Monitoring the spatiotemporal dynamics of waterlogged area in southwestern Bangladesh using time series Landsat imagery

Abstract

Waterlogging is becoming a major environmental problem and challenge for socio-economic development in the southwestern part of Bangladesh. In this study, the Satkhira district was selected as the study area to quantify waterlogging area delineation. To portray these dynamics, Landsat imageries from 1973, 1989, 1995, 2000, 2005, 2010 and 2015 were used. A training dataset was generated in ArcGIS, and a supervised classification was carried out using the random forest algorithm in the R Studio. The overall classification accuracy and kappa statistics was 95% and 91% respectively. Post-classification change detection comparisons were made in QGIS to calculate the transformations of the respective land cover areas in the study site. Areas of approximately 832, 3033, 13,562, 11,547, 27,162, 40,056, and 35,606 ha were observed as waterlogged in the above mentioned years, respectively, which indicates that the acreage of waterlogged areas increased approximately 43 fold from 1973 to 2015. Moreover, the Land Use Land Cover (LULC) change matrix showed that waterlogged area was increased from 5% to 12% during 2000–2005, further rose from 12% to 18% during 2005–2010 and decreased from 18% to 16% during 2010–2015. The most water logged sub-district was Debhata (38%) while Koloroa has the lowest (4%) waterlogged area. The study is an effort to reconstruct the history to understand the dynamics of the waterlogging as past ground monitoring information is absent. Regarding environmental degradation, the government and development agencies should consider these results a critical issue in the entire southwestern part of Bangladesh.