Risk Assessment Of Water-logged Disaster Expressed By The LST-EVI Space In Jianghan Plain

Water-logged low-yield farmlands are distributed widely along rivers and lakes plain in the South China. Water-logging disaster could impact on food production seriously and its long history. Therefore, it plays an important role in the improvement of water-logged farmlands. The traditional measures to identify and water-logged area are based on field experiments and measurements, which are time-consuming. To seeking an effective method is urgent to solve this problem and RS will provide technical support. This paper selects four areas in Jianghan Plain as the study area. It is trying to extract the water-logged farmlands and then spatial analysis based on Land sat 5 TM and DEM data. The results can provide a method for rapid monitoring and identification of water-logged farmland, enriching the theory of soil moisture content in humid plains, which is important theoretical and practical significance.Firstly, Land sat TM data is April 20 in 2005. Radiometric calibration and atmospheric correction conduct after the preprocessing such as layer stacking and subset; spectral changes before and after atmospheric correction are analyzed; classification and accuracy assessment of classification are proceed after atmospheric correction; then it is LST inversion, and emissivity estimated by pixel decomposition method. Secondly, LST-NDVI and LST-EVI spaces are established, and the latter are better reflecting the relationship between temperature and vegetation than LST-NDVI by the fitting of "dry edge" and "wet edge". So the fitting coefficients of the LST-EVI space are used to calculate the TVDI. Finally, waterlogged farmlands are extracted by spatial analysis by ArcGIS10.0 on the basis of farmland pixels, DEM data and TVDI; analyzing three main factors of water-logged disaster:water networks, relief and elevation, and to statistic and spatial analysis the distribution of water-logged farmlands.This study verifies the scientific, reasonable, feasible and practical of the principle of water-logged farmlands detection. And it provides a quick and effective method for the identification of water-logged farmlands. The main results are as follows:(1) First time to use the LST-VI space for recognition of waterlogged farmlands. LST-VI space is representation of the relationship between soil moisture, temperature and vegetation, so it can be used in the research of waterlogged. This study proposes and verifies the feasibility of the method;(2) Proposed the thresholds and criteria for identification of waterlogged farmlands using remote sensing with the support of GIS spatial analysis:farmland pixels satisfying the 0.12<EVI< 0.31 and 23.1 m< DEM< 34 m; TVDI of farmland pixel satisfying 0<TVDI< 0.45;(3) Our results show that the area of waterlogged farmlands is about 339.69km2 in study area in 2005, the elevation of waterlogged farmlands concentrated ranges from 24 m to 29m.