Yield Estimation Based On Multi-Source Satellite In Typical Grassland

Grassland productivity is an important indicator for estimating grassland yield and diagnosis grassland health. It is immense significance for guiding pastoral livestock production and protecting environment through monitoring spatial distribution and dynamics of productivity in large natural grassland. Remote sensing, characted with no damage, simple, fast and real time, is the major technology of the current yield estimation in large-scale grasslands. Combining multi-source satellite data and enhancing the ovservation frequency will be helpful to understand grassland dynamics and provide scientific advices for grassland management.The study was located in the Xilin River basin of typical steppe. Combination of aboveground biomass from202field sites in July of2012and4kinds of satellite data: Thailand THEOS, Chinese environmental mitigation satellite images, Terra/MODIS, and SPOT/VGT, grassland yield estimation models were constructed based on different programs. On ont hand, we explored the potential of multi-source remote data in yield estimate of typical grassland; on the other hand, we clarified the application of the fresh and dry weight on grassland yield estimate. Finally, the optimization program of yield estimation in typical grassland was chosen. The results show that:(1) Accuracy of estimation model based on medium-resolution data source was higher than that of low-resolution data source. Morever, accuracy based on15m THEOS images showed highest accuracy during medium-resolution data; accuracy based on500m MODIS NDVI product was higher than any other low-resolution data;(2) Both of fresh weight and dry weight can be applied to the grassland yield estimation. Furthermore, fresh weight was used for more precipitation year, and dry weight was used fo less precipitation year;(3) With increasing precipitation, the exponential model was more superior than the linear model in typical steppe area;(4) For heavy precipitation year in typical grassland, this eight vegetation index, such as GNDVI, NDVI, RVI, SAVI, IPVI, OSAVI, TVI and MSAVI were all suitable for yield estimation;(5) Vegetation types and distribution of Xilin River watershed decided the turning green pattern, as east earlier and west later, and yield pattern, as east higher and west lower;(6) Because of mowing grass, aboveground biomass declined begin as the end of July in all Xilin River watershed, and started in mid-August in the eastern and central of the watershed;(7) In the end of August, plants stop growing. Thus, aboveground of all watershed showed decrease, and it was extremely low in the end of September.This study explored the optimization program, and analyzed the biomass dynamics and pattern of Xilin River watershed. It was hoped to improve the theory and the technical level of the grassland yield estimation, and provide scientific basis for the production practice.