Microwave Signatures of Inundation Area

Abstract:

The goal of this research is to use the inundation data available from 250 m Moderate-Resolution Imaging Spectroradiometer (MODIS) Near Real Time (NRT) as supplementary data to quantify the signature of different levels of intra-monthly percentages of water-cover on spectral channels of the Tropical Rainfall Measuring Mission Microwave Imager (TMI). The analysis relies on an unsupervised clustering methodology, which uses information content of the magnitudes and polarization differences of all TMI channels. This clustering method is applied to the monthly averages of the spectral brightness temperatures in calendar year 2013, focusing on two important deltaic regions: Mekong and Ganges-Brahmaputra-Meghna. The results illustrate that during the non-raining condition, the high-frequency brightness temperatures at 85V,H GHz can be used as a surrogate of inundation while for the “all-sky” condition, the low frequency channel of 10 GHz, is the one that is hardy influenced by atmospheric constituents but is not very suitable for the detection of small-feature inundation because of its coarse resolution.

The monthly inundation dynamics of the two aforementioned delta regions are demonstrated on each frequency channel separately and on the clusters of all TMI spectral channels for non-rainy and “all-sky” scenes. The mean and variance of the identified clusters are used to detect the inundated percentage in another deltaic region, the Yellow River delta, while the MODIS NRT data are exploited to characterize the error of inundation mapping.