Urban wetland fragmentation and ecosystem service assessment using integrated machine learning algorithm and spatial landscape analysis
Subhasis Das, Partha Pratim Adhikary, Pravat Kumar Shit, Biswajit Bera
Abstract
Dynamics of ecosystem service value (ESV) of various wetlands has been assessed by researchers globally. But the impact of spatio-temporal variation of landscape metrics on ESV in the lower Gangetic plains has not been examined, fully. The present work has established linkages between landscape metrics and ESV in Kolkata urban agglomeration using support vector machine and multivariate regression analysis. Result indicates that wetland area has been reduced by 5.26%, 13.67% and 9.03% during the periods 1990–2000, 2000–2010 and 2010–2020, respectively and the ESV contributed by wetlands has been decreased by $131428, $323674 and $184649, respectively during the same period at an annual rate of 0.85%. Number of patches, mean patch area and edge density are the main determinants of wetland fragmentation and decreased by 44.12%, 10.23% and 8.65%, respectively during the last three decades. A wetland restoration strategy based on dynamic restoration, reactive restoration and wetland creation for the study area has been formulated, which can guide for sustainable management of wetland resources in Kolkata urban agglomeration.