Accuracy Evaluation Of Different Remote Sensing Precipitation Products And Applicability Of Drought Monitoring In Central Asi

In recent years,drought has become increasingly severe in Central Asia,and climate warming has further exacerbated its occurrence.However,the monitoring of drought in this region is severely limited by problems such as data gaps due to the limited number and uneven distribution of ground observation stations.Remote sensing precipitation products have become one of the main sources of data for drought monitoring due to their wide coverage,high spatial resolution,and long time series.However,due to factors such as product algorithms,data distribution range,short time series,and complex underlying surfaces,the accuracy of various remote sensing precipitation products varies and their error characteristics are diverse,making them difficult to apply directly to drought monitoring.To meet the needs of drought monitoring in Central Asia,this thesis selected three remote sensing precipitation products: PERSIANN-CDR,PERSIANN-CCS-CDR,and MSWEP,and one mainstream reanalysis precipitation product,ERA5-Land,based on comprehensive consideration of factors such as coverage and time series length(>30years)in Central Asia.Based on ground station rainfall gauge observation precipitation data,the precipitation error and drought characterization ability of different precipitation products were evaluated,and it was found that different products have different error characteristics,making it difficult to effectively apply them to drought analysis.In order to solve the problem of error correction for remote sensing precipitation products,this thesis developed a grid-scale proportional coefficient precipitation error correction method based on spatial interpolation technology on the basis of the traditional station-scale proportional coefficient method and demonstrated its effectiveness by applying it to different precipitation products.Finally,using this correction algorithm,high-precision precipitation data covering Central Asia were generated,and drought monitoring research was carried out in this region.The results of the error evaluation and correction in this thesis can provide scientific references for upgrading the inversion algorithms of different remote sensing products and selecting user data.At the same time,the results of drought monitoring applications can provide scientific references for local water resource management.The main conclusions of this thesis are as follows:(1)In terms of precipitation inversion in Central Asia,different remote sensing precipitation products show significant differences in performance.Among them,MSWEP has the best precipitation capture ability,while the performance of CCS-CDR is the worst,with the lowest correlation,relatively high relative bias and root mean square error,and poor simulation accuracy of ground rainfall gauge data.(2)Different precipitation products have different abilities to characterize drought features.CDR has the highest correlation with ground rainfall gauge data,but there is a problem of overestimating the severity of drought.MSWEP and ERA5 have lower correlation,and MSWEP has the lowest relative bias.(3)The station-scale proportional coefficient method can effectively reduce errors in different precipitation products,but the correction effect varies depending on the product and region.Even after the proportional coefficient method is used for error correction,if the original product performs poorly,it is difficult to achieve the desired application accuracy.(4)The grid-scale proportional coefficient precipitation error correction method based on spatial interpolation technology can achieve good error correction,but different interpolation techniques can affect the correction effect of the proportional coefficient method,and the optimal interpolation method for the same remote sensing precipitation product varies between high and low altitude areas.The proportional coefficient method based on co-kriging can effectively improve the accuracy of precipitation products in high altitude areas.(5)After error correction,MSWEP has the best overall performance.Drought monitoring analysis based on the corrected MSWEP precipitation product found that Central Asia has shown a slightly significant trend of becoming wetter in the past 40 years,with 5-10 year dry-wet fluctuations.(6)Between 1983 and 2022,there were 10 large-scale drought events in Central Asia,among which the drought event from August to December 1997 was the longestlasting(5 months)and most severe(8.55)drought event.Comparing the characteristics of different drought events,it was found that the duration and severity of drought can to a certain extent affect the severity of drought events.