Vegetation land fragmentation has had numerous negative repercussions on sustainable development around the world. Urban planners are currently avidly investigating vegetation land fragmentation due to its effects on sustainable development. The literature has identified a research gap in the development of Artificial Intelligence [AI]-based models to simulate vegetation land fragmentation in urban contexts with multiple affecting elements. As a result, the primary aim of this research is to create an AI-based simulation framework to simulate vegetation land fragmentation in metropolitan settings. The main objective is to use non-linear analysis to identify the factors that contribute to vegetation land fragmentation. The proposed methodology is applied for Western Province, Sri Lanka. Accessibility growth, initial vegetation large patch size, initial vegetation land fragmentation, initial built-up land fragmentation, initial vegetation shape irregularity, initial vegetation circularity, initial building density, and initial vegetation patch association are the main variables used to frame the model among the 20 variables related to patches, corridors, matrix and other. This study created a feed-forward Artificial Neural Network [ANN] using R statistical software to analyze non-linear interactions and their magnitudes. The study likewise utilized WEKA software to create a Decision Tree [DT] modeling framework to explain the effect of variables. According to the ANN olden algorithm, accessibility growth has the maximum importance level [44] between -50 and 50, while DT reveals accessibility growth as the root of the Level of Vegetation Land Fragmentation [LVLF]. Small, irregular, and dispersed vegetation patches are especially vulnerable to fragmentation. As a result, study contributes detech and managing vegetation land fragmentation patterns in urban environments, while opening up vegetation land fragmentation research topics to AI applications.

Land fragmentation can define as the “situation where one area/unit is composed of a large number of parcels that are too small for their rational utilization” [5]. Land fragmentation affects sustainable development through its multiple impacts on environmental, economic, and social costs [13]. Effective land use management and policy decisions are always based on understanding, modeling, and predicting land-use changes in cities [9]. Therefore, the land fragmentation process should systematically investigate to provide a wide-ranging set of land use indicators to support sustainable development [12]. Built-up land fragmentation is the fragmentation or division of the built-up plots or units within the built-up land-use area horizontally. The objective of this study is to frame a Decision Tree (DT) model to identify the non-linear relationships between the Level of Built-up Land Fragmentation (LBLF) and its influencing factors in urban areas. The sub-objective is to quantify the LBLF in the Western Province, Sri Lanka. The study scope limits to LBLF and Decision Tree (DT) non-linear classifier. The study further quantifies the LBLF from 2000 to 2010 in Western Province, Sri Lanka as an initiation to frame the DT model.

Muthurajawela Marsh and Negombo Lagoon have lost their marshland and mangroves from 41% and 21% accordingly from 1997 to 2017. Vegetation land fragmentation (VLF) is the foremost reason that accelerates the vegetation cover change. However, Sri Lanka has not been given sufficient attention to investigate VLF with case studies. Further, fragmentation patterns like dissecting, shrinking, and disappearing are still essential to investigate with empirical studies to comprehend the phases of VLF. Therefore, this study attempts to quantify the level of vegetation land fragmentation (LVLF) in the Western Province’s Grama Niladhari Divisions (GNDs) from 2000 to 2020 by applying the Landscape Division Index (FRAGSTATS). It further analyses the patterns of VLF by using 200 fragmented locations. The key findings demonstrate that 54% of GNDs have more than 60% of the LVLF in the Gampaha District from 2000 to 2010. Although it has reduced up to 49% from 2010-2020, the Gampaha District remains in the highest position. It additionally discloses the significant increment in LVLF along the expressways from 2010 to 2020. 46% of fragmented patches indicate the mixed of dissecting, shrinking, and disappearing patterns. Therefore, the study provides methodological application and quantitative insights for planners to investigate VLF and its patterns.