Prediction Of Vegetation Phenology With Atmospheric Reanalysis Over Semi-arid Grasslands In Inner Mongolia

Vegetation phenology is a sensitive indicator of climate change and vegetation growth.In the last decade,the warming trend has accelerated,gradually affecting the physical changes on the earth.Accurate understanding of vegetation phenology can not only understand vegetation growth and change,but also predict sudden climate events,which is of great significance to the study of ecosystem and human activities.The Inner Mongolia is dominated by continental semi-arid climate.As an important part of the Eurasian grassland,the grassland covers a vast area and it has the largest grassland pasture in China.Therefore,the prediction of vegetation phenology in the semi-arid grassland in Inner Mongolia can provide a certain theoretical basis for future research.In this study,we used 32 years(1983-2014)remote sensing NDVI data set to estimate the start time of vegetation growing season(SOS)and the peak vegetation growth(POS),and explored their responses to independent atmospheric variables such as air temperature,precipitation,solar radiation,wind speed and soil moisture.First,the accuracy of three important atmospheric variables,air temperature,precipitation and wind speed,was verified.After excluding autocorrelation factors,the forward feature selection method was used to determine the linear or nonlinear correlation between each atmospheric variable and SOS and POS.Secondly,a generalized additive model(GAM)was established to analyze the correlation between phenological parameters and their variables at different time scales.Finally,the sensitivity analysis was carried out to explore the most sensitive factors.The results show that :(1)the accuracy of air temperature data in ERA5-Land data set is higher,with an average difference of 1.5K from the observed value.However,the accuracy of precipitation data in ERA5-Land data set is not as good as that of CHIRPS precipitation data and.Under the same correlation coefficient,the deviation of CHIRPS precipitation data set is smaller.At the same time,the ERA5-Land data set significantly underestimated the wind speed,with a deviation of 1.15m/s compared with the observed value.(2)Soil moisture and precipitation were linearly correlated with SOS,while other variables were nonlinearly correlated.At the same time,air temperature and precipitation were linearly correlated with POS,while wind speed,soil moisture and solar radiation were linearly correlated with POS.(3)Furthermore,it was found that the aforementioned independent variables from the previous year could contribute to approximately 63%-85%of the SOS variations in the present year,whereas the atmospheric variables from April to June could contribute to approximately 70%-85% of the POS variations in the same year.Finally,the SOS and POS predicted by the GAM exhibit significant agreement with the root mean square error of approximately 3 to 5 days when compared to those obtained from the satellite NDVI dataset.(4)The results of sensitivity analysis show that SOS is more sensitive to air temperature and precipitation gathering wind speed,while POS is more sensitive to air temperature,wind speed and solar radiation.