanized countries,
with 19.7 percent of its people living in urban areas (Bank,2021).In Ethiopia, as in many other African nations,
urbanization is accelerating, and competition for land between agricultural and non-agriculture is intensifying
in peri-urban areas (Ibido,2020).According to official figures from the Ethiopian Central Statistics Agency, the
urban population is expected to nearly triple from 15.2 million in 2012 to 42.3 million in 2037, growing at a rate
of 3.8 percent per year, and the rate of urbanization will be even faster, at around 5.4 percent per year (Central
Statistical Agency, 2016). This expansion facilitates the acquisition of land by both governmental and private
groups (Central Statistical Agency, 2016) Knowledge of the LULC shift and its impact on agricultural areas in
the town region was critical for improving food security and good land use management (Agidew & Singh,
2017). Recently, information on LULC transformation has become very important to many groups; a large
amount of data from the Earth's surface is collected using remote sensing tools, which provides an excellent
source of data from which updated LULC information and changes can be extracted and analyzed efficiently
(Hassan et al.,2016).It is also critical for evaluating the amounts and rates of deforestation, habitat fragmentation,
urbanization, wetland and soil depletion, and many other landscape-level phenomena. The detection of LULC
changes is the process of recognizing alterations in the condition of a pixel or phenomena by evaluating image
recorded on different dates. (Yousafzai et al.,2021) The best instrument for managing land use land cover change
research and natural resources is the combination of remote sensing (RS) and geographical information system
(GIS). The analysis and tracking of regional and temporal LULC shifts aids scientists, environmentalists,
agriculturalists, politicians, and urban planners (Hamad et al.,2018). Remote sensing techniques aid in the
efficient preparation of natural resources, as well as land management and long-term change dynamics tracking
(Hamad et al.,2018).
On the other hand, LULC transition models usually attempt to anticipate when and how frequently these changes
will occur. Land prediction models like as IDRISI's CA MARKOV, CLUE-S/ Dyna-CLUE, DYNAMICS EGO,
and Land Change Modeler are utilized by researchers all around the world (Hamad et al.,2018). The future
prediction model was incredibly useful in evaluating how prior and future LULC changes may affect soil
erosion, particularly on agriculture (Perovic et al.,2018). In recent years, several spatiotemporal prediction
models, including the Markov chain (MC) model, the Cellular Automata (CA) model, and the conversion of
land use and its effects (CLUE) model, have been created to forecast LULC and their change (Alam et al.,2021).
Among the several spatio-temporal dynamic modelling approaches, the CA model has been widely used for
land-use change analysis (Temesgen et al., 2021). Most researchers are using MOLUSCE (Modules of Land Use
Change Evaluation), a new QGIS plugin that can assess prospective LULC changes and contains a transition
probability matrix (Hamad et al., 2018).The MOLUSCE was created to look at a variety of applications, such
as studying temporal LULC shifts and forecasting future land use, anticipating potential shifts in land cover and
forest cover, and identifying deforestation in sensitive areas (Aneesha Satya et al., 2020). This LULC change
model was generated with GIS software and is based on a multi-criteria analysis methodology (Mzava et
al.,2019).The parameters are dependent on the researcher's response weight to predict their relative impact on
the model (Hassan et al., 2016; Singh et al.,2014)
There have been various studies on LULCC identification using GIS and RS in Ethiopia .However, previous
studies did not address the question of the effects of LULC change on adjacent agricultural land in the urban
context, and they did not utilize the most recent Model to analyses and predict LULC change. Furthermore, no
research on urban LULC change and its effects on agricultural land at Fitche Town central Ethiopi were done in
the study area. Therefore, this study attempted to fill the gap by analyzing changes in LULC patterns and their
consequences on adjacent agricultural area at Fitche town central Ethiopia by using multi-temporal satellite
imagery based on the AC-ANN-Multi-Layer perception approach.

Materials and Methods

Description of the Study Area

The study area is located in North Shoa zone Oromia National Regional State of Ethiopia in the northern
direction of Addis Ababa. The study area is located about 112 km along distance the northern of Addis Ababa.
The research work conducted in North Shoa around Fitche town Central Ethiopia. Geographically, the study
area is located at 9°48′N latitude and 38°44′E longitude with an average altitude of 2,738 m (8,983ft.) above
mean sea level as shown in (Fig.1). Fitche is a rapidly expanding town. In this town the fast growth of population
has been observed in 20th century. The town has four kebeles occupying a total population of around 136,541,
of whom 66,758 were men and of whom 70,143 were women. The rapid growth of population of the city has
put great pressure on the demand for urban spaces. In response to this demand, efforts are being made by the
city government to incorporate the peripheral areas of the city, which is resulting in hastening the expansion of
the built-up area of the city.

Fig.1 Location map of study area

Data source and Methods

Data Source

The primary and secondary data sources used in this study. Geospatial information data were collected to gain
information on spatiotemporal land use land cover change analysis. Collecting accurate and reliable data is the most
determinant factor for any research as it determines the quality of the research. In this particular study, the first steps
was identify the relevant LULCC factors to be considered for analysis and modeling LULC followed by setting
criterion based on international and local rules. The table below outlines the data needed, the types of data utilized
in the study, the formats used, and where the data was acquired (Table 1).
Table 1: Data Source
S. Data Source Data format Resolution Purpose
No
. Satellite image Year

1 Landsat7 ETM+ 2002 to map the earth and

track changes as seen
Landsat 8 OLI 2013 USGS Tiff 30m
from space
Landsat 8 OLI 2022

2 Digital - To provide
Elevation Model USGS Tiff 30m topographical details
(DEM)
3 Structural Plan of Fitche Fitche CAD file - To subset / clip
municipality office boundary
Town

4 Population data Fitche - To explore population

municipality office Number density in the study
area.

Data Processing

The Landsat images of Remote Sensing data utilized in this study were obtained from the United States
Geological Survey (USGS), which covers the whole study area according to (Li et al., 2014). Despite its Medium
spatial resolution and heterogeneous pixel issue, the Landsat images is widely utilized for urban LULC mapping
(Lu & Weng, 2007).The United State Geological Survey included all images from each year that had less than
10% cloud pollution (USGS).Therefore, Landsat images of 2002, 2013, and 2022 were downloaded to proceed
to subsequent phases(Li et al., 2014).
Image Data Processing

After collecting data, the study proceeds through the processes of manipulating, sorting, and pre-processing gathered
data in order to assess the LULC dynamics of the study area which is specifically committed to attaining the objective
of the study area. The satellite image of 2002, 2013, and 2022 was processed using ENVI 5.3 software. The
coordinate system, layer stacking, and layer extraction all took place. After that, Landsat images were corrected for
the atmospheric, sensor, and an illumination variance through radiometric calibration procedures (Chavez, 1989).

Image classification

Image categorization was defined as the extraction of different classes or themes, as well as land use and land
cover categories, from raw remotely sensed digital satellite images (Qrenawi, 2019). It was a strategy for
identifying different elements such as urban land cover, vegetation kinds, manmade structures, mineral
resources, or changes in any of these qualities from satellite images (García-Álvarez and Paegelow, 2022).
Furthermore, the classified raster images can be transformed to vector features (e.g., polygons) for comparison
with other data sets or for calculating spatial attributers (area, perimeter) using various statistical approaches,
including QGIS software.
Land covers classification Scheme

The land use/land cover classes used in this research are based on the AFRICOVER land use and land cover
categorization scheme, which is widely used in East African countries (Di Gregorio & Latham, 2003). Land use
and land cover maps for 2002, 2013, and 2022 were created using supervised classification with the ENVI 5.3
software's support vector machine classification technique (Mariye et al., 2022). The LULC classes were
mapped from digital remotely sensed data using a supervised classification method (training sample procedure)
and field expertise to perform the classification (F.Y et al., 2017). Drive signatures were classified with the
supervised classification type and the support vector machine classification algorithm in ENVI 5.3 software
based on a selected training sample (Mariye et al., 2022). Supervised classification relies on the quantity and
quality of training data to provide high-quality classification results according to (F.Y et al., 2017). The
researcher adjusted the descriptions of some of the land use/land cover classes for the sake of simplicity, taking
into account the study area's land use/land cover diversity. As a result, the images were categorized into key
land-use/landcover classes such as built-up area, farm land, open-space land, and shrub according to (Kaul &
Sopan, 2012), as indicated Table 2 below.

LULC Descriptions

Built-up area Urban and rural built-up including homestead area such as residential, commercial, industrial
areas, villages, settlements, road network, pavements.

Farmland Includes all agricultural lands (both annuals and perennials) and grazing.
Open-space land Includes a surface with little vegetation, open land exposed soil, rocks, and sand and also
include the vacant space & small amounts of grassland.

Shrubs Includes all forest vegetation types (evergreen, deciduous and wetland), and non-forest
vegetation futures that are not typical of forest (pasture grasslands and recreational grasses.

Table 2 Land cover classification scheme for the study area.

Accuracy Assessment

The accuracy assessment stage is the ultimate level of satellite image classification (Abbas & Jaber, 2020). The
term accuracy is commonly used to represent the degree of precision of a map or classification derived from
remotely sensed data in thematic mapping (Congalton, 2001). Accuracy assessment is a measurement of
estimation using a remotely sensed dataset to classify circumstances, and it is valuable for evaluating
classification approaches as well as determining the inaccuracy that may be involved (Abbas & Jaber, 2020).
The expanded use of remote sensing data and methodologies has made geospatial analysis faster and more
powerful; nevertheless, the added complexity has also raised the possibility of errors (Gelan, 2021).As a result,
evaluating the accuracy of mapping created from remote sensing data is an important and necessary stage in the
categorization process (Gelan, 2021).The error matrix is the most commonly used tool for determining the
accuracy of categorization results (Fan et al., 2007). User accuracy, producer accuracy, overall accuracy and
Kappa statistics were then derived from the error matrices for LULC classes. According to Moller-Jensen (1997),
Kappa analysis is a discrete multivariate technique used in accuracy assessments. In other words, Kappa analysis
is a standard component of accuracy assessment and is regarded an essential component of most image analysis
(Abbas & Jaber, 2020).

Producer’s Accuracy (Errors of omission):- Producer‟s Accuracy measures the percentage of correctly
identified pixels in a sample data set or, more indirectly, mistakes of omission for a specific class (Kaul and
Sopan, 2012). The accuracy of the producer was determined by dividing the number of successfully identified
pixels in each category (on the principal diagonal) by the number of training set pixels utilized for that category
(the column total) calculated according to (Kaul and Sopan, 2012) as follows using (Eq.1).

Producer‟s Accuracy = 100 Eq. (1)

Where, Ci = correctly classified sample locations of the reference data or column and Ct = total number of
sample locations of the column.

User’s Accuracy (Commission error):- The user accuracies are calculated by dividing the number of correctly
categorized pixels in each category by the total number of pixels classified in that category (the row total)
according to (Kaul and Sopan, 2012). This outcome is a measure of commission error calculated as follows
using (Eq.2).

User‟s Accuracy 100 Eq. (2)

Where, Ri = correctly classified samples in the row and Rt = total

number of samples in the row.
Overall Accuracy:- Overall accuracy was calculated by dividing the total diagonal (the sum of properly
categorized sample units) by the total number of sample units in the error matrix (Kaul and Sopan, 2012).This
is the most frequently mentioned accuracy assessment statistic as indicated (Eq.3).

Overall Accuracy = 100 Eq. (3)

Where, Sd = sum of values along diagonal and, n = total number of samples.

Kappa Coefficient:- The Kappa coefficient is another measure of agreement or accuracy (Foody, 2020). Kappa
was used to calculate the degree of agreement or accuracy between the remote sensing-derived categorization
map and the reference data, as represented by the major diagonals and the chance agreement (Foody, 2020).
This is the row and column totals demonstrated. According to Foody (2020) the Kappa values can range from
+1 to -1. However, because there should be a positive correlation between the remotely sensed categorization
and the reference data, positive values were expected. According to Landis and Koch (1977) divided the possible
kappa ranges into three categories: values greater than 0.80 (i.e., 80 percent and above) represent strong
agreement; values between 0.40 and 0.80 (i.e., 40 percent -80 percent) represent moderate agreement; and values
0.40 (i.e.,40 percent and below) represent poor agreement. According to Foody (2020) It’s determined as the
Eq.4 below.

K^ Eq. (4)

Where r = number of rows in the error matrix; Xii = number of observations in row i and column i (on the major
diagonal); Xi+ = total of observations in row i (shown as marginal total to right of the matrix); X+i = total of
observations in column i (shown as marginal total at bottom of the matrix) and N = total number of observations
included in matrix.

Data Analysis Method

The study used both qualitative and quantitative methodologies to address the objective. The methodology
included using geospatial tools and remote sensing data to create LULC maps for 2002, 2013, and 2022, as well
as analyzing the effects of LULCC on agricultural land. The transition probabilities from the ANN learning
process are used in the Cellular Automation model to characterize the LULC changes (Saputra and Lee, 2019).
This approach use the (MOLUSCE) plugin in QGIS 2.18 software (Fig.2). The MOLUSCE plugin has six LULC
prediction phases (Hakim et al.,2019).
LULC Change Analysis

The Panel of Change Analysis provides a quick quantitative assessment of changes, allowing researchers to
provide gain and loss evaluations, net change, persistence, and specific transitions in both map and graphical
formats (Saputra and Lee, 2019). The main issue is defining the dominant changes that can be grouped and
modelled, referred to as sub-models (Christman et al., 2016). Change detection was evaluated in this study
utilizing two software programmers, ENVI 5.3 and QGIS. The image classification technique in ENVI software
was used to construct LULC maps for the years 2002, 2013, and 2022 that produces change maps from 2002 -
2013, 2013-2022, and 2002-2022 in the QGIS MOLUSCE Plugin. The method acts on continuous data by
subtracting pixel values from the prior theme from the after theme and producing two outputs: image difference
and highlight change image (Kafy et al.,2021). Because change detection computes the change in brightness
values over time, the image difference file displays those changes with a grey scale image made of single band
continuous data. The highlight changes file creates a thematic layer clustered into decreased, unchanged, and
increased classes (Kafy et al.,2021). Based on the land change analysis principle, maps of gains and losses,
contributions to net change, transitions of land cover classes across different categories, and spatial trend analysis
were analyzed in both map and graphical form.
Total LULC changes in hectares were calculated according to Kafy et al.,2021,using the following (Eq.5).

Total LULC change = Area final year – Area initial year Eq. (5)

Where, area is extent of each LULC type, Positive values suggest an increase whereas negative values imply a
decrease in extent.
Percentage LULC change were calculated also according to Kafy et al.,2021 using the following (Eq.6).

Percentage LULC changes Eq. (6)

Where, area is extent of each LULC type. Positive values suggest an increase whereas negative values imply a
decrease in extent according to Kafy et al.,2021 by using the Eq.7 below.

Rate of change (ha/yr.) Eq. (7)

Transition Potential Modeling

While there are various approaches for constructing transition potential maps, the MOLUSCES plugin supports
artificial neural networks (ANN), weights of evidence (WoE), logistic regression (LR), and Multi-Criteria
Evaluation (MCE) (Kamaraj & Rangarajan, 2022). In this study, a multi-layer perceptron artificial neural
network (MLPANN) based on the geographic significance of LULC transitions was utilized to examine LULC
trends between 2002 and 2022. It is possible to generate class statistics as well as a transition matrix of land use
and land cover change (Saputra and Lee, 2019).The class statistics describe the changes in LULC that happened
between the first and final periods. The transition matrix indicates the number of pixels that move from one
LULC type to another (Saputra and Lee, 2019). The results of this analysis were utilized to forecast future land
use land cover using MOLUSCE (Modules for Land Use Change Evaluation).) (Kamaraj & Rangarajan, 2022).
Sensitivity of ANN-CA

There are several methods for producing a transitional potential map; this module incorporates computational
intelligence features such as artificial neural networks (ANN).LULC data is used as an input in all approaches
for calibrating and modelling LULC change (Saputra and Lee, 2019). This method is appropriate when dealing
with challenges where the algorithm must deal with massive amounts of unknown or difficult-to-implement
input data (Kafy et al.,2021). As a result, a continuous index is constructed that describes the terrain on a scale
of 0 to 1 (Kamaraj & Rangarajan, 2022). Because ANN requires fuzzy logic, a continuous range, such as 0 and
1, is established based on terrain usability. The interactions between linked neurons, as well as the changing of
the weight connections between them, are critical components of ANN (Kamaraj & Rangarajan, 2022).The
following parameters were finally determined while projecting the LULC map for the year 2050: neighborhood
- 1, iterations - 411 1000 nos., hidden layer - 10 nos., momentum value - 0.06, learning rate - 0.001 (Kamaraj &
Rangarajan, 2022).
Selection of Spatial Variables

The potential power of the independent variables must be included in the LULC change simulation (Gharaibeh
et al., 2020). This study's key spatial factors were: The distance from the road, which determines accessibility,
is a crucial motivator in promoting greater urban uses and expansion (Gharaibeh et al., 2020). Distance from
urban can be a powerful engine of change, with the closer the land is to urban centers, the easier it is for land to
be converted to urban uses (Gharaibeh et al., 2020). The most major anthropogenic determinant in land use
change is population density, with higher densities resulting in more frequent land use change. Anthropogenic
disturbance is strongly related to land cover type (Chowdhury et al.,2021). Elevation is acknowledged as one of
the most important topographic elements influencing LULC evolution (Girma et al.,2022). Slope influences the
spatial trends of land cover change. Assume that the softer a land slope, the easier it is to alter land use
(Gharaibeh et al., 2020). The correlation of spatial variables between the two raster images, which are used to
investigate the correlation among the spatial variable factors, is evaluated using coefficients (Hakim et al.,2019).
The category of each region and the LULC changes are then calculated between the first year (2002) and the
last year (2022).The method also generates the transition matrix, which represents the percentage of pixels that
change from one kind to the next.
Model Validation and LULC Prediction

The Validation is performed at this stage by computing the total kappa value between the reference map (real
LULC map) of 2022 and the simulated LULC map 2022.(Hakim et al., 2019). Return to the cellular automata
simulation stage to predict future LULC after the total kappa value meets the assessment standard. (Hakim et
al.,2019).The LULC maps of 2030, 2040, and 2050 were forecasted using the LULC maps of 2002, 2013, and
2022, as well as the same spatial variable factor combinations.
The flow chart below describes the general analytical procedure used to carry out this study effort as shows
(Fig.2).
 Data Set

Satellite imagery Master plan/

Structural Plan

Landsat-7 0f Landsat-8 0f Landsat-8 0f

2002 2013 2022

Image Pre-processing (Mosaicking,

GPS Data
Subset/clipping, and Layer stacking)

Image Classification Google

Accuracy Assessment

Generate LULC map of the study

period

Change Analysis (CA-ANN)

Validation and Pred iction (CA-ANN)

Predicting the LULC maps of 2030,

2040 and 2050 (CA)

Fig. 2 General flow chart of research work of the study area

Results and Discussions

Dynamics of Land Use Land Cover Changes

This section investigates trends in urban LULC through an analysis of Fitche town's physical expansion and area
coverage over the past 20 years. Several data sets were examined to determine how changes in LULC patterns
affected the land use class. Land cover maps created by supporting vector machine supervised classification
with ENVI software, followed by a post-classification algorithm. On the basis of the collected data and the
analysis performed, LULC maps were developed for 2002, 2013 and 2022, as shown in (Fig.5) and (Fig.6). The
classified Landsat satellite images summarized in (Table 6) demonstrate the spatial extents, percentage changes
over time and the annual rate change of the four LULC classes (built-up area, agricultural land, open space and
shrubs) in Fitche town between 2002 and 2022. As shown in Figure 10, the LULC of all three reference years
had changed over time. The results show that the LULC maps in Fitche town have changed spatially over time.
During the study period, farmland was the dominant land class, but it declined as built-up areas and open spaces
increased. From 2002 to 2022, the proportion of farmland in the total area ranged from 66.4% to 36.4%. The
LULC spatial distribution from 2002 to 2022 is shown in Figures 5 and 6. From 2002 to 2022, Tables 3, 4 and
5 summarize the percent of area covered by each LULC class.

Accuracy Assessment

In general, the study was accurate to 94.9%, 96.6%, and 98.7%, with Kappa statistics of 0.932, 0.955, and 0.983,
respectively, for 2002, 2013 and 2022. This results in an overall accuracy of greater than 85% and a Kappa
coefficient greater than 0.80, indicating strong agreement. As a result, the maps met the accuracy standards for
change detection analysis, and there is a positive correlation between the remotely sensed classed samples and
the reference data, as shown in Table 3.
Table 3: Confusion matrix for the land use land cover map of 2002

Reference Map

Built-
Farmland Open- Grand
Up Area Shrubs Space -Total User Accuracy

Built-Up Area 28 0 0 1 29 96.6%

Farmland 0 26 1 1 28 92.9%

Shrubs 0 0 29 1 30 96.7%

Open-Space 0 2 0 28 30 93.3%

Grand-Total 28 28 30 31 117
Producer 96.7%
Accuracy 90.3%
100% 92.9%
Overall Accuracy 94.9%

Overall, Kappa Coefficients

= 0.932

0.932

Table 4: Confusion matrix for the land use land cover map 2013

Reference Map

Built-Up
Farmland Open- User
Area Shrubs Space Grand Total Accuracy
Built-Up Area
37 0 0 0 37 100%
Farmland
0 36 0 2 38 94.7%
Shrubs
0 1 35 0 36 97.2%
Open-Space
0 2 0 34 36 94.4%

Grand-Total 37 39 35 36 147

Producer Accuracy 100% 92.3% 100% 94.4%

Overall Accuracy = 96.6%
Overall, Kappa Coefficients = 0.955
Table 5: Confusion matrix for the land use land cover map of 2022

Reference Map

Open- Grand User

Built-Up Area Farmland Shrubs
Space -Total Accuracy

Built-Up Area 39 0 0 1 40 97.5%

Farmland 0 40 0 0 40 100%

Shrubs 0 0 36 0 36 100%

Open-Space 0 1 0 38 39 97.4%

Grand-Total 39 41 36 39 155

100% 97.4%
Producer 97.6%
100%
Accuracy

Overall Accuracy = 98.7%

Overall, Kappa Coefficients = 0.983

During the study period, the results demonstrate an uneven shift in land use land cover due to fast urban
expansion in Fitche town. In 2002, the built-up area accounted for 10.43 percent of the total area Within an 11-
year period, it gradually grew to 16.92 percent. This proportion increased to 25.76 percent in the last year.
Temporarily, the pace of built-up expansion was comparably low from 2002 and 2013, with a nearly 6-
percentage-point increase, but it grew by about 10-percentage points, reaching around 26 percent of total area
in the last year. In comparison, at the start of the study period, the area statistics of farmlands were 66.4 percent.
It marginally reduced to 54.93 percent in 2013 year, these numbers rapidly decreased to 36.4 percent, indicating
a linear fall in agriculture and halving from its initial spatial extent. Open-space area, on the other hand, increased
from 20.3 percent to 36.53 percent throughout the research period, with a 3.65 percent annual increase rate. By
comparison, over the study period, the area coverage of shrubs land declined from 2.9 percent to 1.3 percent in
Fitche town.
 a) b)

c)

Fig.3: LULC Map of Study Area in a 2002, b 2013 and c 2022

During 2002-2013, built-up area increased from 361.30 ha to 586.07 ha (10.435% - 16.93%) and from 586.07
ha to 891.89 ha (16.93% - 25.75%) between 2013 and 2022.Furthermore, open space increased from 702.31 ha
to 910.92 ha (20.28% - 26.30%) between 2002 and 2013, and from 910.92 ha to 1265.4 ha between 2013 and
2022. During the period 2002-2013, farmland and shrubs areas declined from 2299.65 ha to 1902.54 ha (66.4%
- 54.93%) and 100 ha to 63.73 ha (2.89% - 1.84%) respectively. Similarly, farmland and shrubs declined from
1902.54 ha to 1261.18 ha (54.93% -36.42%) and 63.73 ha to 44.79 ha (1.84-1.3%) from 2013 to 2022. In general,
the LULC maps, area statistics, and annual rate of change were displayed in (Fig.3a, b,and c) and (Table 6).

Table 6 Area statistics and Annual rate change of the LULC from 2002-2022

LULC Types 2002 2013 2022

ARC

(%) (%)
(ha) (%) (ha) (%) (ha)

Built-up Area 361.30 10.43 586.07 16.92 891.89 25.75 6.69

Farmland 2299.65 66.40 1902.54 54.93 1261.18 36.42 -2.05

Open-space 702.31 20.28 910.92 26.30 1265.4 36.55 3.65

Shrubs 100.00 2.89 63.73 1.84 44.79 1.29 -2.5

Total 3463.26 100% 3463.26 100% 3463.26 100

Land Cover Variations of Built-up Area

In this study period, the LULC pattern has experienced spatial dynamic variations. According to the results from
2002 to 2022, built-up areas have expanded significantly and farmland and shrub land have shrunk. The (Fig.4)
and (Table 7) illustrate the spatial-temporal area and percentage change in built-up area. According to the study,
in 2002 there was 10.4% of built-up area, 16.9% in 2013 and 25.8% in 2022.
Table 7 Built-up and non-built-up areas (2002- 2022).

LULC 2002 2013 2022

Ha % Ha % ha %

Built-up Area 361.30 10.43 586.07 16.92 891.89 25.76

Non-Built-up Area 3102 88.57 3077 83.08 2571 74.24

30

25

20

15

10

0
2002 2013 2022

Fig.4 Trends of Built-up Area from 2002-2022

 a) b)

c)

Fig. 5 built up and non-built up map of a 2002, b 2013 and c 2022

Change trends and Magnitude of Agricultural lands over the study period

The agricultural land was exported from classified LULC maps in order to visualize and examine the spatial
decline of agricultural land during the study time periods. As a result of the horizontal expansion and conversion
of land cover classes during the distinct study periods, farmland areas contributed most to the increase of built-
up areas. In the years 2002 to 2022, there was a dramatic decrease in farmland areas. The Fig.6 shows the
spatiotemporal destruction pattern of farmland in the study area. During the study period at Fitche town, the
proportion of farmland areas was 66.4% in 2002, 54.93% in 2013, and 36.31% in 2022 in table 8.

Table 8 Trends of agricultural land during the study period

LULC 2002 2013 2022

Ha % Ha % Ha %

Farmland 2300 66.4 1903 54.93 1261 36.31

Farmland Area (%)

70
60
50
40
30
20
10
0
2002 2013 2022

Farmland Area (%)

Fig.6 Change Trends of Farmland Area from 2002-2022

 a) b)

c)

Fig.7 Distributions of farmland map in a 2002, b 2013 and c 2022

LULC Change Analysis

After the land use maps were prepared, the changes occurring in each land class during the studied periods were
analyzed. Land cover in research areas such as shrub land and farmland has been steadily declining over the last
two decades. In general, built-up areas and open-space in the study region have steadily increased over the last
20 years. The change of spatial area extent in hectares, percentage change and the annual rate change within the
study area of the each individual land cover categories were presented in (Table 9 and table 10) below over the
period of 2002 to 2022.
Table 9 Change of Land Use Land Cover between 2002 and 2022.

LULC Types 2002-2013 2013-2022 2002-2022

Δ(ha) Δ (%) Δ(ha) Δ (%) Δ(ha) Δ (%)

Built-up Area 224.8 62.2 305.4 52.1 530.59 146.9

Farmland -397.1 -17.3 -640.5 -33.7 -1038.47 -45.2

Open-space 208.6 29.7 354.03 38.9 563.09 80.2

Shrubs -36.3 -36.3 -18.94 -29.8 -55.21 -55.2

Table 10 Annual Rates and trends of Dynamics occurred in LULC over the study period

LULC Types 2002-2013 2013-2022 2002-2022

ARC(ha) ARC(ha) ARC(ha)

Δ(ha) Δ(ha) Δ(ha)

Built-up Area 224.8 20.4 305.4 33.9 530.59 26.5

Farmland -397.1 -36.1 -640.5 -71.2 -1038.47 -51.9

Open-space 208.6 18.9 354.03 39.3 563.09 28.2

Shrubs -36.3 -3.3 -18.94 -2.1 -55.21 -2.8

The LULC change analysis investigates the spatial dynamic variations in the LULC pattern during the study
period. The amount of change that occurred between 2002 and 2022 was extensive, as seen in (Tables 9 and
10).The data reveal that between 2002 and 2013 (11 years), the areas covered by built-up area and open space
land increased by +224.8ha (+62.2%) and +208.6 ha (+29.7%), with annual rate changes of 20.4ha and 18.9ha,
respectively, indicating a significant increasing trend. However, farmland and shrub cover declined by -17.3
percent (-397.1ha) and (-36.3%) -36.3ha, respectively, at annual rates of -36.1ha and -3.3ha, indicating a
downward trend.
The results also show that between 2013 and 2022 (9 years), the built-up area increased by +305.4ha of land,
accounting for +52.1 percent of the built-up area, at an annual rate of 33.9ha, while farmland decreased by -
640.5ha, accounting for -33.7 percent of farmland, at an annual average rate of -71.2ha. This demonstrates that
urban land mass increased fast between 2013 and 2022 (305.4ha) compared to 2002 to 2013.
(224.8ha).Furthermore, the area covered by open space expanded by +354.03ha (+38.9%) with an annual rate
change of 39.3ha, while the area covered by shrubs declined by -18.94 (-29.8%) with an annual rate change of
-2.1ha.
Over the study period (2002-2022), the area had significant land-use changes, with 530.59ha (146.9 %) of built-
up areas rising at an annual rate of 26.5ha/per, indicating a consistently increasing tendency.In comparison, at
an annual rate of -51.9ha/yr, 1038.47ha (-45.2%) of farming was transferred to other land uses. while open space
land coverage grew by 563.09ha (+80.2%). The area covered by shrubs declined by 55.2 percent throughout this
time period, with an annual rate change of -2.8ha.
During the study period, there was a considerable growth in built-up area, an increase in open-space land, and a
shrinkage phenomenon in farmland and shrubs (2002 to 2022).Due to the associated horizontal growth and
conversion of land cover classes during the various study periods, the study area has realized spatial variations
and trends (Fig.7) on different land use and land cover classes.. The reclassified images in (Fig.9 10 and 11),
showed that there had been a rapid land cover conversion of each LULC categories to another classes.

Trends of LULC Change Analysis of 2002-2022

70.00%

60.00%

50.00%

40.00% 2002
2013
30.00%
2022
20.00%

10.00%

0.00%
Built-up Area Farmland Open-space Shrubs
LULC Classes

Fig.8 Trends of LULC Change from 2002 to 2022

 Fig.9 LULC change map of 2002-2013.

Fig.10 LULC change map of 2013 - 2022.

 Fig.11 LULC change map of 2002-2022

LULC Transition Analysis

Analyzing spatiotemporal changes within a collection of LULC categories requires the use of the transition
matrix. The proportions of pixels that shift between different land use categories are shown in the matrix. The
LULC categories in the first year are represented by the rows of the matrix table, and the final year's categories
are shown in the same sequence in the columns. The diagonal entries indicate the degree of class stability,
whereas each off-diagonal entry shows the degree of class transition. A diagonal entry's stability is shown by
values that are near to 1. Researchers compare the temporal changes in various regions primarily using transition
matrices. We utilized the MOLUSCE plugin to construct the transition potential matrix for the years 2002-2013,
2013-2022, and 2002-2022, based on the explanatory variables and current LULC circumstances, in order to
illustrate the expected changes for each LULC type in our study area. Build-up area and farmland area were the
most stable categories, with probabilities of 0.783 and 0.689, respectively, during the period of 2002–2013, as
shown by the transition potential matrix (Table 11). The probability value of 0.082 of farmland was contributed
to the built-up area, as shown on the transition potential matrix. Additionally, open space and shrubs received
contributions from farmed area with probability values of 0.214 and 0.015, respectively. However, the
probability matrix value of the open-space transition probabilities remained constant at 0.556. Open space land
contributed 0.142, 0.299, and 0.002 correspondingly to built-up area, farmland and shrubs respectively. The
transition probabilities of shrubs remained constant at 0.256, whereas contributions to farming, open-space land,
and built-up area were 0.141, 0.456, and 0.147, respectively.
The diagonal values of the cross tabulation matrix for the years 2013–2022 (Table 12) indicate LULC that
remained constant during that time. Built-up, farmland, open-space, and shrubs had transition probability matrix
values of 0.866, 0.578, 0.743, and 0.382, respectively, which remained constant during the time period. Built-
up, open space, and shrub land all received contributions from the agricultural probability values of 0.129, 0.285,
and 0.009, respectively. The Open space land contributed 0.142, 0.114, and 0.001 to built-up, farmland and
shrubs, respectively. The shrubs class supplied 0.152, 0.434, and 0.031 values to agriculture, open spaces, and
built-up areas, respectively. Furthermore, the transition matrix between 2002 and 2022 (Table 13) revealed that
built-up area was still stable with a value of 0.833, while farmland, open-space, and shrubs had fragmentation
values of 0.452, 0.572, and 0.180, respectively. This is because LULC data from 2002–2022 was used along
with spatial factors to predict the LULC map of 2050 and the transition probability matrix. Farmland provided
0.182 of its transition probability value to built-up area, 0.356 to open space, and 0.010 to shrub areas. Moreover,
the transition values 0.003 of open space was turned into shrubs, 0.217 of open space land was turned into
farming, and 0.208 of open space was transformed into built-up areas. The shrubs feature class also contributed
to the built-up, agricultural, and open space area classes with probability values of 0.275, 0.301, and 0.244,
respectively. The areas that were acquired and lost for each category throughout each time interval are detailed
in Figs. 9, 10, 11, and Table 14.
Table 11 Transition matrix from 2002-2013.

2013

Categories
Built-up Area Farmland Open-space Shrubs

Built-up Area 0.782857 0.169938 0.038758 0.008447

Farmland 0.082322 0.689449 0.21367 0.01456

Open-space 0.142127 0.29908 0.556493 0.002301

Shrubs 0.140934 0.456014 0.147217 0.255835

Table 12 Transition matrix from 2013-2022.

2022

Categories
Built-up Area Farmland Open-space Shrubs

Built-up Area 0.86636 0.052567 0.077088 0.003985

Farmland 0.128606 0.57764 0.284547 0.009206

Open-space 0.142069 0.114286 0.743054 0.000591

Shrubs 0.152327 0.434415 0.03103 0.382228

Table 13 Transition matrix from 2002-2022.

2022

Categories
Built-up Area Farmland Open-space Shrubs

Built-up Area 0.832712 0.105891 0.058663 0.002734

Farmland 0.181293 0.452374 0.355949 0.010384

Open-space 0.20803 0.217491 0.571666 0.002813

Shrubs 0.274933 0.300988 0.244385 0.179695

Table 14 Gains and losses of each category

LULC Categories 2002- 2013 (ha) 2013 – 2022 (ha) 2002 – 2022 (ha)

Built-up Area 224.8 305.4 530.59

Farmland -397.1 -640.5 -1038.47

Open-space 208.6 354.03 563.09

Shrubs -36.3 -18.94 -55.21

Selection of Spatial Variables

This model's initial phase was to incorporate the LULC maps for the start year (2002) and the end year (2022).
From the land cover change map, which was produced by feeding the model with spatial variable parameters
including slope, distance from a road, and distance from a built-up region, the study area's changing pattern
between 2002 and 2022 was detected. For each dataset, the attributes of the explanatory maps are extracted using
the same raster format, with a resolution pixel size of 30 m, and the same geographical projected coordinates of
the UTM 37N zone. In addition to calculating the percentage of area change in a particular year, the MOLUSCES
plugin produces a transition matrix that illustrates the percentage of pixels that change from one land use cover
to another. Furthermore, the plugin generates an area change map that illustrates the land's transformation in
built-up areas, farms, open spaces, and bushes across all classes between 2002 and 2022. ANN-(Multi-layer
perception) was the plugin used to project the change in LULC (Mzaya et al.,2019) Moreover, LULC transitions
were projected for the years 2030, 240, and 2050 using the categorized raster images of 2002 and 2022. Predicted
LULC maps assume that the current LULC dynamics and pattern will persist.
The correlation of geographic variables between the two raster images, which are used to examine the correlation
among the spatial variables factors, is evaluated using Pearson‟s correlation, Crammer's coefficient, and Joint
information uncertainty. (Hakim et al.,2019). According to Hakim et al. 2019, next, each area's categorization
and the LULC changes between the first year (2002) and the last year (2022) are computed. The program also
generates the transition matrix, which shows the percentage of pixels that switch between different types. The
method also generates the transition matrix, which shows the percentage of pixels that change from one kind to
the next.

Fig.12 Spatial parameter maps of distance from built-up and distance from major road
 Fig.13 spatial parameter maps of DEM and Slope.

Model Validation and Prediction

Transition Potential Modeling and Model Validation

The MOLUSCE plugin incorporates several popular transition potential modeling algorithms, including logistic
regression, weights of evidence, multi-criteria evaluation, ANN (multilayer perceptron), and CA algorithm, for
use in upcoming simulations. According to Cramer's coefficient, the spatial variables for model calibration were
selected because of their comparatively strong connection with LULC. The transition potential modeling and
prediction in this study was done using the CA-ANN technique. The combining spatial variables with LULC
data from 2013 to 2022, LULC for 2022 was projected, yielding a validation kappa value of 0.95. Once the
predicted LULC was produced, the actual LULC of 2022 was compared to the anticipated data, yielding an
overall accuracy of 88.8%, Kappa (overal) of 0.83, Kappa (histo) of 0.90, and Kappa (loc) of 0.93.The (Fig.14
) and (Table 9) show the actual and forecasted maps and statistics for 2022.
Table 15 spatial extent for Actual and projected LULC maps of 2022.

LULC Actual 2022 Projected 2022 Kappa Value

Ha % Ha % Accuracy ANN Validation

Built-up Area 891.89 25.75 1031.1 29.77

Farmland 1261.18 36.42 1030.4 29.73 88.8 0.95 0.83

Open-space 1265.4 36.55 1347.2 38.90

Shrubs 44.79 1.29 55 1.56

Fig.14 Actual and Projected LULC 2022

Prediction of LULC

During the study period, the spatial dynamic variations in the LULC pattern are examined using the LULC
change analysis. The LULC for 2030, 2040, and 2050 were estimated following the acquisition of good results
from model validation. LULC for 2030, 2040, and 2050 were predicted by utilizing the transition probability
matrix (Table 14), the spatial variables, and the temporal LULC data from 2013 and 2022. A kappa value of 0.95
was found. (Fig. 15a, b, and c) and (Table 16) show the projected LULC for 2030, 2040, and 2050 together with
accompanying area information.
The simulation results were presented for the Molusces model projection for the years 2022–2030.The built-up
area will increase from 891.89ha in 2022 to 1043.6ha in 2030. Farmland area will decrease from 1261.2ha in
2022 to about 1030.3ha in 2030. The open space land will increase from 1265.4ha to about 1347.2ha during the
period between 2022 and 2030. The area of shrubs land will decrease from 44.79ha in 2022 to 42.29ha in 2030.
Similarly, the built-up area is expected to rise from 891.89ha in 2022 to approximately 1140.5ha in 2040,
according to the simulation model running from 2022 to 2040.Farmland area cover will increase from 1261.2ha
in 2022 to about 920.2ha in 2040. The area of open space land will increase from 1265.4ha in 2022 to nearly
1363ha in 2040. Shrubs land will decrease considerably from 44.79ha in 2022 to 40.2ha in 2040 during the study
period of 2022 and 2040years.
According to model simulation showed that the built-up area will expand from 891.89ha in 2022 to 1268.7ha in
2050 due to changes in land cover classes between 2022 and 2050. On the other hand, farmland is predicted to
decline, from 1261.2 hectares in 2022 to 762.2 hectares in 2050. However, open space area 1265.4ha covered in
2022 will be increased to 1400.3ha in 2050. Furthermore, between the years 2022 to 2050, the covered area of
shrubs will drop from 44.79ha to 35.5ha. For the LULC category, Tables 15 and 16 generally showed the
predicted changes in areal extent, percentage change (%), and annual rate change. The findings indicate that,
between 2022 and 2050, the built-up area will continuously rise by 376.81 hectares, with a percentage change
of 42.2 percent and an annual rate change of 13.5 hectares each year. In contrast, agriculture lands decreased by
499 hectares, with a percentage change of -39.6% and an annual rate change of -17.8 hectares per year. In
addition, shrub land declined by -9.29ha (-20.7 percent) at an annual rate of -0.3ha/yr between the years of 2022
and 2050 timeframe, as shown by the simulation model, which also showed an increase in open space land of
134.9ha and a percentage change of 10.7 percent at an annual rate of 4.8ha/yr.
 b)
a)

c)

Fig.15 Predicted LULC maps for the year’s a 2030, b 2040 and c 2050
Table 16 Spatial area extent of predicted LULC of 2030, 2040 and 2050 years
LULC
2022 2030 2040 2050
Classes

Area(ha) % Area(ha) % Area(ha) % Area(ha) %

Built-up Area 891.89 25.8 1043.6 29.8 1140.5 32.3 1268.7 35.7

Farmland 1261.2 36.4 1030.3 29.7 920.2 26.6 762.2 22.0

Open-space 1265.4 36.6 1347.2 38.9 1363 39.4 1400.3 40.4

Shrubs 44.79 1.29 42.6 1.60 40.2 1.77 35.5 1.8

Table 17 Spatial area extent change and percentage changes of LULC from 2022 to 250

LULC Classes 2022-2030 2022-2040 2022-2050

Area(ha) % Area(ha) % Area(ha) %

Built-up Area 151.71 17.0 248.61 27.9 376.81 42.2

Farmland -230.9 -18.3 -341 -27.0 -499 -39.6

Open-space 81.8 6.5 97.6 7.7 134.9 10.7

Shrubs -2.19 -4.9 -4.59 -10.2 -9.29 -20.7

Table 18 Spatial area extent change and Annual Rates change of LULC from 2022- 2050

2022-2030 2022-2040 2022-2050

LULC Classes

Area(ha) ha/yr. Area(ha) ha/yr. Area(ha) ha/yr.

Built-up Area 151.71 19 248.61 13.8 376.81 13.5

Farmland -230.9 -28.9 -341 -18.9 -499 -17.8

Open-space 81.8 10.2 97.6 5.4 134.9 4.8

Shrubs -2.19 -0.3 -4.6 -0.3 -9.29 -0.3

The MOLUSCE plugin for QGIS software is used in this work to create a transition probability matrix for each
interval. The analysis of the LULC shift from 2002 to 2022 is done utilizing spatiotemporal LULC data and
accessible spatial variable factors. The study's use of the CA-ANN multilayer perceptron technique with the
MOLUSCE plugin to anticipate the LULC map for 2030, 2040, and 2050 offers an answer to the research topic
of how the LULC has changed over the previous 20 years (Fig.15a b and c). The influence of land use and land
cover change (LULC) on agricultural land in the study area was evaluated using the existing and projected land
cover maps. In the research area throughout the study period, most LULC categories exhibited declining trends
in farmland and shrubs, while open-space showed slight increases, over the previous 20 years, built-up area was
the land cover that showed the greatest expansion. This suggests that there has been an ongoing increase in
small-scale open space and built-up areas, leading to a decrease in agricultural land in the research area. The
Providing pertinent information about changes with regard to their extent, trend, location, and geographical
distribution is the most widely held view of the application of change analysis. According to the current study
and previous research, unsustainable agricultural land conversion in urban periphery areas has detrimental
effects on farmlands and crop yields, endangering the livelihoods of the local population. The Urban LULC
Change has a variety of effects on farmers in periurban settings, according to Ayele & Tarekegn (2020).
Furthermore, Efa & Gutema (2017) contended that the fast conversion of agricultural lands into other land uses
brought about by urbanization negatively affects farmers, rendering them landless and forcing them to relocate
farther from their homes in order to obtain land for farming. In order to improve the prediction of future land
cover, it is crucial that the study highlights the pattern of changing land cover. These results show that rapid
urban LULC dynamics have caused a dramatic transformation of the Fitche town LULC during the past 20
years, most notably in the fast conversion of farmland and shrubs to built-up areas in the study area. Moreover,
during the study period, small-scale farmlands and shrub classes were transformed into open space. As seen in
Table 3, between 2002 and 2022, the built-up portion expanded from 10.43 percent to 25.76 percent, while the
open-space portion increased from 20.28 to 36.53 percent. Nonetheless, the percentage of farmland fell from
66.4 to 36.42%. Furthermore, the results of the future simulation indicate that, between 2022 and 2050, the
proportion of built-up areas would rise by 25.73 to 35.74 percent, while the percentage of open-space will rise
by somewhat, from 36.55 to 40.41 percent. From 2022 to 2050, it is anticipated that the percentage of farmlands
will drop from 36.42 percent to 22.01 percent. The Natural resources, the ecosystem, and food security may
ultimately be lost as a result of significant changes in LULC, notably urban expansion and farmland
fragmentation. As a result, policymakers will benefit from the spatiotemporal and future LULC simulation
results in assessing the shift in LULC intensity and the socioeconomic factors influencing it, as well as in
advocating for policies related to environmental preservation and sustainable development.

Conclusions
This study has been carried out to analyze land use land cover dynamics and its impact on agricultural land in
Fitche town central Ethiopia. The goal of the research is to identify changes in urban land use and land cover
based on Landsat satellite images by using geospatial tools to analyze the changes in these areas and their effects
on agricultural land. LULC changes throughout the years 2002, 2013, and 2022 were analyzed using Landsat
imagery to anticipate the LULC map in the research area. The land cover maps in this study were divided into
four main land use land cover classifications (built-up area, farmland, open-space and shrubs). This was done to
forecast the future and gain a thorough understanding of the town's changes in land cover and land use. A total
kappa index of 0.83 percent indicates that the model was verified using simulated and observed LULC 2022
using kappa parameters. This suggests that the model has strong predictive power for land use and land cover
changes in 2030, 2040, and 2050. According to the study's findings, there have been notable changes to the
quantitative and spatial distribution of land cover and use. The findings indicated a consistent decline in the
town's agricultural area between 2002 and 2022, despite an increase in the built-up area over the same period.
The findings indicate that the percentage of land used for farming was 66.4 percent in 2002, 54.93 percent in
2013, and 36.42 percent in 2022. The annual rate change from 2002 to 2022 was -2.05 hectares per year.
Similarly, the percentage of built-up area was 10.43 percent in 2002, 16.92 percent in 2013, and 25.75 percent
in 2022. The annual rate change from 2002 to 2022 was 6.69 hectares per year.
Planning strategies to balance conflicting users, conservation, and development pressures depend heavily on the
prediction of LULC. Future LULC maps for the Fitche town are simulated and predicted using the ANN Cellular
Automation model. The four spatial variable factors such as DEM, slope, distance from an urban area, and
distance from a road, had a significant impact on the research area's LULC map prediction. The Kappa value of
0.83 shows a maximum level of accuracy between the observed and predicted 2022 LULC maps. The LULC
maps for 2013 and 2022, together with the same combinations of geographical variable components, were used
to anticipate the LULC maps for 2030, 2040, and 2050. The projected land use and cover (LULC) for the years
2030, 2040, and 2050 indicate a considerable growth in built-up areas, which were 29.8%, 32.3 percent, and
35.7 percent, respectively, and a decline in farmland areas, which were 26.7 percent, 29.7 percent, and 22
percent, respectively. The findings show that agricultural land and other land types would eventually be
transformed to built-up and open-space land as a result of human demands. The study's conclusions illustrated
the past LULCC and emphasized the changes that will occur over the following 28 years. The study effectively
illustrated the effectiveness of the cellular automata model by tracking spatiotemporal changes in land cover and
land use while predicting future patterns of land cover using remotely sensed data and QGIS techniques. This
study aids in identifying certain land-use changes and predicting the land uses that will be impacted by upcoming
changes. Additionally, ecological concerns and the decrease of biodiversity can be detected. The results support
the long-term development of natural resources by helping farmers and policymakers create the best land use
planning and management strategies. These kinds of data are essential for making well-informed decisions in
urban planning since they offer the prospective information needed to monitor growth and enhance
environmental sustainability.
 Authors’ contributions: All authors contributed to the study. The data collection, and analysis was conducted by
Dereje Ketema. The final manuscript was prepared by Addisu Bekele. These authors read and approved the final
manuscript.

Funding: No funding was received for conducting this study.

Data availability and Materials

All materials used for this article compilation are properly cited and acknowledged.

Declarations

Conflict of interest: The authors declare no conflict of interest with any individual, group or organization.

Ethics and approval: This is an original article that contains unpublished material. All experimental procedures
were conducted in accordance with the Guide for the Care.

Reference
Abbas, Z., & Jaber, H. S. (2020). Accuracy assessment of supervised classification methods for extraction land use
maps using remote sensing and GIS techniques. IOP Conference Series: Materials Science and Engineering,
745(1). https://doi.org/10.1088/1757-899X/745/1/012166
Agidew, A. meta A., & Singh, K. N. (2017). The implications of land use and land cover changes for rural
household food insecurity in the Northeastern highlands of Ethiopia: The case of the Teleyayen sub-
watershed. Agriculture and Food Security, 6(1), 1–14. https://doi.org/10.1186/s40066-017-0134-4.
Alam, N., Saha, S., Gupta, S., and Chakraborty, S. (2021). Prediction modelling of riverine landscape dynamics
in the context of floodplain management.
Aneesha Satya, B., Shashi, M., & Deva, P. (2020). Future land use land cover scenario simulation using open
source GIS for the city of Warangal, Telangana, India. Applied Geomatics,12(3), 281–290.
https://doi.org/10.1007/s12518-020-00298-4
Angel, S. (2023). Urban expansion: theory, evidence and practice. Buildings and Cities, 4(1), 124–138.
https://doi.org/10.5334/bc.348
Ayele,A. &Terakegn,K. (2020). The impact of urbanization expansion on agricultural land in Ethiopia: A review
Bank, W. (2021). Climate Risk country profile: Ethiopia. World Bank Group, 3(4), 1–32. www.worldbank.org
Berihun, M. L., Tsunekawa, A., Haregeweyn, N., Meshesha, D. T., Adgo, E., Tsubo, M., Masunaga, T., Fenta, A. A.,
Sultan, D., & Yibeltal, M. (2019). Exploring land use/land cover changes, drivers and their implications in
contrasting agro-ecological environments of Ethiopia. Land Use Policy, 87(March), 104052.
https://doi.org/10.1016/j.landusepol.2019.104052
Central Statistical Agency. (2016). Demographic and societal patterns of urban populations in Ethiopia. 5–8.
Chavez, P. S. (1989). Radiometric calibration of Landsat Thematic Mapper multispectral images. Photogrammetric
Engineering & Remote Sensing, 55(9), 1285–1294.
Chowdhury, S., Peddle, D. R., Wulder, M. A., Heckbert, S., Shipman, T. C., & Chao, D. K. (2021). Estimation
of land-use/land-cover changes associated with energy footprints and other disturbance agents in the
UpperPeaceRegionofAlbertaCanadafrom1985to2015usingLandsatdata.InternationalJournalofAppliedEar
thObservationandGeoinformation,94(November2019),102224.https://doi.org/10.1016/j.jag.2020.102224
Christman, Z., Rogan, J., Eastman, J. R., & Turner, B. L. (2016). Distinguishing land change from natural
variability and uncertainty in central Mexico with MODIS EVI, Conference Series: Materials Science
and Engineering, 745(1).
Cohen, B. (2004). Urban growth in developing countries: A review of current trends and a caution regarding existing
forecasts. World Development, 32(1), 23–51. https://doi.org/10.1016/j.worlddev.2003.04.008
Congalton, R. G. (2001). Accuracy assessment and validation of remotely sensed and other spatial information.
International Journal of Wildland Fire, 10(3–4), 321–328. https://doi.org/10.1071/wf01031
Di Gregorio, A., & Latham, J. (2003). Africover Land Cover Classification and Mapping Environmental
Challenges, 4(January), 100084. https://doi.org/10.1016/j.envc.2021.100084
Efa, T. and Gutema, I. (2017). „Prospects and Challenges of Urbanization on the Livelihood of Farming
Community Surrounding Finfinne‟, American Research Journal of Humanities and Social Sciences, 315:
2378-7031
F.Y, O., J.E.T, A., O, A., J. O, H., O, O., & J, A. (2017). Supervised Machine Learning Algorithms: Classification
and Comparison. International Journal of Computer Trends and Technology, 48(3), 128–138.
https://doi.org/10.14445/22312803/ijctt-v48p126
Fan, Y., Yang, R. G., & Wei, Y. M. (2007). A system dynamics based model for coal investment. Energy, 32(6), 898–
905. https://doi.org/10.1016/j.energy.2006.09.015
Foody, G. M. (2020). Explaining the unsuitability of the kappa coefficient in the assessment and comparison of the
accuracy of thematic maps obtained by image classification. Remote Sensing of Environment, 239(August
2019), 111630. https://doi.org/10.1016/j.rse.2019.111630
García-Álvarez, D., & Paegelow, M. (2022). Spatial Metrics to Validate Land Use Cover Maps. In Land Use Cover
Datasets and Validation Tools. https://doi.org/10.1007/978-3-030-90998-7_11
Gelan, E. (2021). GIS-based multi‐criteria analysis for sustainable urban green spaces planning in emerging
towns of Ethiopia: the case ofSululta town. Environmental Geospatial approach. Annals of GIS, 27(3),
299–314. https://doi.org/10.1080/19475683.2020.1870558
Gharaibeh, A., Shaamala, A., Obeidat, R., & Al-Kofahi, S. (2020). Improving land-use change modeling by
integratingANNwithCellularAutomataMarkovChainmodel.Heliyon,6(9),e05092.https://doi.org/10.1016/j
.heliyon.2020.e05092
Girma, R., Fürst, C., & Moges, A. (2022). Land use land cover change modeling by integrating artificial neural
network with cellular Automata-Markov chain model in Gidabo river basin, main Ethiopian rift.
Environmental Challenges, 6(August 2021), 100419. https://doi.org/10.1016/j.envc.2021.100419
Hakim, A. M. Y., Baja, S., Rampisela, D. A., & Arif, S. (2019). Spatial dynamic prediction of landuse / landcover
change (case study: Tamalanrea sub-district, makassar city). IOP Conference Series: Earth and
Environmental Science, 280(1). https://doi.org/10.1088/1755-1315/280/1/012023
Hamad, R., Balzter, H., & Kolo, K. (2018). Predicting land use/land cover changes using a CA-Markov model
undertwodifferentscenarios.Sustainability(Switzerland),10(10),1–23. https://doi.org/10.3390/su10103421
Haregeweyn, N., Fikadu, G., Tsunekawa, A., Tsubo, M., & Meshesha, D. T. (2012). The dynamics of urban expansion
and its impacts on land use/land cover change and small-scale farmers living near the urban fringe: A case
study of Bahir Dar, Ethiopia. Landscape and Urban Planning, 106(2), 149–157.
 https://doi.org/10.1016/j.landurbplan.2012.02.016
Hashidu, B. R., Abbas, A. M., & Kamaludeen, A. M. (2019). Urban Growth Pattern and Agricultural Land Use
Dynamics in Gombe City , Nigeria. 2(2), 43–49.
Hassan, Z., Shabbir, R., Ahmad, S. S., Malik, A. H., Aziz, N., Butt, A., & Erum, S. (2016). Dynamics of land
use and land cover change (LULCC) using geospatial techniques: a case study of Islamabad Pakistan.
Springer Plus, 5(1). https://doi.org/10.1186/s40064-016-2414-z.
Ibido, M. N. (2020). Pollution Control 2020- Urban expansion and its impact on peri urban household : The case
of Gelan and Dukem Towns , Ethiopia- Mesfin Nigussie Ibido- Nanjing Agricultural University. 4(2), 3–
4. India. International Journal of Emerging Technology and Advanced Engineering, 3(1), 1–5.
http://www.krishisanskriti.org/Publication.html
Kafy, A. A., Naim, M. N. H., Subramanyam, G., Faisal, A. Al, Ahmed, N. U., Rakib, A. Al, Kona, M. A., &
Sattar, G. S. (2021). Cellular Automata approach in dynamic modelling of land cover changes using Rapid
Eye images in Dhaka, Bangladesh.
Kamaraj, M., & Rangarajan, S. (2022). Predicting the future land use and land cover changes for Bhavani basin,
Tamil Nadu, India, using QGIS MOLUSCE plugin. Environmental Science and Pollution Research.
https://doi.org/10.1007/s11356-02117904-6
Kaul, H. a, & Sopan, I. (2012). Land Use Land Cover Classification and Change Detection Using High Resolution
Temporal Satellite Data. Journal of Environment, 01(04), 146–152.
Landis, J. R., and Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics,
33, 159-174.
Leulsegged, K., Gete, Z., Dawit, A., Fitsum, H., & Andreas, H. (2012). Impact of urbanization of addis abeba
city on peri-urban environment and livelihoods. The Tenth Conference on Ethiopian Economy, 1–30.
Li, M., Zang, S., Zhang, B., Li, S., & Wu, C. (2014). A review of remote sensing image classification techniques:
The role of Spatio-contextual information. European Journal of Remote Sensing, 47(1), 389–411.
https://doi.org/10.5721/EuJRS20144723
Lu, D., & Weng, Q. (2007). A survey of image classification methods and techniques for improving classification
performance. International Journal of Remote Sensing, Monitoring and Assessment,
Mariye, M., Maryo, M., & Li, J. (2022). The Study of Land Use and Land Cover (LULC) Dynamics and the
Perception of Local People in Aykoleba, Northern Ethiopia. In Journal of the Indian Society of Remote Sensing
(Vol. 50, Issue 5, pp. 775–789). https://doi.org/10.1007/s12524-021-01462-y
Misra, A. K., Masoodi, M., Poyil, R. P., & Tewari, N. K. (2018). Water demand and waste management with respect
to projected urban growth of Gurugram city in Haryana. Beni-Suef University Journal of Basic and Applied
Sciences,7(3),336.
Moller-Jensen, L. 1997. Classification of urban land cover based on expert systems, objects models and texture.
Comp. Environ. Urban Syst. 21(3/4): 291–302.343.https://doi.org/10.1016/j.bjbas.2018.03.003-190(11),675.
Mzava, P., Nobert, J., & Valimba, P. (2019). Land Cover Change Detection in the Urban Catchments of Dar es
Salaam, Tanzania using Remote Sensing and GIS Techniques. Tanzania Journal of Science, 45(3), 315–329.
www.ajol.info/index.php/tjs/
Patel, S., Verma, P., & Singh, G. (2019). Agricultural growth and land use land cover change in peri-urban India.
Environmental Monitoring and Assessment, 191. https://doi.org/10.1007/s10661-019-7736-1
Perović, V., Jakšić, D., Jaramaz, D., Koković, N., Čakmak, D., Mitrović, M., & Pavlović, P. (2018). Spatio-temporal
analysis of land use/land cover change and its effects on soil erosion (Case study in the Oplenac wine-producing
area, Serbia). Environmental Project. Encyclopedia of Life Support Systems (EOLSS), I, 236–254.
 Qrenawi, L. (2019). Faculty of Graduate Studies PhD Program in Water Technology Remote Sensing and GIS Image
Classification Prepared by : Luay I . Qrenawi , PhD Candidate. March.
Saputra, M. H., & Lee, H. S. (2019). Prediction of land use and land cover changes for North Sumatra, Indonesia,
using an artificial-neural-network-based cellular automaton. Sustainability (Switzerland), 11(11), 1–16.
https://doi.org/10.3390/su11113024
Seyam, M. M. H., Haque, M. R., & Rahman, M. M. (2023). Identifying the land use land cover (LULC) changes
using remote sensing
andGISapproach:AcasestudyatBhalukainMymensingh,Bangladesh.CaseStudiesinChemicalandEnvironmental
Engineering,7(December2022),100293.https://doi.org/10.1016/j.cscee.2022.100293
Singh, P., Gupta, A., & Singh, M. (2014). Hydrological inferences from watershed analysis for water resource
management using remote sensing and GIS techniques. Egyptian Journal of Remote Sensing and Space
Science, 17(2), 111–121. https://doi.org/10.1016/j.ejrs.2014.09.003
Stillwell, J. (2021). Modeling land-use change : theories and methods MODELLING LAND-USE CHANGE. March.
Sudhira, H. S., Ramachandra, T. V., & Jagadish, K. S. (2004). Urban sprawl: Metrics, dynamics and modelling using
GIS. International Journal of Applied Earth Observation and Geoinformation, 5(1), 29–39.
https://doi.org/10.1016/j.jag.2003.08.00
Sunil Sankhala, B. K. S. (2014). Evaluation of Urban Sprawl and Land use Land cover Change using Remote Sensing
and GIS Techniques: A Case Study of Jaipur City ,Systems Research ,10(1.)
Temesgen, H., Wu, W., Legesse, A., & Yirsaw, E. (2021). Modeling and prediction of effects of land use change in
an agroforestry dominated southeastern Rift-Valley escarpment of Ethiopia. Remote Sensing Applications:
Society and Environment, 21(October 2020), 100469. https://doi.org/10.1016/j.rsase.2021.100469
Terfa, B. K., Chen, N., Liu, D., Zhang, X., & Niyogi, D. (2019). Urban expansion in Ethiopia from 1987 to 2017:
Characteristics, spatial patterns, and driving forces. Sustainability (Switzerland), 11(10),1–21
Weith, T., Barkmann, T., Gaasch, N., Rogga, S., Strauß, C., & Zscheischler, J. (2021). Sustainable Land Management
in a European Context: A Co-Design Approach. In Human-Environment Interactions.
Yousafzai, S.,Saeed, R., Rahman, G.et.al., (2021),Spatiotemporal assessment of land use dynamic and urbanization
linking with environmental aspects and DPSIR frame work approach

View publication stats