摘要
Abstract
Based on 18-day observation data of global navigation satellite system(GNSS)buoys,the precipitable water vapor(PWV)in the atmosphere was retrieved,and the accuracy was verified.The relationship between GNSS PWV and actual precipitation was explored by introducing a random forest method,and the precipitation during rainfall was predicted based on PWV data.The results show that the difference between the GNSS PWV of buoys and the GNSS PWV of fixed stations is small.The average difference is less than 1 mm,and the standard deviation is 0.4 mm,which indicates that PWV inversion based on the GNSS buoys has high accuracy.The analysis of the relationship between GNSS PWV and actual precipitation shows that the PWV value will gradually increase before the occurrence of rainfall,and rainfall starts after the PWV value reaches the peak.PWV fluctuates during rainfall.After the end of the rainfall,the PWV value decreases significantly and gradually recovers to the normal level.The correlation between PWV and precipitation is as high as 65.5%during the concentrated rainfall period.The random forest algorithm is used to construct the precipitation prediction model,and the predicted results of the model are highly consistent with the actual observation results.When the precipitation exceeds 0.5 mm,the relative error of the predicted precipitation is less than 25%.The control variable method is used to analyze the importance of input characteristics,and the results show that PWV is much more important than relative humidity for marine precipitation prediction.关键词
可降水汽含量(PWV)/全球导航卫星系统(GNSS)浮标/降水量/随机森林Key words
precipitable water vapor(PWV)/global navigation satellite system(GNSS)buoy/precipitation/random forest分类
天文与地球科学
