Abstract: This study analyzed the changes in grassland ecosystem health from 2001 to 2017 in the Ebinur Lake Basin. The research was based on MODIS and Landsat data, and used information entropy principles and a pressure-state-response (PSR) model to evaluate the health status of grassland in the Ebinur Lake Basin in 2017, based on the change trends and current data. It was found that: 1) From 2001 to 2017, the proportions of land area in the Ebinur Lake Basin for which environmental health of the grassland ecosystem has improved, remained unchanged or deteriorated were, respectively, 29.5%, 59.4% and 11.1% of the total grassland area. The ecological environmental health of the environment is generally stable or is improving. 2) In 2017, grassland ecosystem ecological environment health index ranged between values of 0.1097 and 0.3892, indicating that the overall health level was not high. Hence there is a need strengthen the protection and management provisions for grassland resources. 3) The results were used to devise and map three proposed grassland management categories for the grassland ecosystems in Ebinur Lake Basin: a governance area, a protection area and a prevention area. The governance area (in which management should be strengthened to ensure current health status is maintained) accounts for 39.1% of the total grassland area, mainly located in mountain foothills and desert grassland. The protection area (where effort is needed to reverse a current degradation trend) accounts for 34.2% of the total grassland area, mainly distributed in particular mountain foothill regions. The prevention area (generally with good health status but needing management of population pressures from nearby cities) accounts for 26.7% of the total grassland area, mainly in the areas with relatively concentrated towns.

Key words: grassland ecosystem, time information entropy, PSR model, health diagnosis