Land-Use/-Cover Dynamic Modeling Using RS and GIS with Emphasis on Maximum Likelihood Rule and Transition Matrix

Aims: Understanding the land-use and land-cover (LULC) change pattern is important for prospering environmental restoration. The present study aimed to study changes in LULC patterns of the Koupal Watershed in Khuzestan Province. Materials Methods: This study focused on changes in LULC patterns using remote sensing techniques and geographic information systems (GIS). For this purpose, the Multi-temporal satellite images of the Landsat series (1998 and 2020) were acquired, preprocessed, and used to extract LULC maps by machine learning method including the Bayes discriminant and Maximum likelihood rule over 22 years. The reliability of classified maps was checked using a confusion matrix. The transition matrix and change rate were extracted by change detection analysis. Findings: Change detection analysis shows that vegetation cover witnessed of dramatic decrease and changed from 27.6% to 0.06%, followed by water body reduction from 8.59% to 0.79% and bare land decrease from 57.9% to 51% of the whole area and a rapid expansion of cropland from 5.44% to 41.25%. The change matrix revealed that 93% of cropland remained unchanged, followed by bare land (71%), built-up (53%), water body (7%), sand dune (6%), and vegetation (0.05%). Conclusion: These results establish LULC trends in the past 22 years and provide useful data for planning and sustainable land-use management. The findings presented in the study should be applied as an approach to create awareness and increase land-cover protection and halt land-cover change.