Abstract: The grassland of YunnanGuizhouSichuan Province serves as an important part of grassland in China, which plays an important role in the construction of ecological civilization and global climate change. It is of great significance to analyze the change of net primary productivity (NPP) of grassland. This study calculated the NPP of  grassland using the improved CASA model, and analyzed spatiotemporal characteristics and influencing factors of grassland ecosystem productivity in YunnanGuizhouSichuan  from 1996 to 2015 based on the remotesensing image data of AVHRR and MODIS, and the meteorological data of the surrounding areas. The results showed that: (1) The NDVI values  of grassland in Yunnan, Guizhou and Sichuan Province obviously showed a regular pattern during the past 20 years, and the annual average value of  NDVImax showed an increasing  trend in 97.18% of the grassland area. (2) The NPP of YunnanGuizhouSichuan grassland showed a spatial pattern of ‘low value in the northwest, high value in the southwest  and medium value in the east’. The annual average NPP was 420.85 gC·m-2 in the past 20 years. (3) The annual average NPP value of YunnanGuizhouSichuan area had a certain  fluctuation, and it increased at the rate of 7.232 2 gC·m-2·a-1 in the past 20 years. The NPP was the highest in summer with an average value of 160.25 gC·m-2. It showed an  average NPP value of 109.10 gC·m-2 in spring and 106.10 gC · m-2 in autumn. The average value in winter was 54.03 gC·m-2, which is the lowest.
Key words：grassland; net primary productivity (NPP); vegetation index (NDVI); spatiotemporal characteristics; YunnanGuizhouSichuan Provinces