Vegetation Parameter And Evaporation Estimation Based On Remote Sensing For Assessing Regional Drought

The water shortage and uneven distribution in time and space is a major limiting factor for the social economic development especially for agricultural production in China. For better managing the limited water resources, it demands quantitatively assessing water consumption and monitoring drought situation. Due to great heterogeneity in land surface, it is difficult to make regional estimation from the traditional field observations. Remote sensing have obvious advantages on quick monitoring of large scale land surface and has been widely used for regional water cycle study. This thesis focused on physically-based model of vegetation parameter retrieval and evapotranspiration estimation based on MODIS data, and its application to assessing regional drought.Firstly, the uncertainties in the MODIS surface reflectance products were discussed. And the effects of uncertainty on the vegetation parameter retrieval were investigated using both the vegetation index method and radiative transfer model inversion method. It was found that the MODIS Collection 4 (C4) product had greater uncertainty than its following product Collection 5 (C5). After temporal composite, the uncertainty in the daily C4 product was reduced. It was also found that the vegetation parameter retrievals using the two methods from the C5 data and 8-day composite C4 data were greatly improved comparing to the daily C4 data.Based on the above understanding, a vegetation parameterization retrieval scheme using the temporal composite of MODIS data was proposed, and the single-layer and double-layer models were developed for estimating the actual evapotranspiration during the winter wheat growing season. The flux observation in the Weishan Irrigation Zone was used to validate the model results. It was proposed the Peman-Monteith algorithm with surface resistance derived from MODIS LAI data was used to make initial estimation of the canopy transpiration in the double-layer model. It was shown that the double-layer model can simulate the actual evapotranspiration better than the model using the Priestley-Taylor method for initialization. It was also found the double-layer model was sensitive to land surface parameters and had no obvious advantages than the single-layer model in this semi-humid zone with dense vegetation cover and small difference between land surface and air temperature.The land surface data assimilation based on the SiB2 model and MODIS data was also tested in this research. The initial soil moisture had a large impact on the modeling results. With the error in the initial soil moisture, the soil moisture results were much improved with the assimilation scheme using soil moisture observation, and the surface heat fluxes results were improved through the assimilation scheme using MODIS surface temperature. It was confirmed that the land data assimilation system combining land surface model and remote sensing had great potential application for studying the water and energy cycles and assessing regional drought.Finally, the crop water stress index (CWSI) based on the modeled actual evaporation was used for assessing the water stress of winter wheat during the main growing period from 2005 to 2007 in Weishan Irrigation Zone. And the temperature vegetation dryness index (TVDI) was used for assessing the drought occurred in the eastern Sichuan Province and Chongqing in 2006. Comparing with the field observations, it was found that the CWSI was less affected by the regional climate conditions and crop growth, and therefore it was a more reliable index for assessing the drought situation.