Exploring physical wetland vulnerability of Atreyee river basin in India and Bangladesh using logistic regression and fuzzy logic approaches

Abstract

In spite of the immense ecological importance of the wetland, wetland habitats are losing their existence in the floodplain regions. In the present study, we have explored the nature of the physical vulnerability of wetland using logistic regression (LR) and fuzzy logic (FL) approaches for both pre and post-dam periods. LR is selected because initial vulnerable is identified and FL is used to explore the best possible alternative interwoven relation among the parameters. Seven parameters like (Water presence frequency (WPF) map, Flood inundation map (2d hydraulic model) (FIM), Frequency of pixel being non permanent (NPW), Agriculture presence frequency (APF) map, Fragmentation of wetland (FoW), Normalized differences built-up index (NDBI), and Wetland changes (WC)) are employed for this study. Result reveals that 9.97 and 8.99% wetland areas were under high and very high vulnerability as per LR model but it is increased to 23.35% and 12.52% in post-dam period. Similarly, as per FL model, very high vulnerable wetland habitat is increased from 9.93% to 25.28%. Sensitivity analysis using Kappa coefficient and Receiver operating curve shows the validity of the models. About 31–68% flow reduction in Atreyee river in post-dam period, squeezing of 25.80% active flood-prone zone, agricultural extensions are identified as the major reasons behind growing wetland vulnerability. Vulnerable wetland habitat will automatically reduce productivity and service capability of the wetland ecosystem. Cause optimization may be the vital step for reducing the vulnerability. Ecological flow estimation and downstream release of water can play a decisive role for arresting the growing trend of vulnerability.