摘要
海表温度(Sea Surface Temperature,SST)是干旱预报的主要因子之一,传统预报模型主要采用固定海域SST(如ENSO),未从全球大范围角度搜寻可利用的SST信号。通过回归分析筛选全球具有预报意义的SST区域,结合随机森林算法构建了一种气象干旱预报新模型,以珠江流域为例进行应用检验。结果表明:①该模型可有效预报干旱的时空变化规律,且预见期越长,预报精度相应下降;②非汛期干旱预报准确率高于汛期,沿海地区比内陆地区有更好的预报效果;③珠江流域干旱发生可能与典型的气候波动有关,如厄尔尼诺南方振荡和北大西洋振荡。
Sea surface temperature(SST)is one of the main factors for drought forecasting.Conventional forecasting models mainly use SST from fixed sea areas(e.g.,ENSO),without searching for available SST signals from a global large-scale perspective.Combining with the random forest algorithm,this paper constructs a new meteorological drought forecasting model through regression analysis to screen global SST areas of forecasting significance and takes the Pearl River Basin as an example for application tests.The results are as follows.①The model can effectively forecast the temporal and spatial evolution of drought,and as the forecast period becomes longer,the forecast accuracy decreases accordingly.②The accuracy of drought forecast is higher in non-flood season than in flood season,and the coastal area has a better forecast effect than the inland area.③The occurrence of droughts in the Pearl River Basin may be related to typical climate fluctuations,such as the El Niño Southern Oscillation and the North Atlantic Oscillation.
作者
冯鑫
刘艳菊
童宏福
钱姝妮
FENG Xin;LIU Yanju;TONG Hongfu;QIAN Shuni(School of Civil Engineering and Transportation,The South China University of Technology,Guangzhou 510640,China;Comprehensive Technology Center of Pearl River Water Resources Commission of Ministry of Water Resources,Guangzhou 510611,China;Qingyuan Hydrology Sub-bureau of Guangdong Province,Qingyuan 511599,China;Guangzhou Water Planning Survey and Design Institute Co.,Ltd.,Guangzhou 510665,China)
出处
《人民珠江》
2024年第5期96-102,共7页
Pearl River
基金
国家自然科学基金(52109019)
广东省自然科学基金(2023A1515030191)。
关键词
干旱预报
海表温度
随机森林
珠江流域
drought predication
sea surface temperature
random forest
Pearl River Basin
