| Xinjiang is located in the arid inland zone,and under the coupled influence of global climate change and human activities,drought-generated losses in Xinjiang are on an expanding trend.Grassland ecosystems are extremely sensitive and fragile,and grassland phenology is highly sensitive to climate change,but no detailed investigation has been conducted on the response of grassland phenology to drought events in Xinjiang.In this study,we use remote sensing parameter extraction,machine learning,model simulation,time series analysis,trend analysis,and correlation analysis to clarify the trends and internal fluctuations of dry and wet conditions and climate extremes in Xinjiang in the past 60 years(1960~2019),and to grasp the situation of meteorological drought(1960~2019),actual evapotranspiration(2003~2019),and agricultural drought(2000~2019)in Xinjiang.Meanwhile,this study predicted the spatiotemporal evolution of drought based on machine learning methods,elucidated the spatiotemporal variation of grassland vegetation phenology in Xinjiang in the last 20 years(2000~2019),and explored the response mechanism of grassland phenology to climate change and drought in Xinjiang.The main conclusions of the study are as follows:(1)Xinjiang has warmed significantly in the past 60 years,with air temperature and ground temperature tending to increase at a higher rate than the national average of0.33°C/10 a and 0.53°C/10 a,respectively,and the ground temperature increasing more than the air temperature,with the winter and spring increasing more than the summer and autumn,and the nighttime increasing more than the daytime.Precipitation tends to increase at a rate of8.45 mm/10 a,with large interannual fluctuations,uneven spatiotemporal distribution of precipitation over the years,and there is a small decrease in precipitation at the early stage and a continuous increase at the later stage on an interdecadal scale,with the highest and largest increase in precipitation in summer among all seasons,and greater precipitation in mountainous areas than in basins and plains.The average potential evapotranspiration(ETp)in Xinjiang was 2737.67 mm,with a tendency rate of 36.28 mm/10 a.The ETp increase the most in summer among all seasons,while it showed a small decrease in winter,and the ETp was higher in the basin plains than in the mountains.The number of sunshine hours shows the distribution characteristics that the eastern part of Xinjiang is higher than the western part,and the plains are higher than the mountainous areas,but the sunshine hours are decreasing in the whole Xinjiang probably due to the influence of cloudiness,etc.And the tendency rate of sunshine hours is-23.20 h/10 a.The extreme climate change in Xinjiang in recent 60 years shows that the 5 extreme low temperature indices are all decreasing in different degrees,the 4extreme high temperature indices are all increasing in different degrees,and the 9 extreme precipitation indices are all increasing in different degrees.(2)Standardized Precipitation Index(SPI)and Standardized Precipitation Evapotranspiration Index(SPEI),which characterize the meteorological drought in Xinjiang have similar time series trends.Different time scales(1,3,6 and 12 months)of meteorological droughts indicate different durations of dry and wet periods in Xinjiang,as shown by monthly scale,seasonal scale and semi-annual time scale of SPI and SPEI,Xinjiang exhibits a high frequency of dry and wet period transitions from 1960 to 2019,while the 12-month scale SPI and SPEI indicate a gradual change from mild drought to mild wet in Xinjiang since 1987.Based on the Surface Energy Balance Algorithms for Land(SEBAL),the Actual Evapotranspiration(ETa)in Xinjiang was fitted with the FAO Penman-Monteith formula with an accuracy of R2=0.76,p<0.001,indicating that the SEBAL model has good applicability in extracting ETa in Xinjiang region,where the multi-year means ETa ranged from 3.60 to 1483.21 mm between 2003 and 2019,and the region-wide multi-year ETa averaged 523.83 mm.To explore the spatial and temporal evolution characteristics of agricultural drought in Xinjiang,the Temperature Vegetation Drought Index(TVDI)and Temperature Fluorescent Drought Index(TFDI)were calculated based on the principle of feature space,and the results showed that TVDI and TFDI have similar trends,and the average multi-year agricultural drought in Xinjiang is in "mild drought" and "moderate drought".The spatial distribution of agricultural drought in Xinjiang has a vertical zonal distribution pattern,from mountainous areas to oasis and then to desert,the values of TVDI and TFDI gradually increase,and the degree of agricultural drought gradually deepens,corresponding to the distribution of vegetation types in the area.Based on two datasets of vegetation indices(NDVI,EVI)and based on Sun-induced chlorophyll fluorescence(GOME-2,GOSIF),two sets of self-calibrating Palmer drought severity index(sc PDSI)datasets were extracted using Random Forest model,respectively.The results found that the multi-year average sc PDSIVI and sc PDSISIF values in Xinjiang were 0.52 and 0.66,respectively,which were in the normal range,and the results of both sc PDSI datasets indicated that the spatial trends of drought in Xinjiang were similar,and the change trends showed a small wetting trend in the currently relatively dry basin plain areas,while the currently relatively wet mountainous areas showed an arid trend.(3)The multi-year start of season(SOS)of Xinjiang grasslands was between 110 th to150th days,and the trend of SOS was-5.47 d/10 a,spatially 82.45% of the grasslands had an early trend of SOS;the end of season(EOS)was concentrated between 270 th and 290 th days,and the trend of EOS was 6.61 d/10 a,87.63% of the grasslands had a delayed trend of EOS;the length of season(LOS)was concentrated between 140 th and 160 th days,and the trend of LOS was 11.97 d/10 a,87.24% of the grasslands had an increasing trend of LOS.Climate change and drought have a significant impact on grassland phenology,in which the increase of temperature will lead to the overall advancement of grass SOS in the whole territory,especially the most significant impact of spring temperature,while the ground temperature has a similar effect on grass SOS,i.e.with higher ground temperatures causing an earlier SOS of grasslands.The increase precipitation in summer and autumn was an important factor in delaying the EOS of grasslands.In addition,the increase in the number of sunshine hours in spring will also promote the growth of grassland.Both meteorological drought and agricultural drought had inhibitory effects on grass growth,i.e.,the higher the degree of drought the worse the grass growth,among which agricultural drought(TVDI,TFDI)had the most significant effects on grass phenology. |
