Bao Gang,ao Yulong，Alateng Tuya，Bao Yuhai，Qin Zhihao，Wang Mulan，Zhou Yi
Based on the third generation GIMMS NDVI time\|series datasets during 1982~2011，we extracted the start of growing season (SOS)，end of growing season (EOS) and length of growing season (LOS) in the Mongolian Plateau using cumulative NDVI based logistic regression curves，change rate of curvature in NDVI logistic regression curves and change rate method of NDVI and further analyzed the spatio\|temporal changes of phenology.The results showed that the cumulative NDVI based logistic regression curves and change rate method of NDVI performed better predictions in SOS and EOS modeling，and the mean value of these two methods improved the extraction accuracy of phenology in the Mongolian Plateau.SOS in the Mongolian Plateau mostly started from the middle of April to the end of May and ended from the end of the September to the middle of the October.Most LOS ranged from 125 to 175 days.Spatially，the earlier SOS，later SOS and longer LOS occurred in the humid and sub\|humid area of the plateau，and later SOS，earlier EOS and shorter LOS occurred in arid and semi\|arid regions of the plateau.Temporally，during the 30\|year observation period，approximately，51.6% and 33.9% of the plateau experienced advanced and delayed SOS，respectively，and 21.2% and 12.4% of which are statistically significant；Approximately，35.6% and 49.8% of the study area experienced delayed EOS，respectively，and 8.2% and 12.0% of which are statistically significant；Accordingly，40.3% (17.8% are significant) and 44.8% (18.9% are significant) of the plateau showed shortening and lengthening of the LOS.
