Remote sensing monitoring of green-up dates in the Xilingol grasslands of northern China and their correlations with meteorological factors

The grassland vegetation green-up date helps to initiate the start of grassland grazing and is an important scientific issue in grassland phenology research. This study examined green-up in the Xilingol grasslands of northern China. A double logistic function was used to reconstruct a time series of Satellite Pour l'Observation de la Terre Vegetation (SPOT Vegetation) Normalized Difference Vegetation Index (NDVI) data from 1999 to 2012. The dynamic threshold method was used to monitor the green-up date via remote sensing. The study found that green-up dates occurred between early April and mid-May in most areas. Sample plots indicated that 68% of ground monitoring results were consistent with the remotely sensed green-up date. The root mean square error of the green-up date measured via remote sensing was 8.7days. The results are summarized as follows. (1) The green-up date of temperate desert grassland usually occurred around the first half of April but could be quite variable (standard deviation=28.65days). The green-up date of temperate grassland usually occurred in mid- to late April, and was somewhat variable (standard deviation=18.45days). The green-up date of temperate meadow grassland usually occurred in late April or early May with little variability (standard deviation=12.97days). (2) Linear trend analyses showed that only 13.9% of the pixels exhibited significant changes over the 14-year time series. However, the linear trend increased from the southeast to the northwest, with 57% of pixels demonstrating earlier green-up. Across the region, the linear trend exhibited an average rate of change of - 1.5days 10year(-1). In addition, we discuss correlations between green-up dates and meteorological factors at eight meteorological stations and analyse partial correlation coefficients between green-up dates and meteorological factors. The correlations measured at seven meteorological stations were statistically significant (p<0.05). The average correlative value of deterministic factors was 0.54. The partial correlation coefficients between green-up date and temperature prior to seedling establishment were negative at five stations, but this relationship was significant at only three stations. Partial correlation coefficients between green-up date and precipitation prior to seedling establishment were negative at all eight stations, of which five relationships were significant.