Author(s): Zuoqiang Han; Ruipeng Song; Yiqi Yan; Lu Lu
Linked Author(s): Yiqi Yan
Keywords: Random forest Prediction Extremeprecipitation Midsummer the Yellow River Basin
Abstract: In this paper, we define the summer precipitation extremes in different regions of the Yellow River Basin, and construct a random forest decision tree model for typical regions. The Nino3.4 SST, the intensity, the position of ridge line and the westward ridge point of subtropical high from July to August were used as forecasting factors to study the decision analysis of different factors on the midsummer precipitation in above Lanzhou and Sanhua regions. The results show that the uncertainty of summer precipitation in different regions can be well interpreted by the threshold of extreme precipitation in different regions defined by percentile method and probability density function. With Nino3.4 SST and subtropical high index as the forecasting factors, we have good forecasting skills for the study area, and the decision tree gives the decision rules of extreme precipitation. For the Lanzhou above region, the probability of extreme precipitation is greater when the ridge line of subtropical high is north in July and the ridge point of subtropical high extending west is east in July-August. For the Sanhua region, when the El nino event occurs, the precipitation in the Sanhua region is extremely low when the subtropical high is obviously northward in the midsummer. When the Nino3.4 SST is normal or cooler, the subtropical high is normal or slightly stronger, and the ridge is northward, it is conducive to the extreme precipitation in San-Hua region.
Year: 2025