¹ Department of Urban and Regional Planning, Faculty of Engineering, Universitas Islam Bandung, Bandung, West Java, Indonesia
² Department of Geological Engineering, Faculty of Geological Engineering, Universitas Padjadjaran, Bandung, West Java, Indonesia
³ Department of Civil and Environmental Engineering, Faculty of Engineering, Universitas Indonesia, Depok, West Java, Indonesia
⁴ Department of Urban and Regional Planning, School of Architecture, Planning and Policy Development, Institut Teknologi Bandung, Bandung, West Java, Indonesia
⁵ Department of Urban and Regional Planning, Faculty of Engineering, Universitas Gadjah Mada, Yogyakarta, Special Region of Yogyakarta, Indonesia
⁶ Department of Electrical and Information Engineering, Faculty of Engineering, Universitas Gadjah Mada, Yogyakarta, Special Region of Yogyakarta, Indonesia
⁷ Geomatics Engineering, Department of Geodetic Engineering, Faculty of Engineering, Universitas Gadjah Mada, Yogyakarta, Special Region of Yogyakarta, Indonesia
⁸ Department of Civil and Environmental Engineering, Faculty of Engineering, Universitas Gadjah Mada, Yogyakarta, Special Region of Yogyakarta, Indonesia
⁹ Faculty of Civil and Environmental Engineering, Institut Teknologi Bandung, Bandung, West Java, Indonesia
¹⁰ Central South University, Changsha, Hunan, China

Randy Victor Wahab Wattimena
Roles: Conceptualization, Data Curation, Formal Analysis, Methodology, Project Administration, Writing – Original Draft Preparation, Writing – Review & Editing

Saraswati Saraswati
Roles: Conceptualization, Supervision, Validation, Writing – Review & Editing

Yulia Asyiawati
Roles: Conceptualization, Supervision, Validation, Writing – Review & Editing

Nabila Kalsum Tuanany
Roles: Project Administration, Writing – Original Draft Preparation, Writing – Review & Editing

Ramadhan Eka Prasetian
Roles: Data Curation, Writing – Review & Editing

Chatarina Yosefina Janggur
Roles: Methodology, Validation, Writing – Original Draft Preparation, Writing – Review & Editing

Ranti Aprilia
Roles: Methodology, Validation, Writing – Original Draft Preparation, Writing – Review & Editing

Luthfiyyah Dzakiyyah Wopa
Roles: Methodology, Validation, Writing – Original Draft Preparation, Writing – Review & Editing

Atris Greisela Pattiasina
Roles: Formal Analysis, Methodology, Writing – Original Draft Preparation, Writing – Review & Editing

Hervin Maitimu
Roles: Data Curation, Formal Analysis, Software

Mesak Ananias Miru
Roles: Conceptualization, Methodology, Software, Writing – Original Draft Preparation, Writing – Review & Editing

Zhuhril Ilmi
Roles: Validation, Writing – Review & Editing

Sintiche Jasso
Roles: Data Curation, Methodology, Software, Writing – Review & Editing

Andre Tamaela
Roles: Data Curation, Methodology, Software, Writing – Review & Editing

Habel Brilian Natalio Taka
Roles: Data Curation, Methodology, Software, Writing – Review & Editing

Ghazi Oktavidi Muslim
Roles: Writing – Review & Editing

Karangnunggal is one of the subdistricts located in Tasikmalaya Regency, with an area of 153.26 km². Geographically, Karangnunggal Subdistrict is located south of the capital of Tasikmalaya Regency.

Karangnunggal Subdistrict consists of 14 villages with rural classifications as described in Table 1, and the administrative map of the villages in Karangnunggal Subdistrict can be seen in Figure 1.

Karangnunggal Subdistrict has 5 elevation criteria: 0–69 m a.s.l., 70–154 m a.s.l., 155–226 m a.s.l., 227–285 m a.s.l., and 286–424 m a.s.l. In general, the area can be distinguished by elevation, namely the northern part being highland and the southern part being lowland, as shown in Table 2, while the distribution of topography in Karangnunggal Subdistrict can be seen in Figure 2.

Karangnunggal Subdistrict has 5 slope gradient classifications, namely 0–8%, 8–15%, 15–25%, 25–40%, and > 40%. The slope gradient in Karangnunggal Subdistrict is dominated by the 8–15% class, covering an area of 6,628.51 Ha. Meanwhile, the smallest slope gradient class is >40%, covering an area of 1,015.15 Ha. For further detail, see Table 3, and the distribution map of slope gradient in Karangnunggal Subdistrict can be seen in Figure 3.

Based on the soil classification system, Karangnunggal Subdistrict has three main soil types: Red-Yellow Podzolic, Brown Forest, and Alluvial. Red-Yellow Podzolic soil has low productivity due to advanced weathering and intensive leaching, resulting in low nutrient content and poor physical and chemical properties (Haryati 2011). This soil has an acidic pH, is poor in macronutrients (Ca, K, Mg), and dries easily due to low water-holding capacity, making it less suitable for annual crops. Brown Forest soil is a mineral soil that develops in humid to moderate climate areas, generally in forested regions (Tamil Nadu Agricultural University 2013) This soil is fairly potential for dryland agriculture, small-scale plantations, settlements, and public facility development due to its stable soil structure. Alluvial soil is formed from sediment deposition processes and has relatively high fertility, making it highly supportive of agricultural activities, particularly with easily applied irrigation systems (Butler 2024). More detailed information on the distribution and characteristics of each soil type can be seen in Table 4 and Figure 4.

Land use is the tangible manifestation of the influence of human activity on part of the earth’s physical surface. Land use in Karangnunggal Subdistrict is dominated by plantation land use, covering an area of 9,177.96 Ha, or approximately 60% of the entire area of Karangnunggal Subdistrict. Meanwhile, the smallest land use type is fish ponds, covering an area of 3.10 Ha, or approximately 0.02% of the entire area of Karangnunggal Subdistrict. For further detail, see the chart in Figure 5, and land use by village administration in Karangnunggal Subdistrict can be seen in Table 5.

Viewed by administrative area, plantation land use is dominated by Cibatu Village, covering an area of 1,077.85 Ha. Meanwhile, the smallest land use is fish ponds, covering an area of 3.10 Ha of the entire area of Karangnunggal Subdistrict. The land use map of Karangnunggal Subdistrict can be seen in Figure 6.

This analysis method aims to process the collected data in order to answer the research questions. In this study, the authors employed a spatial analysis approach with the assistance of GIS software. The main technique used is map overlay and scoring of each physical land characteristic. The entire scoring and weighting process refers to the standard set out in Regulation of the Minister of Public Works No. 20/PRT/M/2007 concerning Technical Guidelines for Physical and Environmental Aspect Analysis (Wattimena et al. 2026). The stages of analysis carried out include:

2.1.1. Land Capability Analysis

This analysis was carried out to obtain an overview of the level of land capability for development as a development area, serving as a reference for suitability directions for cultivation and protected areas. The output of this analysis includes a land capability classification map for development according to the area’s function, as well as the potential and physical constraints of land development (Puntsag 2014; Arfiansyah et al. 2024). Implementation steps:

Mapping the land capability value is the sum of value multiplied by weight, namely:

The determination of land capability classification is not based solely on value intervals, but also takes into account the lowest value = 1 among several land capability units, which serves as the determining value for whether the interval applies. Thus, if a particular area or zone has a fairly high value interval but has a low determining value, its land capability class may not be the same as another area with the same land capability value (Department of Natural Resources and Environment Tasmania 2025). The resulting land capability classification is based solely on physical conditions as they are and has not yet taken non-physical factors into account. The following table of land capability classification can be seen in Table 7.

2.1.2. Analisis Daya Dukung Lahan

Carrying capacity functions as a development planning instrument that maps the relationship between the population and the environment as well as land-use patterns, enabling planners to assess the suitability of interactions within an area. Information on carrying capacity is required by planners to gauge land capacity, as planning must be able to estimate community needs and align them with available land conditions (Mutaali 2012). Land carrying capacity refers to land utilization as well as population data in a systematic manner (Utama and Dionita 2015). Every human activity in meeting the needs of life requires space, so that land availability plays an important role in supporting various human activities. In addition, the amount of land available in a region also determines its capacity to support the population and influences a decent standard of living.

Land carrying capacity refers to the capacity of a region to support the life of living beings, particularly humans. Carrying capacity analysis is conducted to assess the extent to which an area is able to provide suitable settlement land for a given population (Pratamaningtyas Anggraini et al. 2023; Kodri Sitompul, Cut Nuraini, and Abdi Sugiarto 2025; Yuliani, Rahman, and Berliana 2025b). This analysis process requires data on the area of land available for settlement, the population to be accommodated, and the standard land requirement per individual. Land considered suitable for settlement can be identified through two approaches, namely the spatial planning approach and the land capability approach (Hasanuddin, Rahman, and Rasyidi 2022). This study uses the spatial planning approach, so that suitable settlement land is land located outside protected areas and free from disaster threats such as landslides, abrasion, seawater intrusion, and other geological hazards. Through this approach, the area is classified into three main categories, namely:

Based on the definitions of these three land carrying capacity zones, it can be concluded that the zone that can be developed for settlement while also accommodating population is the potential zone. However, the potential zone cannot be developed entirely for settlement; space must also be allocated for other uses, namely utility networks and public infrastructure.

Therefore, the development of settlements must take into account a land coverage ratio of 60% of the existing potential zone area, in accordance with the criteria set out in Regulation of the Minister of Public Works Number 20 of 2007 concerning Technical Guidelines for Physical and Environmental Analysis (Hasanuddin et al. 2022). The area of land that can be developed for settlement from the potential zone can be calculated using the following formula:

LPm=(LWPx60%)

Source: (Mutaali 2012).

LPm = Area of land that can be developed for settlement (ha).

LWP = Area of the Potential Zone.

60% = Land Coverage Ratio

2.1.3. Analysis of Potential Location Determination for Settlement Development Using Cellular Automata (CA)

Analysis of Potential Settlement Location Determination Using Cellular Automata (CA) is a dynamic spatial modeling approach used to predict and identify potential settlement zones in Karangnunggal Subdistrict based on the integration of land carrying capacity results both potential and limitation zones as one of the driving and constraint (Atlas n.d.; Marinescu 2017; Jiang et al. 2022).

Data processing in this study applies the Cellular Automata (CA) model as a cell-grid-based spatial modeling approach. The process begins with downloading land cover and road network data, which is processed in ArcGIS 10.8 to convert the format from polygon to raster; the raster data is then imported into QGIS 3.44 for processing land use change predictions through the MOLUSCE plugin. The determination of settlement development locations is based on spatial variables, namely driving factors, constraint factors, and land cover for 2015–2025. For further detail, see Table 8.

This analysis was carried out through several stages, consisting of:

Involves the preparation of spatial data such as the initial land cover map, driving variables, and constraint variables, to begin the Cellular Automata (CA) modeling process using software such as QGIS with the MOLUSCE plugin, or to integrate raster data. This process ensures that the model has an accurate data foundation for settlement potential analysis (Xing et al. 2020; Khan and Chen 2025; Yeasmin et al. 2025).

The Evaluating Correlation stage in the MOLUSCE plugin is a process for assessing the level of correlation between the spatial variables (driving factors) used in Cellular Automata modeling. At this stage, MOLUSCE calculates the linear relationship between rasters using Pearson’s Correlation method, with a value range of −1 to +1. Values close to 0 indicate a weak relationship (relatively independent variables), while values close to +1 or −1 indicate a strong relationship (variables tend to be mutually explanatory/redundant). This correlation evaluation is important to ensure that the driving variables used do not experience multicollinearity, as overly correlated variables can cause information redundancy and potentially affect the stability and interpretation of the transition potential model (NextGIS 2026).

The area changes analysis continues by identifying spatial patterns of change, such as clustering of settlement growth or fragmentation of agricultural land. Through overlay and cross-tabulation matrices, it is possible to determine which land class is most dominantly undergoing conversion, as well as the direction and intensity of change in the study area. This information is important for understanding spatial dynamics before entering the transition potential modeling stage (Mas et al. 2022; Jafarpour Ghalehteimouri et al. 2022; Noviani, Ahmad, and Marfu’ah 2024; Edosa and Nagasa 2024).

Transition Potential Modelling in the MOLUSCE plugin is the process of building a prediction model for land cover change transition potential based on predetermined spatial variables. In this study, transition potential is formed using the Artificial Neural Network (ANN) – Multi-layer Perceptron (MLP) method, which functions to learn the relationship between historical land cover change and the driving factors used (Muhammad et al. 2022; Dian Hudawan Santoso et al. 2025; Dadzie et al. 2026).

Cellular Automata Simulation is carried out by combining transition probabilities from the Markov Chain with local rules based on the cell’s neighborhood. Each cell in the grid changes its status based on the condition of the cell itself, its neighboring cells, and transition probability, with iteration over one or several time periods. Using software such as LanduseSim or the MOLUSCE plugin, the simulation produces a land use prediction map for the target year, taking into account growth scenarios and spatial constraints (Kang et al. 2019; Pinto et al. 2025; Saharan and Nehra 2025).

Validation in the MOLUSCE plugin is the process of evaluating the level of agreement between the simulated map produced and the reference map used for comparison. In this study, the Reference Map is the reference map selected to assess the simulation output, while the Simulated Map is the land cover change model output generated by the Cellular Automata (CA) model. This validation is carried out to ensure that the change pattern produced by the model is not only mechanically formed, but also has a high degree of agreement with the reference pattern considered representative. In addition, the Number of validation iterations = 5 parameter indicates that the evaluation process was carried out through several testing iterations, so that the assessment results are more stable and do not depend on a single calculation (Bramantio, Hizbaron, and Khakhim 2024; Tefera and Thatiparthi 2026).

The land capability classification for Karangnunggal Subdistrict was carried out through an overlay (intersect) of each Land Capability Unit (LCU), in which the final capability level value of each LCU was multiplied by its weight in stages to produce a cumulative total score map across all LCU. The multiplication of the final value by the weight for each LCU is called the score, using the formula Score = Final Value x Weight, which is then accumulated to obtain the overall land capability map. The detailed results of this cumulative score calculation are presented in Table 9 and the chart in Figure 7.

Land capability in Karangnunggal Subdistrict is classified into four land capability classes: class b (low), class c (moderate), class d (fairly high), and class e (high). This high land capability dominates the area of Karangnunggal Subdistrict, enabling the development of various agricultural activities and sustainable settlements. Meanwhile, the smallest land capability area is the low capability class, covering an area of 1,087.68 Ha, or only 7%. Land capability in Karangnunggal Subdistrict by administrative area can be seen in Table 10.

Viewed by administrative area, the smallest land capability area is found in Cikupa Village, covering 13.48 Ha. High-classification land capability is distributed across all villages in Karangnunggal Subdistrict, with Cibatuireng Village having the largest area at 800.84 Ha.

The dominance of the fairly high and high land capability classes indicates that the physical land conditions in Karangnunggal Subdistrict generally support settlement development, as reflected by the moderate-to-flat slope gradient that minimizes erosion risk. However, this result is a cumulative score from all nine LCU parameters and therefore does not automatically reflect the area’s safety from disaster. Given that Karangnunggal Subdistrict remains vulnerable to hydrometeorological hazards (floods, landslides, earthquakes), the land capability analysis map can be seen in Figure 8.

Karangnunggal Subdistrict has three land carrying capacity categories based on the physical aspect analysis of land capability, namely the Constraint Zone, the Limitation Zone, and the Potential Zone; of the three, only the Potential Zone is suitable for development as settlement land. The Potential Zone has the most suitable physical conditions for housing without extensive modification, while the Constraint Zone requires additional investment such as land filling and engineering intervention to be suitable as settlement land, and the Limitation Zone is prohibited due to natural hazards. This land carrying capacity determination is based on land capability; however, since the result is intended for ideal settlement land, a slope gradient intervention limited to 0–15% was added, based on the guidance of SNI 03–1733-2004 concerning Procedures for Urban Housing Environmental Planning. For further detail, see Table 11 and the chart in Figure 9.

Karangnunggal Subdistrict has a land carrying capacity classification dominated by the Potential Zone, covering an area of 9,154.36 Ha, or 60% of the area of Karangnunggal Subdistrict. This potential zone is based on physical land capability analysis with high and fairly high land capability classifications. Land carrying capacity in Karangnunggal Subdistrict by administrative area can be seen in Table 12.

Viewed by administrative area, the Potential Zone is dominated by Karangnunggal Village, covering an area of 1,044.32 Ha. Meanwhile, the smallest Potential Zone area, 219.82 Ha, is located in Cintawangi Village.

The proportion of potential land carrying capacity reaching 60% indicates the high capability of Karangnunggal Subdistrict to support sustainable development. This land availability serves as both a supporting factor and a limiting factor for the direction of settlement development, given that disaster-prone areas have been excluded through the land capability analysis. This region is thus shown to have sufficiently extensive safe space to accommodate its function as the prospective new regional capital, provided that its development remains controlled through the regulation of land coverage ratios and building height directives. The land carrying capacity map for Karangnunggal Subdistrict can be seen in Figure 10.

In planning settlement development, the proportion of spatial utilization for physical building structures is limited to a maximum of 60% of the potential zone area. The remaining 40% is absolutely allocated for non-built-up areas that integrate ecological functions and disaster risk reduction. In addition to maintaining soil hydrological function, this 40% open space plays an important role as a structural mitigation area to prevent mass erosion in sloped areas and reduce settlement vulnerability to geological and climate-related hazards. This mitigation-based spatial planning approach refers to the criteria set out in Regulation of the Minister of Public Works Number 20 of 2007 concerning Technical Guidelines for Physical and Environmental Analysis (Hasanuddin et al. 2022). The calculation results for the area of land that can be developed for settlement are as follows.

LPm=LWP×60%

LPm=9154,36Ha×60%=5492,62Ha

Based on the potential zone area of Karangnunggal Subdistrict of 10,863.87 Ha, an LPm of 5,492.62 ha is obtained. This value uses a realistic land area that can be utilized for settlement needs. The potential zone area by administrative area can be seen in Table 13.

By administrative area, the area that can be developed into settlement (LPm) is distributed across all villages in Karangnunggal Subdistrict. Settlement development has the largest area in Karangnunggal Village, covering 626.59 Ha. Meanwhile, the settlement development zone (LPm) with the smallest area, 131.89 Ha, is located in Cintawangi Village.

In this study, modeling of settlement land cover change was carried out using the Cellular Automata (CA) approach through the QGIS MOLUSCE tool, utilizing baseline land cover data from 2015 and 2020 as a representation of historical change dynamics. These years were used to capture the transition pattern of settlement land cover over time, so that the model could identify land change tendencies, particularly for built-up land. Based on this historical change pattern, this study then developed a settlement development projection and determined potential settlement land for the year 2045 as a future scenario reflecting the tendency of spatial development if the change pattern and its driving conditions were to relatively continue.

3.3.1. Pemodelan Pengembangan Pemukiman dengan Cellular Automata

In modeling settlement development, the main variables used in the CA model consist of two groups, namely driving factors ( Table 14) and constraint factors ( Table 15). Driving factors play a role in increasing the likelihood of a location undergoing land cover (settlement) change because they represent the level of suitability, accessibility, and ease of development. The selection of driving and constraint variables in this study is based on two main foundations, namely theoretical and normative foundations. Theoretically, the division of variables into driving factors and constraints is a standard component of the transition rules in the Cellular Automata model. This is in line with previous Cellular Automata research (Pratomoatmojo 2018), which applied similar variables in settlement development modeling. Normatively, constraint variables were established based on legally binding regulations, namely Law Number 26 of 2007 on Spatial Planning and Law Number 41 of 1999 on Forestry. This study also establishes a number of constraint factors that function to limit or suppress the likelihood of land cover (settlement) change in certain areas, whether due to environmental risk or protected function.

With this configuration, the CA model forms the transition potential of each cell/grid based on the combined influence of driving and constraint factors. Conceptually, locations with higher carrying capacity, closer proximity to roads and infrastructure, and relatively gentle slopes tend to obtain higher transition potential values, making them more likely to undergo land cover class change. Conversely, locations situated on water bodies, forest areas, riparian zones, and disaster-prone areas have a reduced likelihood of transition. The integration of historical change patterns (2015–2020) with the influence of the driving-constraint variables enables the model to produce spatial predictions for 2030–2045 that do not merely follow area change trends, but also account for the spatial distribution of change according to environmental characteristics and applicable spatial constraint policies.

Overall, the use of multi-temporal baseline data (2015, 2020, 2025) together with a clear separation between driving and constraint factors makes the CA simulation in this study more systematic and accountable. The resulting projection of potential settlement development for 2045 is expected to provide an overview of the direction of potential settlement locations in the future, identify areas potentially experiencing high change pressure, and support further analysis related to the need for spatial utilization control, protection of important areas, and the formulation of development strategies that are more adaptive to environmental risks and constraints.

The following are the stages of the results of the analysis for determining the potential settlement development location in Karangnunggal Subdistrict. The following is the process of settlement development analysis using cellular automata:

The Evaluating Correlation stage in the MOLUSCE plugin is an analysis stage used to measure the degree of relationship (correlation) between the spatial variables or driving factors entered into the Cellular Automata model. The results of this analysis can be seen in Figure 11.

Area Changes in the MOLUSCE plugin functions to analyze the magnitude of land cover area change between classes as well as the land transition pattern from the initial 2010 map (initial state) to the final 2020 map. The results of this analysis can be seen in Figure 12 and Table 16.

Based on the resulting land transition pattern, it can be described that settlement areas in particular show an increasing trend, where in 2015 the area of 1,244.34 Ha increased to 1,771.47 Ha, or rose from 8% to 11% in 2020.

Conceptually, the information in the Area Changes tab is highly important because it forms the basis for establishing the historical change pattern to be used in subsequent modeling, particularly for developing transition probabilities and building the Transition Potential Modelling before the Cellular Automata simulation is run. In other words, the class statistics table provides a quantitative overview of area change. This result can be interpreted as an indication that Karangnunggal Subdistrict, during the 2015–2020 period, exhibited a relatively limited and focused level of change, with most land cover classes remaining stable. Nevertheless, the change observed in settlement land is a signal of a certain transition tendency that requires further analysis (for example, in relation to accessibility characteristics, land carrying capacity, and constraint factors), as this transition pattern is what will influence the shape of the land cover projection for 2045.

Transition Potential Modelling in the MOLUSCE plugin is the process of building a prediction model for the transition potential of land cover change based on the selected spatial variables or driving factors. The results of this analysis can be seen in Figure 13.

The display above represents the Transition Potential Modelling stage in the MOLUSCE plugin. Transition potential is formed using the Artificial Neural Network (ANN) – Multi-layer Perceptron (MLP) method, which functions to learn the relationship between historical land cover change and the driving factors used, enabling the model to estimate the probability of change at each cell/grid. In other words, the ANN is trained to recognize change patterns that occurred in the past and to relate them to location characteristics (accessibility, carrying capacity, slope gradient, and proximity to infrastructure), then produces a transition potential map as the basis for the Cellular Automata simulation toward the projection year (2040).

The learning rate value of 0.100 in Figure 14 indicates the speed of network weight updates during the training process, while the maximum iterations of 1000 indicates the maximum number of iterations permitted until the model reaches a stable (convergent) condition. Using 6 hidden layers to capture non-linear relationships between variables, as well as a momentum of 0.006 to help stabilize learning so that weight changes are not overly fluctuating. These parameters essentially control the balance between the model’s ability to learn complex patterns and maintain stability so as not to produce predictions that are overly sensitive to the training data.

The summary of model performance values shown reinforces this interpretation. A Current Validation Kappa of 0.88 indicates a strong level of agreement between the model’s predicted results and the validation data, so that the model’s ability to represent change patterns is considered very good. In addition, a Min Validation Overall Error of 0.01845 in MOLUSCE indicates a very low minimum mean squared error on the Neural Network validation set, meaning the model’s predictions are relatively accurate in distinguishing areas likely to change from areas that tend to remain stable. The small negative difference between the Min Validation Overall Error and the Current Validation Error indicates that the current accuracy is nearly as good as the best point in Neural Network training, which can technically be understood as a condition where the model has reached stability, such that further iterations no longer meaningfully improve accuracy. Overall, the results of the Transition Potential Modelling stage indicate that the ANN model built has been able to produce a good transition potential map, which can be used as the primary basis for running the Cellular Automata simulation to project land cover from 2030 to 2045 according to the scenarios and variables established in this study.

Validation in the MOLUSCE plugin is the stage for evaluating model accuracy by comparing the simulated map produced by CA-MLP against the validation reference map (Reference Map). The validation results can be seen in Figure 15.

The display above represents the Validation stage in the MOLUSCE plugin. In this study, the Reference Map is the reference map selected to assess the simulation output, while the Simulated Map is the land cover change model output generated by the Cellular Automata (CA) model. This validation is carried out to ensure that the change pattern produced by the model is not only mechanically formed, but also has a high degree of agreement with the reference pattern considered representative. In addition, the Number of validation iterations = 5 parameter indicates that the evaluation process was carried out through several testing iterations, so that the assessment results are more stable and do not depend on a single calculation.

Based on the quantitative indicators shown, model performance can be categorized as very good. A % of Correctness value of 97.17529% indicates that, overall, most cells/grids in the simulated map have the same class as the reference map. Furthermore, a Kappa (overall) value of 0.96743 indicates a high level of agreement after accounting for the possibility of agreement occurring by chance, so that the model’s prediction quality is not only high in percentage terms but also statistically robust. MOLUSCE also displays two additional components, namely Kappa (histogram) of 0.98966 and Kappa (location) of 0.95675. Kappa histogram represents agreement from the composition/quantity aspect (whether the area proportion of each class in the simulation result aligns with the reference map), while Kappa location assesses agreement from the spatial allocation aspect (whether the location of change occurs in a consistent place). A Kappa location value close to 1 indicates that the model is very good at placing change patterns in appropriate locations, while the also very high Kappa histogram confirms that the model is able to maintain consistency in the proportion of each land cover class.

The Multiple-resolution budget graph at the bottom shows the evaluation of model agreement at several resolution levels (from the most detailed scale to a more “loose” scale). In principle, this graph shows that when evaluation is carried out at a coarser resolution (toward the right), the level of agreement tends to increase because small positional differences at the detailed scale become more “tolerated” at a more general scale. The upward pattern of the curve in this graph indicates that the model has successfully formed a spatially consistent change structure, and the remaining differences are generally at the micro-detail level (for example, the shifting of a few cells around class boundaries). Thus, these validation results confirm that the Cellular Automata model built has high reliability for use in developing the 2040 land cover projection, as it is able to represent both the quantitative aspect of change and the spatial distribution pattern of change with a very strong level of agreement.

The Analysis Results based on the stages of analysis for determining the potential location of settlement development using Cellular Automata (CA) with the MOLUSCE plugin can be seen in Table 17.

Based on the final results of the Cellular Automata (CA) simulation, the land cover structure of the study area in 2015, 2020, and the 2040 projection shows that land use composition is generally still dominated by the Plantation/Garden class, followed by Paddy Field and Shrubland, with a consistent total area of 15,325.95 Ha in each period. This consistency in total area indicates that the dynamics occurring represent a shift between land cover classes (conversion) within the study area, rather than a change in the area’s overall extent. Overall, the projection results show that land cover change occurring up to 2045 is moderate, with more pronounced change in the built-up land (settlement) class and certain cultivation land, while water body classes remain relatively stable.

During the 2015–2020 period, the most pronounced change was the increase in the area of Settlement and Activity Area from 1,248.39 Ha to 1,770.39 Ha. This increase was accompanied by a decrease in Plantation/Garden from 8,899.59 Ha to 8,579.09 ha, while water body features remained relatively stable during that period. This change pattern shows that spatial dynamics in the 2010–2020 decade were primarily reflected in the expansion of built-up areas, accompanied simultaneously by a contraction of the dominant cultivation class (plantation/garden), although the magnitude of change remained relatively limited compared to the total area.

Furthermore, in the 2025–2045 projection, the CA model predicts a tendency consistent with the previous period, namely an increase in Settlement and Activity Area from 1,770.39 Ha to 2,618.07 Ha. At the same time, Plantation/Garden again experienced a decrease from 8,579.09 Ha to 8,115.45 ha.

When viewed cumulatively from 2015 to 2045, the most significant change in land cover structure is the increase in Settlement and Activity Area. Interpretively, the consistent tendency of increasing settlement land cover from 2015–2020 and continuing to 2045 is in line with the logic of CA modeling, which relies on the influence of spatial driving factors, particularly land carrying capacity, proximity to the road network, proximity to infrastructure, and slope gradient. Locations that are more accessible, gentler in slope, and supported by facilities tend to have higher transition potential toward built-up function. On the other hand, the application of constraint factors such as water bodies, forest areas, and riparian buffers serves to restrain change so that it does not expand uncontrollably, so that the projection results show change that remains focused and does not disrupt the stability of certain classes to an extreme degree, such that settlement growth and settlement growth locations represent locations with development potential that can be realized without damaging the environment. A comparison map of historical land cover in 2015 and existing 2020 conditions and the settlement simulation up to 2045 can be seen in the following figures.

3.3.2. Development of Potential Residential Areas

Based on the results of the potential settlement development simulation analysis using Cellular Automata (CA), settlement development in 2045 not only illustrates the trend of increasing built-up area, but also represents the combined influence of historical dynamics and spatial constraints that can realistically shape the direction of land use change in the future. The settlement land change map can be seen in Figures 16, 17, and 18. Meanwhile, settlement development data for 2010–2020 together with the 2045 simulation are presented in Table 18. In addition, the settlement development chart is shown in Figure 19, and details of settlement development by village administration for 2015–2020 and the 2045 simulation are listed in Table 19.

The distribution of settlement development by village based on the Cellular Automata simulation shows an increase in the 2045 simulation results for every village, with the highest increase found in Karangnunggal Village, with an area of 155.34 Ha in 2015 and a projected increase in potential settlement development to 411.04 ha by 2045. In field conditions, Karangnunggal Village is the center of activity within Karangnunggal Subdistrict and the southern part of Tasikmalaya Regency, with adequate supporting infrastructure, road access, and sufficient potential land carrying capacity to drive the conversion of land to built-up use. The distribution of potential settlement locations by village administration can be seen in Table 20.

The distribution of potential settlement development by village administration is spread across all villages in Karangnunggal Subdistrict, with the highest potential settlement location found in Karangnunggal Village, covering an area of 33.75 Ha. The map of potential settlement locations can be seen in Figure 20.

The authors would like to express their highest gratitude to the Indonesia Endowment Fund for Education (Lembaga Pengelola Dana Pendidikan/LPDP) for the financial support provided during this academic journey.

•  Adebayo WG: Resilience in the Face of Ecological Challenges: Strategies for Integrating Environmental Considerations into Social Policy Planning in Africa. Sustain. Dev. 2025; 33(1): 203–220. Publisher Full Text
•  Arfiansyah D, Hawken S, Zlatanova S, et al.: Cellular Automata Modelling to Simulate Patterns of Urban Growth for Nusantara: Indonesia’s New Capital. Spat. Inf. Res. 2024; 32(6): 829–849. Publisher Full Text
•  Atlas: n.d. Cellular Automata in GIS.Reference Source
•  Ayuningtias GM, Istanabi T: Predicting Land-Use Changes in Sustainable Food Agriculture Areas in the Southern Suburbs of Surakarta Using Spatial Modeling [Prediksi Perubahan Penggunaan Lahan Pada Kawasan Pertanian Pangan Berkelanjutan Di Suburban Selatan Kota Surakarta Menggunakan Pemodelan Spasial]. Desa-Kota: Jurnal Perencanaan Wilayah, Kota, Dan Permukiman. 2024; 7(1): 175–187. Publisher Full Text
•  Bramantio B, Hizbaron DR, Khakhim N: Prediction of the Future Landuse and Land Cover Changes in the Parangtritis Sand Dune: A Spatio Temporal Analysis Using QGIS MOLUSCE. IOP Conf. Ser.: Earth Environ. Sci. 2024; 1313(1): 012014. Publisher Full Text
•  Butler A: Agricultural Potential of Alluvial Soils in River Basins. African Journal of Geography and Regional Planning. 2024; 11(1). Reference Source
•  Dadzie E, Twumasi YA, Ning ZH, et al.: Integrating Remote Sensing and MOLUSCE to Map and Predict Land Degradation in Baton Rouge. Int. Arch. Photogramm. Remote. Sens. Spat. Inf. Sci. 2026; XLVIII-M-10–2025: 107–112. Publisher Full Text
•  Department of Natural Resources and Environment Tasmania: Land Capability Handbook.2025.
•  Santoso DH, Puryani P, Algary TA, et al.: Land Use Change Analysis Using Plugin MOLUSCE in Yogyakarta Urban Agglomeration Area. INSOLOGI: Jurnal Sains Dan Teknologi. 2025; 4(2): 160–169. Publisher Full Text
•  Edosa BT, Nagasa MD: Spatiotemporal Assessment of Land Use Land Cover Change, Driving Forces, and Consequences Using Geospatial Techniques: The Case of Naqamte City and Hinterland, Western Ethiopia. Environmental Challenges. 2024; 14: 100830. Publisher Full Text
•  Haryati L: Identification of Disaster Risks Associated with the Eruption of Mount Galunggung and Disaster Mitigation Measures in Tasikmalaya Regency [Identifikasi Resiko Bencana Letusan Gunung Api Galunggung Dan Upaya Arahan Mitigasi Bencana Di Kabupaten Tasikmalaya]. Bandung: Universitas Pasundan; 2011.
•  Hasanuddin AS, Rahman R, Rasyidi ES: Analisis Pengembangan Kawasan Kota Baru Pattallassang. Journal of Urban Planning Studies. 2022; 2(3): 230–241. Publisher Full Text
•  Hu L, Duan X, Liu J: Application and Assessment of a CA-ANN Model for Land Use Change Simulation and Multi-Temporal Prediction in Guiyang City, China. Sustainability. 2026; 18(3): 1518. Publisher Full Text
•  Idajati H, Umilia E, Nurliyana FU, et al.: Identification of Environmental Conditions Based on the Carrying Capacity and Holding Capacity Analysis (Case Study: Kecamatan Barat, Magetan). IOP Conf. Ser.: Earth Environ. Sci. 2021; 778(1): 012022. Publisher Full Text
•  Iskandar B, Saidah AA, Kurnia AJ, et al.: Modeling Land Cover Change Using MOLUSCE in Kahayan Tengah Forest Management Unit, Kalimantan Tengah. Jurnal Sylva Lestari. 2024; 12(2): 242–257. Publisher Full Text
•  Ghalehteimouri J, Kamran AS, Mousavi MN, et al.: Predicting Spatial and Decadal of Land Use and Land Cover Change Using Integrated Cellular Automata Markov Chain Model Based Scenarios (2019–2049) Zarriné-Rūd River Basin in Iran. Environmental Challenges. 2022; 6: 100399. Publisher Full Text
•  Jiang X, Li B, Zhao H, et al.: Examining the Spatial Simulation and Land-Use Reorganisation Mechanism of Agricultural Suburban Settlements Using a Cellular-Automata and Agent-Based Model: Six Settlements in China. Land Use Policy. 2022; 120: 106304. Publisher Full Text
•  Kang J, Fang L, Li S, et al.: Parallel Cellular Automata Markov Model for Land Use Change Prediction over MapReduce Framework. ISPRS Int. J. Geo Inf. 2019; 8(10): 454. Publisher Full Text
•  Khan M, Chen R: Assessing the Impact of Land Use and Land Cover Change on Environmental Parameters in Khyber Pakhtunkhwa, Pakistan: A Comprehensive Study and Future Projections. Remote Sens. 2025; 17(1): 170. Publisher Full Text
•  Sitompul K, Nuraini C, Sugiarto A: Analysis of the Carrying Capacity and Land Capacity for Residential Areas in the Southern Part of Medan City. International Journal of Mechanical, Electrical and Civil Engineering. 2025; 2(1): 215–236. Publisher Full Text
•  Kristiadi Y, Herdiansyah H: Environmental Carrying Capacity Modeling Using System Dynamics in the Context of Smart Sustainable City: Jakarta Case Study. Sustainable Urban Development and Environmental Impact Journal. 2024; 1(2): 82–99. Publisher Full Text
•  Kusumastuti AC, Kolopaking LM, Barus B: Faktor Yang Mempengaruhi Alih Fungsi Lahan Pertanian. Jurnal Sosiologi Pedesaa. 2018; 6(2): 130–136.
•  Law on Environmental Protection and Management [Undang-Undang tentang Perlindungan dan Pengelolaan Lingkungan Hidup]: Vol. Official Gazette of the Republic of Indonesia, 2009, No. 140; Supplement to the Official Gazette of the Republic of Indonesia, No. 5059 [Lembaran Negara Republik Indonesia Tahun 2009 Nomor 140, Tambahan Lembaran Negara Republik Indonesia Nomor 5059].2009.
•  Marinescu DC: Nature-Inspired Algorithms and Systems. Complex Systems and Clouds. Elsevier; 2017; Pp. 33–63. Publisher Full Text
•  Mas J-F, García-Álvarez D, Paegelow M, et al.: Metrics Based on a Cross-Tabulation Matrix to Validate Land Use Cover Maps. Land Use Cover Datasets and Validation Tools. García-Álvarez D, Olmedo MTC, Paegelow M, et al., editors. Cham: Springer International Publishing; 2022; Pp. 127–51. Publisher Full Text
•  Miswar D, Suyatna A, Zakaria WA, et al.: Geospatial Modeling of Environmental Carrying Capacity for Sustainable Agriculture Using GIS. Int. J. Sustain. Dev. Plan. 2023; 18(1): 99–111. Publisher Full Text
•  Muhammad R, Zhang W, Abbas Z, et al.: Spatiotemporal Change Analysis and Prediction of Future Land Use and Land Cover Changes Using QGIS MOLUSCE Plugin and Remote Sensing Big Data: A Case Study of Linyi, China. Land. 2022; 11(3): 419. Publisher Full Text
•  Mutaali L: Daya Dukung Lingkungan Untuk Perencanaan Pengembangan Wilayah. ogyakarta: adan Penerbit Fakultas Geografi (BPFG) Universitas Gadjah Mada; 2012.
•  Nandar U: The Proposed New Autonomous Region of Tasela Has Finally Been Included in the Draft of the 2025–2029 Tasikmalaya Regency Medium-Term Development Plan (RPJMD) Calon Daerah Persiapan Otonomi Baru Tasela Akhirnya Masuk Draft RPJMD Kabupaten Tasikmalaya 2025–2029. Radartasik.id; n.d.
•  Nasr M, Orwin JF: A Geospatial Approach to Identifying and Mapping Areas of Relative Environmental Pressure on Ecosystem Integrity. J. Environ. Manag. 2024; 370: 122445. PubMed Abstract | Publisher Full Text
•  NextGIS: MOLUSCE Documentation.2026. Reference Source
•  Noviani, Rita A, Marfu’ah IN: Spatial Patterns Analysis of Land Use Changes Using Spatial Metrics in the Peri-Urban Area of Surakarta City 2023. IOP Conf. Ser.: Earth Environ. Sci. 2024; 1314(1): 012089. Publisher Full Text
•  Pantow P, Lakat RMS, Kapugu H: The Effect of Land-Use Change in the Geothermal Area on the Community Economic Sector in Tompaso District [Pengaruh Perubahan Tata Guna Lahan pada Kawasan Geothermal terhadap Sektor Perekonomian Masyarakat di Kecamatan Tompaso]. Jurnal Spasial. 2022; 9(2): 197–208.
•  Pinto LV, Inácio M, Gomes E, et al.: A Protocol to Model Future Land Use Scenarios Using Dinamica-EGO. MethodsX. 2025; 14: 103283. Publisher Full Text
•  Anggraini P, Tyas NH, Hidayati, et al.: Spatial Mapping Based on the Settlement Carrying Capacity Value in Gunungpati District, Semarang City. BHUMI: Jurnal Agraria Dan Pertanahan. 2023; 8(2): 216–244. Publisher Full Text
•  Pratomoatmojo: Permodelan Perubahan Penggunaan Lahan Berbasis Cellular Automata Dan Sistem Informasi Geografis Dengan Menggunakan LanduseSim [Modeling Land-Use Change Using Cellular Automata and Geographic Information Systems with LanduseSim].2018; 13(1): 26. Publisher Full Text Reference Source
•  Puntsag G: Land Suitability Analysis for Urban and Agricultural Land Using GIS: Case Study in Hvita to Hvita, Iceland. Reykjavik, Iceland: Final Project, United Nations University Land Restoration Training Programme (UNU-LRT); 2014.
•  Rahmawati M, Pratomoatmojo NA: Cellular Automata-Based Modeling of Land-Use Change in the Peri-Urban Area of Surabaya City in Sidoarjo Regency [Pemodelan Perubahan Penggunaan Lahan Berbasis Cellular Automata Pada Wilayah Peri Urban Kota Surabaya Di Kabupaten Sidoarjo]. Jurnal Teknik ITS. 2020; 8(2): C200–C206. Publisher Full Text
•  Riyadi DS, Rustiadi E, Widiatmaka, et al.: Assessing Environmental Carrying Capacity and Disaster Risk in Spatial Utilization: A GIS-Based Study of East Java Province, Indonesia. Land. 2026; 15(4): 537. Publisher Full Text
•  Saharan DD, Nehra DS: Geospatial Assessment And Prediction Of Land Use Changes In Bikaner Urbanizable Area: A QGIS-Based Approach Using MOLUSCE Plugin. Int. J. Environ. Sci. 2025; 130–142. Publisher Full Text
•  Sudipa N: Status Daya Dukung Lahan Untuk Keberlanjutan Pangan Di Kabupaten Klungkung. Jurnal Ilmu Pertanian Indonesia. 2021; 26(4): 597–604. Publisher Full Text
•  Swasto DF, Rahmi DH, Rahmawati Y, et al., editors. Proceedings of the 6th International Conference on Indonesian Architecture and Planning (ICIAP 2022): Beyond Sustainability through Design, Planning and Innovation. Singapore: Springer; 2023. Lecture Notes in Civil Engineering. Publisher Full Text
•  Tamil Nadu Agricultural University: Resource Management:: Soil:: Soil Related Constraints. Tamil Nadu Agricultural University (TNAU); 2013. Reference Source
•  Tasikmalaya Regency Disaster Management Agency (BPBD): Distribution of Disasters in Tasikmalaya Regency [Sebaran Bencana Di Kab.Tasikmalaya]. SIRENA-Tasikmalaya Regency Disaster Management Agency (BPBD) [SIRENA-BPBD Kabupaten Tasikmalaya]; 2025. Reference Source
•  Tefera SA, Thatiparthi VL: Optimizing Land Use and Land Cover Prediction Using an Integrated MOLUSCE Framework in Gubalafto District Ethiopia. Discov. Sustain. 2026; 7(1): 958. Publisher Full Text
•  Utama MS, Dionita NF: Pengaruh Produksi, Luas Lahan, Kurs Dollar Amerika Serikat Dan Iklim Terhadap Ekspor Kacang Mete Indonesia Beserta Daya Saingnya. E-Jurnal Ekonomi Pembangunan Universitas Udayana. 2015; 4(5).
•  Wattimena RV, Wahab NK, Tuanany GO, et al.: ANALYSIS OF THE CARRYING CAPACITY AND SETTLEMENT CAPACITY IN ASSESSING THE READINESS OF KARANGNUNGGAL DISTRICT AS A CANDIDATE FOR THE CAPITAL OF SOUTH TASIKMALAYA REGENCY. F1000Res. 2026; 15: 1279. Publisher Full Text
•  Wattimena RVW, Tuanany NK: Land Capability Unit Scoring, Karangnunggal, Tasikmalaya, Indonesia. Zenodo. 2026. Publisher Full Text
•  Xing W, Qian Y, Guan X, et al.: A Novel Cellular Automata Model Integrated with Deep Learning for Dynamic Spatio-Temporal Land Use Change Simulation. Comput. Geosci. 2020; 137: 104430. Publisher Full Text
•  Yeasmin T, Sourav Karmaker M, Islam S, et al.: Prediction of Land Cover Changes in an Urban City of Bangladesh Using Artificial Neural Network-Based Cellular Automata. Urban Lifeline. 2025; 3(1): 7. Publisher Full Text
•  Yuliani E, Rahman B, Berliana A: The Settlement Development Based on Analysis of Settlement Carrying Capacity and Land Capacity in Pekalongan City. IOP Conf. Ser.: Earth Environ. Sci. 2025a; 1524(1): 012005. Publisher Full Text
•  Yuliani E, Rahman B, Berliana A: The Settlement Development Based on Analysis of Settlement Carrying Capacity and Land Capacity in Pekalongan City. IOP Conf. Ser.: Earth Environ. Sci. 2025b; 1524(1): 012005. Publisher Full Text
•  Zhou L, Feng Y: GIS Spatial Analysis of Land Use Planning Based Onspatial Database and Geographical Weighted regression Model.2025.

Current Reviewer Status:

AWAITING PEER REVIEW

AWAITING PEER REVIEW

?