GENERATING TIME SERIES OF MEDIUM-RESOLUTION TEMPERATURE VEGETATION DROUGHT INDEX IMAGES USING A KALMAN FILTER METHOD FOR SOIL MOISTURE CHANGE ANALYSIS
Keywords: Temperature Vegetation Drought Index, Data fusion, Time series, Kalman filter, MODIS, Landsat
Abstract. Soil moisture is one of key environmental variables that affect vegetation cover and energy exchange between the land surface and the atmosphere. Satellite remote sensing technology can provide information for monitoring large-scale soil moisture dynamics quickly. The temperature vegetation dryness index (TVDI) acts as an effective indicator of inferring soil moisture status which is calculated according to the empirical parameterization of composed of the land surface temperature (LST) and the normalized difference vegetation index (NDVI) characteristic space. In this paper, the MODIS TVDI was calculated based on MODIS LST product (MOD11A2, 1 km) and NDVI data (derived from MOD09A1, 500m). Meanwhile, LST and NDVI from Landsat8 OLI images were estimated to obtain Landsat-based TVDI. Then, a Kalman filter algorithm was used to simulate TVDI time series data with 30m resolution and a revisit period of 8 days combining TVDI derived from Landsat and MODIS data. We selected the west of the Songnen Plain, China as the test area and high quality cloudy-free images during growing season (April to October) of 2018 as the input data. The predicted TVDI time series data of medium resolution not only improved the temporal resolution to capture the changes at fine scale within a short period, but also made up for the deficiency of low spatial resolution MODIS data. The results show that it is feasible to generate medium or high resolution TVDI time series data by applying different remote sensing data by Kalman filtering algorithm.