Prediction of wetland area and depth using linear regression model and artificial neural network based cellular automata

Wetlands are an integral part of the socio-ecological setup of the earth, but their fast transformation beg careful consideration. In this regard, the flood plain wetlands of the Atreyee river basin of India and Bangladesh are not exceptional. The main goal of this study is the mapping of the floodplain wetlands, along with arriving at predictions of their area up to 2039 using the advanced technique of artificial neural network based cellular automata (ANN-CA). Apart from this, prediction of wetland depth using linear regression model is another aim of the present research work. The analysis is executed using 27 Landsat images and Digital Elevation Model (DEM). Results reveal that the present wetland area is 52.92 km2 in pre-monsoon and 518.68 km2 in post-monsoon season, respectively. The composite wetland map from 1987 to 2019 of post-monsoon clearly indicates that 10.48km2 wetlands are identified as hydro-ecologically consistent wetlands. Simulated models reveal that the wetland area from 2009 to 2019 has declined by 66.16km2 and is expected to decrease by 164.62km2 in the next 20 years. Normalized Difference Water Index (NDWI) depth indicates that water availability also may decline significantly in the next 20 years as per regression model-based simulation. All the simulated models were validated with observed wetland area by kappa coefficient, receiver operating curve. The present study will definitely be useful for decision-makers by aiding them in initiatives that take a significant step toward maintaining the wetland landscape, as well as the environment. [Display omitted] •This work intends to delineate wetland area and and monitoring.•The wetland area and depth were predicted for 2029 and 2039.•The simulated wetland area and depth were validated by Kappa coefficient, ROC.•By 2039, the wetland area may be squeezed by 20.4% in comparison to 2019.•Unit wetland can generate 2.5 times greater monetary return than a unit agricultural land.