CollectHomepage AdvertisementContact usMessage

干旱区地理 ›› 2018, Vol. 41 ›› Issue (4): 809-816.doi: 10.12118/j.issn.1000-6060.2018.04.17

Previous Articles     Next Articles

Evaluation of applicability of satellite-based RFE2.0 rainfall data in Xinjiang

LIU Hai-jun1,2, CHEN Ya-ning2, XU Chang-chun1   

  1. 1. College of Resources and Environment Science/Key Laboratory of Oasis Ecology, Xinjiang University, Urumqi 830046, Xinjiang, China;
    2. State Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, Xinjiang, China
  • Received:2017-12-02 Revised:2018-03-26 Online:2018-07-25

Abstract: With the daily precipitation data from 65 meteorological stations in Xinjiang,China,the verification methods including continuous statistics and classification statistics were used to evaluate the applicability of satellite-based RFE2.0(Rainfall estimation2.0)data.The results showed that RFE2.0 overestimated the daily precipitation in Xinjiang, with an average value of 0.4 mm MBE (mean bias error),more than 70% meteorological stations being among 0.5 mm.The average value of correlation coefficient R between RFE2.0 and observation data was 0.4,indicating a low correlation with a descending order of the correlation value from north Xinjiang to east Xinjiang and then to south Xinjiang.The MAE(mean absolute error) was 0.23-4.7 mm,with 1.16 mm average value.More than 50% meteorological stations with MAE were smaller than 1mm.It was the northern,southern,eastern Xinjiang according to the order of relevance from big to small.The RMSE (root mean square error) was 1.03-9.7 mm,with 3.32 mm average value.The descending order list was the north Xinjiang,south Xinjiang,east Xinjiang according to the MAE value.From the perspective of deviation from the real value,the simulated values in east Xinjiang was close to the observed value.There were 45 non-GTS stations selected to verify the possibility of precipitation events through classification statistics. FBI (frequency bias index) in Xinjiang was 0.82-3.71,the average value of FBI in northern,southern and eastern Xinjiang were 1.42,2.1,2.53.It was eastern,southern and northern Xinjiang according to size of FBI.In genenal,RFE2.0 overvalued possibility of rainfall events in Xinjiang,overestimation of precipitation events in northern Xinjiang is lower than the average of the whole Xinjiang.The POD value was 0.12-0.95,with average value 0.67.The average value of POD was 0.74, 0.61, 0.61,respectively from northern,eastern and southern Xinjiang. POD in northern Xinjiang was higher than that of east and southern Xinjiang.The FAR was 0.18-0.92,with 0.58 average value and 33% stations FAR less than 0.5.The average value of FAR in northern Xinjiang,eastern Xinjiang and southern Xinjiang was 0.43,0.74,0.69,respectively.The value of FAR in northern Xinjiang was smaller than eastern and southern Xinjiang.Two stations were selected to analyze respectively,Tuergate and Yuli observation stations from the southern Xinjiang,Caijiahu and Zhaosu observation stations from the northern Xinjiang,Shanshan and Yiwu observation stations from east Xinjiang,to verify the application of RFE2.0.The above rule can provide scientific basis for application of RFE2.0 in Xinjiang.

Key words: rainfall, RFE2.0, validation, Xinjiang

CLC Number: 

  • TP79