Precipitation Estimation In The Tianshan Mountains By Integrating Remote Sensing Products And Ground Data

Object:The precipitation is affected by multiple factors,such as atmospheric movement,land-sea position and underlying surface conditions,and has complex spatial variability.On the macro scale,the precipitation complies with certain zonal distribution law,but on the micro level,there are local differences.Quantitative estimation of spatial distribution of precipitation has long been a hot topic in the field of hydrometeorology.In the context of global warming,precipitation fluctuation is aggravated,and drought disasters are frequent in the Tianshan Mountains which is far away from the ocean.Understanding the precipitation distribution and drought characteristics in the Tianshan Mountains with complex terrain and sparse meteorological stations requires high-precision spatial precipitation data to support.Methods:Based on 36 meteorological stations in the Tianshan Mountains,statistical indexes（CC,BIAS,SDR）and classified indexes（POD,FAR,CSI）were selected in this study to evaluate the applicability of TRMM,GPM and MSWEP in different time scales and spatial subregions.And the accuracy difference caused by terrain was analyzed.Based on the comprehensive evaluation results,the spatial precipitation estimation model was constructed based on GWR,MGWR model and interpolation method,the spatial distribution of precipitation in long time series（1979-2020）with a resolution of 1km×1km was obtained by integrating ground data.Finally,based on the estimation results,the temporal and spatial variation characteristics of precipitation and drought events in the Tianshan Mountains were analyzed in detail.Results:（1）On the time scale,the inversion ability of three remote sensing products on the precipitation of rainy season（CC>0.7）was stronger than that in the dry season（CC>0.5）,the detection ability of monthly precipitation is generally poor(CC_TRMM<0.5<CC_MSWEP<0.6<CC_GPM<0.7);On the spatial scale,the differentiation characteristics of MSWEP and the Tianshan Mountain precipitation were closest（SDR>0.75）.The inversion performance of remote sensing products on precipitation in mountainous areas was not only related to topographic factors such as elevation and slope,but also to meteorological factors such as temperature and sunshine duration,which indirectly affected the performance of products by changing the underlying surface cover type.（2）Compared with the remote sensing products,the fusion results of GWR improved the underestimation of the measured monthly precipitation by GPM(BIAS_GPM=-4.83%,BIAS_{g GPM}=-2.65%).The fusion results of MGWR significantly improved the correlation between MSWEP and measured precipitation(CC_MSWEP=0.55,CC_{m MSWEP}=0.93).In general,m MSWEP had the best fit with the spatial distribution of actual precipitation in the Tianshan Mountains（CC>0.9,-2%<BIAS<+2%）,while preserving the spatial pattern of precipitation in different subregions,it showed a richer and clearer spatial texture of topographic precipitation.（3）From 1979 to 2020,the annual precipitation in the Tianshan Mountains showed an increasing trend,and the annual average precipitation was 268.42mm.From the perspective of spatial scale,the precipitation in the Tianshan Mountains generally presents a changing trend of more in the West and less in the East and more in the North and less in the South.The average annual precipitation in the West and East Tianshan Mountains increased almost linearly with the increase of altitude.Precipitation in North Tianshan increased with the increase of altitude,and the maximum precipitation height zone appeared at around 2749m.The precipitation in the South Tianshan Mountains fluctuated with the increase of altitude,and there are two maximum precipitation zones at 3249m and 4739m,respectively.In the past 40 years,the total amount and frequency of extreme precipitation in the Tianshan Mountains increased and decreased respectively before and after 2000.（4）In recent 40 years,the dry and wet changes in the Tianshan Mountains showed a wetness trend.Among them,71%of the areas showed a significant trend of humidification,15%of the areas showed a slight trend of humidification,only 9%of the areas showed a significant trend of dry,mainly concentrated in the East Tianshan Mountains.From the perspective of different subregions,the West Tianshan has the characteristics of low frequency,long time,and high severity of drought,while the East Tianshan presented high frequency,short time and low severity of drought.Occasional short-term high-severity drought events occurred in North and South subregions.There were 16 drought events in the Tianshan Mountains,among which the drought event from January 1983 to March 1986 was the most severe（the severity reached 27.11）and the longest（39 months）.Conclusion:The correlation between the precipitation estimation results obtained in this study and the measured monthly precipitation was more than 0.9,and the deviation was controlled within±2.0%,which could provide a potential reference scheme for the realization of the spatio-temporal monitoring of drought characteristics in the northwestern mountainous areas of China.