Understanding climate controls on spring phenology in grassland ecosystems is critically important in predicting the impacts of future climate change on grassland productivity and carbon storage. The third-generation Global Inventory Monitoring and Modeling System (GIMMS3g) normalized difference vegetation index (NDVI) data were applied to derive the start of the growing season (SOS) from 1982-2010 in grassland ecosystems of Ordos, a typical semi-arid area in China. Then, the conditional Granger causality method was utilized to quantify the directed functional connectivity between key climatic drivers and the SOS. The results show that the asymmetric Gaussian (AG) function is better in reducing noise of NDVI time series than the double logistic (DL) function within our study area. The southeastern Ordos has earlier occurrence and lower variability of the SOS, whereas the northwestern Ordos has later occurrence and higher variability of the SOS. The research also reveals that spring precipitation has stronger causal connectivity with the SOS than other climatic factors over different grassland ecosystem types. There is no statistically significant trend across the study area, while the similar pattern is observed for spring precipitation. Our study highlights the link of spring phenology with different grassland types, and the use of coupling remote sensing and econometric tools. With the dramatic increase in global change research, Granger causality method augurs well for further development and application of time-series modeling of complex social-ecological systems at the intersection of remote sensing and landscape changes.