Multi-Decade Land Cover/Land Use Dynamics and Future Predictions for Zambia: 2000 - 2030

Accurate and up-to-date information on land use/land cover change (LULCC) is important in land use planning and natural resource management; however, in sub-Saharan Africa, detailed information on LULCC is still lacking. Therefore, this study assessed the dynamics of LULC change (2000–2020) and future projections (2020–2030) for Zambia. The 2000 and 2010 LULC maps were used to simulate the 2020 LULC scenario using Arti�cial Neural Network (Multi-layer Perception) algorithms in Modules for Land Use Change Evaluation (MOLUSCE) plugin in QGIS 2.18.14. The 2010 and 2020 maps were used to predict the 2030 LULC classes. The reference 2020 and predicted 2020 LULC maps were used to validate the model. The validation between the predicted and observed 2020 LULC map, Kappa (loc) was 0.9869. The ANN-MLP simulated the 2020 LULC patterns successfully as indicated by the high accuracy level of more than 95%. LULC classes were predicted for 2030 using the 2010–2020 calibration period. The expected LULC types for 2030 revealed that built-up area will increase by 447.20 km 2 (71.44%), while 327.80 km 2 (0.73%) of cropland will be lost relative to 2020 LULC map. Dense forest (0.19%), grassland (0.85%) and bare land (1.37%) will reduce from 2020–2030. However, seasonally �ooded, sparse forest, shrub land, wetland and water body will increase marginally. The largest LULC change is from forest into other LULC types. The insights from this study show that ANN-MLP can be used to predict LULCC, and that the generated information can be employed in land use planning at a national scale.


Introduction
Land use/land cover change (LULCC) is a signi cance cause of global environmental change (Song et al. 2018; Guidigan et al. 2019).Land cover is described as the physical and environmental characteristics of the land surface area (Turner 1994), while land use is the human modi cation of the surface of the earth (Batunacun et al. 2018).Information on LULCC is key to environmentalists and planners due to its impact on the global environment (Islam et al. 2018).The generation of LULCC maps is important in land use planning and natural resources management for the bene t of human needs.The main drivers of regional land cover changes (LCC) are urbanization, deforestation, expansion of agricultural land and alteration of grassland areas (Lambin et al. 2001;Bowler et al. 2020;Hossain et al. 2023).In developing countries, urbanization and agricultural expansion have led to rapid changes in LULC (Ibrahim and Ludin 2014;Hossain et al. 2023).The major drivers of deforestation and forest degradation in developing countries like Zambia include economic development, population growth and lack of good government policies due to limited per capita land and poor adaptability ( Geospatial technology, Remote sensing and Geographical Information System (GIS), is increasingly used to monitor the Earth's surface at low costs and within the shortest period of time (Mas and Flores 2008;Mas et al. 2017).The integration of satellite remotely sensed data and GIS is recognized as an effective approach for monitoring and managing natural resources and land use planning (Guidigan et al. 2019).In addition, the advancement in computing technology couples with free access policies on datasets have made it possible to monitor changes at large spatial scales (WOODCOCK et al. 2008;Turner et al. 2015).The study of LULCC detection is important to urban planners, policy makers, natural resource managers, agriculture experts and environmentalist (Babalola and Shahi et al. 2020).The Markov-CA approach has been applied in generating dynamics of LULCC modeling at watershed level in Indonesia (Yulianto et al. 2016).Further, Markov-CA approach has been used to predict future changes in LULC types (Allahyari and Salehi 2020).The Cellular Automata (CA) model has been used in simulating the spatial LULCC by estimating the state of a pixel according to its initial state, surrounding neighbourhood effects and transition rules (Eastman 2016 Arti cial neural network (ANN) is a popular tool used in the analysis of satellite remotely sensed data (Mas and Flores 2008) mainly due to the fact that it characterizes a comparatively new approach to developing predictive models (Blackard and Dean 1999).ANN-MLP has been applied in spatio-temporal modeling of soil erosion studies (Cherif et al. 2023).In the medical eld, this approach has been used to simulate biological nervous systems (Gopal 2017 Predicting LULCC is signi cant for a number of developmental issues including urban expansion, deforestation and forest degradation (Li and Yeh 2002).Change detection and predicting future LULC categories can provided patterns of future land development based on current and historical changes.The historical changes can be used to identify future problems and uncertainties associated with LULCC.The LULC prediction maps provides vital information concerning types, direction, location and the magnitude of change.The predicted LULC maps can be used to inform policy, assess the developmental impacts and contribute to land use plans.Researchers such as Guidigan et al. (2019) have applied ESA CCI-LC products (2001,2008,2013) in Benin.This researcher used QGIS MOLUSCE plugin to validate the model and predict LULC scenarios for 2025 and 2037 and the results indicated an increase in cropland, forest and a reduction in Savannah land.The objectives of this study were to assess the dynamics of LULCC and to predict 2030 future LULC scenario in Zambia.The insights from this study can be used in sustainable natural resources management and land use planning at a national scale.

Description of the study site
Zambia is a land linked country located in Southern Africa and shares its borders with eight neighboring countries.Tanzania and the Democratic Republic of Congo (DRC) are in the north; Angola in the west; Namibia, Botswana and Zimbabwe in the south, and Mozambique and Malawi in the east (Fig. 1).It lies between latitudes 7° and 19° South of the Equator and longitudes 21° and 35° East of the Greenwich Meridian.Zambia has a total land surface area of 752,616 km² and its altitude range between 1,000 and 1,600 m above sea level (Chigunta and Matshalaga 2010).Zambia had an estimated population of 10415944, 13605984, and 18383955 according to World Population Prospects (WPP) of the United Nations, Department of Economic and Social Affairs in 2000, 2010 and 2020, respectively (https://www.worldometers.info/world-population/zambia-population/).By 2030, the population is projected to be 24,325,505 (UNDESA 2019).

Agro-ecological regions and climate
Zambia is divided into three agro-ecological regions (AERs; I, IIA, IIB and III) based on the amount of rainfall, soils properties and other climatic characteristics (Suman 2007;Mubanga et al. 2020;Bailey et al. 2021) as shown in Fig. 1.AER I, AER II, and AER III receive approximately < 800, 800-1000 and 1000-1500 mm of precipitation per year, respectively.The average annual temperature and rainfall in the AERs varies mostly by elevation (USAID 2011).The country is characterized by tropical climate with three distinguishable seasons: hot and dry season from mid-August to November, warm and wet (mid-November to April) and cool and dry (May to mid-August) (NAPA 2007).The cold temperature ranges from 3.6˚C to 12.0˚C with an average of 8.1˚C, mean hot temperature is 31.8˚Cranging from 27.7˚C to 36.5˚C (Kasali 2008).The hot summer months are very dry, receiving almost no rainfall between June and August.
Rainfall is extremely variable from year to year with an annual mean rainfall between 600 mm in the southern and 1400 mm in the northern part of Zambia (GIZ 2014).The annual rainfall is strongly in uenced by the shifting of the Paci c Ocean's El Nino Southern Oscillation (ENSO), the Inter-Tropical Convergence Zone (ITCZ) and the Congo Air Boundary and these produces stable rainfall patterns in the northern part of Zambia (MTENR et al. 2007; MTENR 2010; USAID 2011).The oscillating of the ITCZ between the northern and southern tropics during the course of the rain season leads to downward gradient of rainfall distribution from the north to the south of the country (Libanda et al. 2016).Multi-decadal trends in these phenomena contribute to annual variations in rainfall patterns and temperatures (Fig. 1).

Topography
Zambia's topographic features are represented by a series of gently undulating and at plateau with hills and low ranges.Broad shallow depressions can often be found in the plateau forming swamps and ats.The western part of the country is covered with loose sediment delivered by the Zambezi River which forms a wide at plain.The plateau is abruptly broken by steep linear escarpments running in a North-east-South-west direction along the Luangwa River and Zambezi River in the south-western peripheral area of Zambia.The plateau has an average elevation of 1300 m above sea level varying from a maximum of 2164 m in the east to a minimum of 325 m at the Zambezi River.Most of the country lies between 900 m and 1500 m and the main cities are mainly situated on the gentle undulating plateau (Chigunta and Matshalaga 2010).
Zambia has a diverse topography that includes high plateaus, mountains, and valleys.The topographic features are represented by a series of gently undulating and at plateau with hills and low ranges.The country's landscape can be divided into three main regions: The Zambezi River Valley.This region is located in the southern part of the country and is characterized by broad valleys and lowlands.The Zambezi River, which forms Zambia's southern border with Zimbabwe, ows through this region and forms the famous Victoria Falls.Broad shallow depressions can often be found in the plateau forming swamps and ats.The western part of the country is covered with loose sediment delivered by the Zambezi River which forms a wide at plain.The Central Plateau.This region occupies the central part of the country and is dominated by a high plateau that rises between 1,000 and 1,600 meters above sea level.The plateau is broken up by numerous hills and valleys and is dotted with large lakes and rivers.The plateau is abruptly broken by steep linear escarpments running in a North-east-South-west direction along the Luangwa River and Zambezi River in the south-western peripheral area of Zambia.The plateau has an average elevation of 1300 m above sea level varying from a maxim of 2164 m in the east to a minimum of 325 m at the Zambezi River.
The majority of the country lies between 900 m and 1500 m and the main cities are mainly situated on the gentle undulating plateau (Chigunta and Matshalaga 2010).
The Eastern Highlands.This region is located in the northeast part of the country and is characterized by high mountains and deep valleys.The highest peak in Zambia, Mount Ma nga, is located in this region and rises to an elevation of 2,339 meters above sea level.

Resource dependency
Zambia is well endowed with resources for reducing poverty through realising its agricultural potential.A wide range of crops, livestock and sh can be produced because of diversity potential in the AERs.The country's land area is approximately 752,000 square kilometres, of which, 12% is suitable for arable use.However, only about 16% of the arable land is presently cultivated.Additionally, about 16 million hectares are t for rangeland grazing (Chigunta and Matshalaga 2010).
Zambia's economy is heavily dependent on its natural resources, particularly copper, which accounts for a signi cant portion of the country's export earnings.The country is also rich in mineral resources, including copper, cobalt, and emeralds, which are found in the Central Plateau region.In fact, copper mining has been the backbone of Zambia's economy for decades, and the country is one of the world's top producers of the mineral.However, Zambia's resource dependency has also made its economy vulnerable to uctuations in commodity prices, as well as to shifts in global demand.In recent years, the country has experienced a decline in copper prices, which has negatively impacted its economy.In addition, the COVID-19 pandemic has caused disruptions in global supply chains, which has further affected Zambia's export earnings.
In an effort to diversify its economy and reduce its dependence on copper, Zambia has sought to develop other sectors, such as agriculture, tourism, and manufacturing.The government has also implemented policy reforms aimed at improving the business environment and attracting foreign investment.While Zambia's resource dependency has provided a source of wealth and employment, the country is also aware of the need to diversify its economy and reduce its vulnerability to external shocks.
Zambia also has abundant surface and underground water resources.It has numerous rivers, lakes and dams.The total country's ground water is estimated at 1,740,380 million cubic metres, while the ground water recharge is estimated at 160,080 million cubic meters per annum.While the irrigable land is estimated at 423,000 hectares, less than 40,000 hectares or nine percent is actually irrigated.This is concentrated mostly on commercial farms for the production of sugar, wheat and plantation crops.

Data source
The European Space Agency (ESA) Climate Change Initiatives (CCI) land cover maps were acquired from Copernicus Climate Data Store  (10) local Classi cations (Table 1) using the r.class module in QGIS 2.18.14.LULCC detection was undertaken on the reclassi ed LULC maps for 2000, 2010 and 2020.The spatial variables maps included in the analysis were elevation, slope and aspect maps.The aspect and slope maps were generated from the Digital Elevation model (DEM) using QGIS.The nal step was to align all LULC maps (2000, 2010, and 2020), elevation and aspect maps.In predicting future changes, we focused on using topographic factors, without including proximity factors as we wanted to understand the in uence of topographic factors alone.
The LULCC from 2000-2010, 2010-2020 and 2000-2020 (Fig. 2, Table 4) were analysed using R Programming/RStudio software.The land cover change transition matrix and percentage change in land cover were developed using R/RStudio.The pixel data were resampled from 300 m spatial resolution to a square kilometre.

Description of Multi-layer Perceptron (MLP) model
The MLP is the most popular neural network architecture (Mutlu et al. 2008).ANN-MLP is one of the approaches used with presence/absence data (Eastman 2016).It is a network of simple neurons called perceptrons.The perceptron computes a single output from multiple real-valued inputs by forming a linear combination according to its input weights and then possibly putting the output through some nonlinear activation function (Oyebode and Stretch 2019).
The MLP is usually trained using the error backpropagation algorithm (Cao et al. 2019).The objective of training of an ANN-MLP model is to obtain optimal model parameters that allow a model to yield the best representation of the input-output relationship of a particular system (Oyebode and Stretch 2019).The ANN-MLP works by iteratively changing a network's interconnecting weights thus minimizing the overall error between observed and simulated LULC map (Sudheer 2000).

Land use/land cover Change Prediction and Validation
Area Changes (Change Analysis) The where and P ij is the i,j-th cell of contingency table, P iT is the sum of all cells in i-th row, P Tj is the sum of all in j-th column, c is the count of raster categories.

Validation statistics
The transition potential modelling between 2000 and 2010 LULC map using a random sample of 1000 yielded a current validation kappa of 0.93984.The validation statistics between the predicted 2020 and observed 2020 LULC map indicated that there was a 99.41% of correctness.The kappa (overall), kappa (histo) and kappa (loc) were 0.99122, 0.99754 and 0.9366, respectively.The transition potential modelling between 2010 and 2020 LULC map using a random sample of 1000 yielded a current validation kappa of 0.93128 (Fig. 3).

Future land use land cover change
The modelled land cover change matrices between 2020 and 2030 are shown in The ndings of this study need to be interpreted by taking into consideration the research limitations.Firstly, we considered analysis period of 10 years, which is a common practise among researchers; however, some changes might have occurred within this period that went unnoticed.For example, cropland used for less than 5 years and left as fallows and settlements which were abandoned.Secondly, we focused more on topographic factors in predicting our future LULC status without considering other factors such as proximity and socioeconomics.This was done deliberately in order to understand the in uence of topographic factors on LULCC in isolation.This approach was ideal for those areas which have limited access and have less socioeconomic activities.
Finally, although we achieve high accuracies (> 95%) in our validation, it is still important to note that there is still room for error on our results; however, this is expected to be minimal.

Conclusions
This study aimed to understand the multi-decade LULC changes, as well as predicting the future status of land cover for Zambia at a national scale.The study employed ESA CCI dataset for the period between 2000 and 2020 and ANN-MLP as the processing tool.Furthermore, the study made predictions for the year 2030 based on a model of 2010 and 2010 maps.The study revealed that cropland (0.73%), dense forest (0.19%), grassland (0.85%) and bare land (1.37) will reduce from 2020 to 2030.However, built-up will increase by 447.20 km 2 (71.44%) from 2020 to 2030.Other LULC classes will increase marginally namely, seasonally ooded grassland, sparse forest, shrub land, wetland and water body.The growth in built-up and the use of ANN-MLP models in LULC prediction could be used for the successful land use planning in forest and natural resource management.The insights from this study are key in conservation, forest monitoring and general LULCC dynamics in order to balance the trade-off between human needs which are driven by population growth on one hand, and conservation needs on the other hand.Being one of the rst studies on predicting LULCC for Zambia, the nding from this study will be key in strategic planning for different developmental aspects that concern land at a national scale.Different government sectors such as the Department of Forestry, Agriculture, Lands and conservation will employ these results in their strategic planning and operations.
This study also demonstrates that open access tools and free access datasets can be employed to provide key information on LULC at large scale.The Diouf and Lambin 2001; Ibrahim and Ludin 2014; Fagan et al. 2020; Hossain et al. 2023).Owing to different factors driving the changes, LULCC occur at different rates and spatio-temporal scale (DeFries et al. 2010; Hansen et al. 2016).
Akinsanola 2016; Saputra and Lee 2019; Hossain et al. 2023).Different algorithms have been used to predict and simulate the changes in LULC (Rahaman et al. 2023; Shahfahad et al. 2023; Roushangar et al. 2023).These include Cellular Automata (CA) model namely; Traditional CA (TCA) model, Agents based Cellular Automata (ACA) Model and Neural Network coupled Agents-based Cellular Automata (NNACA) model (Devendran and Lakshmanan 2018; Saputra and Lee 2019), machine-learning (Saputra and Lee 2019; Elmes et al. 2020), cellular automata-Markov model (Yulianto et al. 2016; Islam et al. 2018) and Markov chain (MC) (Anand et al. 2018; Allahyari and Salehi 2020; Silva et al. 2020; ; Saputra and Lee 2019).Yang et al. (2019) argues that CA models are able to predict the spatial distribution of landscape patterns but fail to predict temporal changes (Islam et al. 2018; Yang et al. 2019).A combination of CA and ANNs have been used to model LULC dynamics (Eastman 2009; Saputra and Lee 2019; Yang et al. 2019).Other researchers (Liu et al. 2017; Saputra and Lee 2019) have highlighted that the CA model can generate patterns and effectively represent nonlinear spatial LULCC processes.The CA model has been applied in urban and forest studies (Saputra and Lee 2019), and have the capacity to simulate the spatial and temporal complexity of urban areas.Further, forests have also been simulated under the in uence of anthropogenic activities and natural hazards (Saputra and Lee 2019).The arti cial neural network-based Markov chain model has been applied in simulating LULCCs (Debnath et al. 2023).
; Talukdar et al. 2020).The ANN models are being used in the urban spatial growth (Kumar et al. 2022), LULC classi cation, LULCC detection and LULC prediction since the 1990s (Gopal 2017).ANN have been integrated with GIS and CA models for predicting LULC (Devendran and Lakshmanan 2018; Buğday and Erkan Buğday 2019; Saputra and Lee 2019).The Arti cial Neural Network based Cellular Automaton (ANN-CA) simulate multiple LULCC and complex land use systems (Li and Yeh 2002; Saputra and Lee 2019).The ANN-CA can be used to determine LULCC by incorporating spatial variables such as slope, aspect and elevation which may in uence the changes (Li and Yeh 2002; Saputra and Lee 2019).An important class of ANN which has been applied and used in many elds is the multilayer perceptron (MLP) (Mas and Flores 2008; Oyebode and Stretch 2019).
(https://cds.climate.copernicus.eu/cdsapp#!/dataset/satellite-land-cover?tab=form) for the year 2000, 2010 and 2020.The land cover maps (LCMs) are provided for versions 2.0.7 (2000 and 2010) and 2.1.1 (2020).The v2.0.7cds provides the LC maps for the years 1992-2015 while v2.1.1 for the years 2016-2019.The ESA CCI dataset provides global wall-to-wall maps that describe the land surface into 22 classes, which have been de ned using the United Nations Food and Agriculture Organization's (UN FAO) Land Cover Classi cation System (LCCS).In addition to the land cover (LC) maps, four quality ags are produced to document the reliability of the classi cation and change detection.The ESA CCI land cover maps are consistent with the series of global annual LC maps from the 1990s to 2015 produced by the ESA CCI.To produce the ESA CCI dataset, the entire Medium Resolution Imaging Spectrometer (MERIS) Full and Reduced Resolution archive from 2003 to 2012 was rst classi ed into a unique 10-year baseline LC map.This is then back-and up-dated using change detected from (i) Advanced Very-High-Resolution Radiometer (AVHRR) time series from 1992 to 1999, (ii) SPOT-Vegetation (SPOT-VGT) time series from 1998 to 2012 and (iii) PROBA-Vegetation (PROBA-V) and Sentinel-3 OLCI (S3 OLCI) time series from 2013 (ESA 2017a; Guidigan et al. 2019).Beyond the climate-modelling communities, this dataset's long-term consistency, yearly updates, and high thematic detail based on a global scale have made it attractive for a multitude of applications such as land accounting, forest monitoring and deserti cation, in addition to scienti c research (ESA 2015).The 2000 and 2010 land use land cover maps were used to simulate the 2020 LULC map using the Modules for Land Use Change Evaluation (MOLUSCE) plugin in QGIS 2.18.14 (https://qgis.org/downloads/).The MOLUSCE is a QGIS plugin for Land Use Change Evaluation.It provides a set of algorithms (Arti cial Neural Network-Multi-Layer Perceptron [ANN-MLP], Logistic Regression [LR], Weights of Evidence [WoE], Multi Criteria Evaluation [MCE]) for land use change simulations including kappa statistics for validation (Hakim et al. 2019; Alam et al. 2021).A MLP is an important class of ANN and it has been widely applied in similar studies (Mas and Flores 2008).MOLUSCE plugin was developed by Asia Air Survey (http://www.asiaairsurvey.com)and NextGIS (http://nextgis.com).Here, the predicted 2020 LULC map was compared with the reference 2020 LULC map to validate the model.Based on the created model, the prediction of future LULC scenario for 2030 was simulated.Data processing The ESA Global land cover datasets for 2000 (C3S-LC-L4-LCCS-Map-300m-P1Y-2000-v2.0.7cds.tif),2010 (C3S-LC-L4-LCCS-Map-300m-P1Y-2000-v2.0.7cds.tif)and 2020 (C3S-LC-L4-LCCS-Map-300m-P1Y-2020-v2.1.1.tif)were extracted (clip raster by mask layer) using country boundary shape les (gadm36zmb0.shp)after being loaded in QGIS.The ESA CCI land cover maps have a spatial resolution of 300 metres (ESA 2017b; European Space Agency 2017).The 31 ESA CCI Classi cation (European Space Agency 2017) were reclassi ed into ten LULCC detection was carried out using land cover maps for 2000, 2010 and 2020.LULCC detection aided in observing the evolution of land class categories during the last twenty years.Area changes were generated from 2000-2010 and 2010-2020.The MOLUSCE plugin allows the generation of class statistics, transition matrix and LULC change map(Guidigan et al. 2019).The class statistics describes changes that have taken place between LULC map at time 0 and time 1 periods.The transition matrix depicts the number of pixels that have changed from one LULC type to another.Transition Potential ModellingThe ANN-MLP was selected to perform the transition potential modelling as a pre-stage(Guidigan et  al. 2019; Mangel and Berhe 2021) for prediction 2020 LULC map using the Cellular Automata Simulation based on initial LULCC in 2000 and 2010.The MLP structure consists of three-layer (input, hidden and output layer) (Eastman 2016; Oyebode and Stretch 2019; Mangel and Berhe 2021).The ANN (MLP) learning process between 2000 and 2010 LULC maps used 1000 iterations, 1-pixel, 0.100 learning rate, 10 hidden layers and 0.050 momentum.The overall accuracy, minimum validation overall error and current validation Kappa were − 0.00095, 0.00385 and 0.93984, respectively between the 2000 and 2010 LULC maps.Cellular Automata Simulation The class statistics, transition matrix and LULC change map were used to predict 2020 LULC map using Cellular Automata (CA) Simulation in MOLUSCE plugin.The CA approach is based on Monte Carlo algorithm and every change can be viewed as a transition of land use categories.The change map is a integer one-band raster that stores information about transitions (Asia Air Survey and Next GIS 2017).Furthermore, category values of change map are mapped one-to-one to transition classes.During the second stage, class statistics, transition matrix and LULC change map between 2010 and 2020 LULC map were used to predict the 2030 LULC map using the Cellular Automata Simulation and plausible changes in LULCC determined.Validation Ten validation iterations were used to compare the predicted 2020 and observed 2020 LULC map using kappa histogram (khisto), kappa overall (kovr) and kappa location (kloc) (Asia Air Survey and Next GIS 2017).Based on previous studies which were conducted in the same geographic area (Phiri et al. 2019), we used a sample of 1000 points for validation.The MOLUSCE plugin calculates three types of kappa statistic as shown in Eqs. 1, 2 and 3 (Pontius Jr.
-Temporal changes in LULC from 2000 to 2020 Observed land cover change Matrix between 2000-2010, 2010-2020, and 2000-2020 are shown in Figures

Figure 1 Map
Figure 1 Map showing the relative position of Zambia with its neighboring countries namely, Tanzania and the Democratic Republic of Congo (DRC), Angola, Namibia, Botswana, Zimbabwe, Mozambique and Malawi.

Table
(Kipkulei et al. 20221;Hossainet al. 2023) and 1.22% under Shrub land (5808.50km2),Grassland(114.70 km 2 ) and Sparse forest (788.60 km 2 ), respectively.During the period 2000 to 2010 and 2010 to 2020, Built-up exhibits a decreasing trend by 59.79% and 71.44%, respectively.Cropland indicated a decrease trend of 0.70%.However, there is an increase of Cropland LULC type from 2000-2010 and 2000-2020 of 1.43% and 0.72%, respectively.The results shows that Built-up area increased while Sparse forest, shrub land and Grassland decreased during the period 2000 to 2020.Moreover, Bare land also increased from 2000 to 2020 by 6.39% (2.60 km 2 ).It has been noted that the increase in bare land is as a result of the decline in forest area.This trend is of great concern to environmentalists and policy makers even in other countries beyond Zambia(Oluwajuwon et al. 2021;Hossain et al. 2023).Figure2shows the trend of LULC between 2000 and 2020.It is clear from the map that land cover classes such as Built-up and Bare land have increased, while dense forest have decline over this period.Moyo et al. 2022; ZamStat 2022).Moreover, as of 2010, the population had been growing at approximately 4.2% year -1 , with 43% of the population living in urban areas (CSO 2012).A high rate of population growth exerts pressure on the land and exacerbates shortfalls in cropland (CSO 2011; Population Reference Bureau 2019).Population growth and expansion in built-up areas is putting more pressure on terrestrial landscapes to meet socioeconomic needs such as food and shelter(Bowler et al. 2020).Evaluating LULCC assessment has revealed that Zambia has been subjected to four diverse rates of land degradation during the periods of 2000, 2010 and 2020.This is as a result of upsurge in built-up from 2000 to 2020 of about 681.60 km 2 (174.01%).Understanding LULC dynamic is vital to sustainable development in Zambia where deforestation is a common problem.The prominent cause of forest loss is the expansion of development projects, including settlements and cropland, into forested area(Hossain et al. 2023).In east African, cropland expansion has led to losses in forest, wetland and grassland LULC types(Kipkulei et al. 2022).Hossain et al. (2023) has reported similar results in Bangladesh where Built-up (settlement) expanded due to massive human migration into urban areas.The decline in the area covered by forests affects other sectors such as conservation as these areas are habit of different fauna species -some are facing different threats and might go into extinction.
1, Table 2 and Table 3.The percentage changes in the ten multi-temporal LULC classes are shown in Table 4.The LULC categories of cropland, seasonally ooded grassland, dense forest, wetland, built-up, bare land and water body increased from 2000 to 2020.Major increase of LULC types from 2000 to 2020 were Built-up (174.01%)followed by Bare land

Table 5 .
(Ackom et al. 2020)ateishi 2015)2030 LCMs indicates that cropland, dense forest, grassland and bare land categories will reduce by 327.80, 814.90, 68.50 and 0.60 km2.However, built-up will increase by 71.44% during the same period.Similar results have been reported byChisanga et al. (2022)who noted that built-up area will increase by 2030.Others studies revealed an increase in built-up LULC types (Buğday and Erkan Buğday 2019).In contrast, other researchers have predicted 2030 LULCs using the 2010-2020 calibration period (Fig.4).The largest changes were associated with conversion of forest lands into other land uses(Shahi et al. 2020).Alongside with this rapid urbanization of the human society, it is estimated that, by 2030, cities will physically expand by 1.2 million km 2 , which is almost same size as the Republic of South Africa(Seto et al. 2012;Tateishi 2015).The expansion of built-up area by 2030 may led to degradation of ecosystems in most developing countries(Ackom et al. 2020).Simulating LULCC provides insights, patterns and identi es human impacts like deforestation and forest degradation.The identi ed environmental problems can be incorporated in the processes of land use planning.Other researchers have noted that monitoring and LULC modeling is imperative in environmental planning, conservation of natural resources and in attaining sustainable development goals (SDGs) number 13 and 15 (Redowan et al. 2014; Rawat and Kumar 2015; Hossain et al. 2023).The ndings of this study contributes to Zambia's Vision 2030 for sustainable socio-economic development (GRZ 2006).The population of Zambia is projected to be 24 million by 2030 (UNDESA 2019).Further, the simulated LULC maps can be used to articulate long-term plans national development, and achieving the desirable socio-economic outcomes by 2030.The simulation of LULCC provides an understanding of future LULC scenarios (Saputra and Lee 2019).Simulation of LULCC is important for a variety of planning, monitoring, management and academic research.It provides historical LULCC, and how future LULC classes will change based on the current development processes into the future (Li and Yeh 2002).Elsewhere, other researchers (Ibrahim and Ludin 2014; Moulds et al. 2015; Al-Rubkhi 2017; Buğday and Erkan Buğday 2019; Guidigan et al. 2019; Hakim et al. 2019; Saputra and Lee 2019; Suprayogi and Subiyanto 2019) have used Open Source Software to analyse and predict future LULCC using tools such as QGIS (MOLUSCE plugin) and R Programming packages (lulcc R package).Suprayogi and Subiyanto (2019) used the lulcc package in R/RStudio to model the land use change of Banyumanik.The use of open-source software and open access dataset is a clear testament that monitoring of LULC can be done with less challenges and information can be generated with easy.