Satellite Remote Sensing Drought Monitoring Methods Based on Different Biophysical Indicators

As a frequent natural hazard, drought causes the heaviest damage to daily life and ecological environment among all natural disasters. The large scale and dynamic drought monitoring has been frontier and hot topic in the global climate change and food security research. Satellite remote sensing is the main method of obtaining distributed information of the land surface. However, the applicability of a drought monitoring method varies by region and period. So it's necessary to inspect and evaluate the drought monitoring method to ensure the validity and accuracy in drought monitoring.Vegetation index based I_VC, land surface temperature based and Surface Temperature-Normalized Difference Vegetation Index space (T_s-I_NDV space) based are compared to analyze and evaluate their characteristics and applicability in drought monitoring. The MODIS 16-day NDVI (MOD13A2) and 8-day T_s (MOD11A2) provided at 1 km spatial resolution as a gridded level-3 product in the Sinusoidal projection are used, and the Heilongjiang Province, the main food production area in China, is chosen as the study area. The monitoring period is from 9 May to 9 June of the year 2000 to 2008. The correlation between I_VC, I_TC, I_VTC and relative soil moisture in 10 cm and 20 cm depth, the correlation between I_VC, I_TC, I_VTC and the precipitation in the satellite monitoring period and the cumulative precipitation in the last 1-, 2-, 3-, 4-period time scales, the difference and the relationship between I_VC, I_TC, I_VTC are analyzed.Significant linear correlation is found between I_VTC and the relative soil moisture in 10 cm and 20 cm depth, especially for the 10 cm depth. The correlation between I_VTC and relative soil moisture is obviously better than that between I_TC, I_VC and the relative soil moisture. So, it can be judged that I_VTC may mirror the soil moisture better than I_VC and I_TC, and is more sensitive to shallow soil moisture. I_VTC is also found linearly correlated with precipitation in current monitoring period, as well as the cumulative precipitation in the last 1-, 2-, 3-, 4-period timescales. The correlations are better than those between I_TC, I_VC and precipitation as well as corresponding cumulative precipitation, which show that I_VTC is more sensitive to precipitation than I_VC and I_TC, and is closely related to not only current precipitation but also past cumulative precipitation. In the early growing season, I_VTC and I_TC are applicable in drought monitoring, while the fractional I_VC cover is too low to monitor drought. It is difficult to compare the drought among different areas with I_VC and I_TC, while based on the energy conservation principle, combining the I_NDV and T_s, I_VTC can reflect soil moisture better and is comparable in different areas. Furthermore, the land surface temperature implied by I_VTC gives more direct hint of drought than the I_NDV implied especially on the grassland, then in the crop land, brush land and forest land in sequence.