Impacts of the Threshold Value of the Enhanced Vegetation Index on Surface Water Area and Extent in MODIS Imagery

Impacts of enhanced vegetation index (EVI) on the monitoring of surface water from the MODIS imagery are presented for the lowland region of the Mekong River consisting of the Tonle Sap Lake and its tributaries, the Tonle Sap channel, the Mekong River from Kraite to Tan Chau, and the Bassac River. Records of 839 MODIS images and daily water elevation at ten locations in the domain of interest are collected in the period from 2000 to 2019. A wavelet-based filter algorithm is applied for detecting the water surface. Using the relation between observed water elevation and surface water area and extent, a sensitivity analysis of EVI is performed to determine and classify the non-flood, mixture, and flood pixels in MODIS images. The results showed that the EVI affects significantly correlation coefficients. A value of 0.95 is obtained for the magnitude of the correlation coefficient at most studied locations when appropriate values of 0.30 and 0.45 are used for classifying non-flood, mixture, and flood pixels in MODIS images. Spatio-temporal variation of surface water area and extent associated with different flow conditions is also discussed.