Water Body Mapping and Monitoring using Landsat Time Series Satellite Images

Abstract

Bangladesh is believed to be extremely vulnerable to climate change, which may result in
abnormal spatio-temporal pattern in rainfall and increased variability of temperature across
the region. Consequently, the frequency and intensity of various natural hazards are expected
to increase, which may affect the availability of fresh water on the surface as well as
underground. That’s why mapping of water body and its continuous monitoring is important
(Regional Water Report 37, FAO 2011).This study aims at identifying water body in the
coastal belt of Bangladesh using Landsat 5 TM time-series satellite images for the year 2000,
2005 and 2010. Satellite derived indices e.g. WRI, NDVI, NDWI, MNDWI, AWEI, NDMI are
computed from Landsat data of 2000, which were compared with the base map for selecting
the best index for water body identification. The result shows that NDWI is more robust in
extracting water bodies compared to other indices. Furthermore, unsupervised and
supervised image classification techniques have been applied on all three years data. Both
the index images as well as the classified images are reclassified to produce binary images
showing water and non-water area. Average accuracy of the classification is 88%. Result
shows that there is remarkable increase in water area after 2005. The reason might be
attributed to the fact that the study area has suffered from several natural calamities during
the study period.