Improving daily precipitation estimates for the Qinghai‐Tibetan plateau based on environmental similarity

Due to the scarcity of gauge observations and inaccuracy of satellite estimation, obtaining reliable daily precipitation estimates over the Qinghai‐Tibetan Plateau (QTP) remains challenging. In this article, an integrated scheme is developed based on the assumption that in a specific climatic region, the similarity of environmental conditions related to precipitation (SEP) in two locations is positively correlated to the similarities of occurrence and magnitude of precipitation between them. First, the QTP was divided into the northwestern arid, middle semi‐arid/semi‐humid, and southeastern humid climatic sub‐regions by grouping analysis. Second, based on modified weighted k‐nearest neighbour model, daily precipitation of target locations in these climatic sub‐regions were predicted by weighted regression of a group of gauge observations that have the largest SEP with the target locations. SEP was calculated by the following auxiliary environmental factors: longitude, latitude, elevation, Normalized Difference Vegetation Index, relative humidity, and CMORPH (Climate Prediction Center's morphing technique) daily precipitation estimates (original CMORPH). The validation results demonstrate the effectiveness of the proposed scheme. Compared with the original CMORPH and PDF‐calibrated CMORPH (CMORPH calibrated by probability density function matching plus optimal interpolation method) at daily, monthly, and yearly scales, the scheme improves the rain/no rain detection capacity and the accuracy of daily precipitation estimates. In addition, the daily precipitation estimates obtained from this scheme can present significant discrimination over specific geographic units, particularly the Qaidam Basin, the great bend of the Brahmaputra River, and Hengduan Mountain. Daily precipitation estimates derived from remote sensing or fused by gauge observation and remote sensing products are inaccurate over the Qinghai‐Tibet plateau (QTP) due to the low quality of remote sensing and the scarcity of gauge observations. Calibration model using auxiliary environmental factors, like regression models, can only be effective at monthly scale. Using a weighted k‐nearest neighbour model revised by the similarity of environmental conditions, an integrated scheme developed here can improve the daily precipitation estimates over the QTP.