Abstract: With MODIS NDVI data from 2001 to 2017 smoothed by the S-G and HANTS methods, we applied three phenological extraction models (the cumulative NDVI based Logistic curvature (CUM), rate of change of curvature (RCC), dynamic threshold method) to identify the start of growing season (SOS) in the Mongolian Plateau. The response characteristics of the SOS to climate change were analyzed on the vegetation type scale. The results showed that although the S-G method was slightly better than the HANTS method, both of them had relatively less influence on extracting the SOS. Among the three phenological extraction methods, CUM method had better extraction accuracy. The average SOS of the Mongolian Plateau was approximately 124 d. Spatially, the earlier SOS was observed in the southwestern part of the plateau and the Hentiy Mountain areas, whereas the later SOS occurred in Hangay Mountains and western part of the Greater Khingan Mountain. Temporally, the SOS generally showed a weak advancing trend (-0.04 d·a^-1). Spatially, the delayed SOS was observed in the Hangay Mountains, and the advanced SOS occurred in the Greater Khingan Mountain, Hentiy Mountain, the Great Lakes basin, and central and western Inner Mongolia. At the vegetation type scale, expect for the forest vegetation (-0.67 d·a^-1), there were no apparent trends of SOS in other vegetation types. The temperature prior to the SOS, especially in March, negatively correlated with the SOS in the Mongolian Plateau, and the precipitation in the autumn and winter of the previous year had a positive correlation with the SOS.

Key words: water requirement, water deficit, winter wheat., climate change, Penman-Monteith equation