基于大气环流因子的西南地区干旱预测模型及应用

Study of Drought Forecasting Models in Southwest China Based on Circulation Parameters

为了有效识别西南地区干旱发生的前兆环流信号,基于标准化降水指数（SPI）和环流特征量月值数据,考虑致旱因子与区域干旱的非线性关系,对满足不同线型的非线性因子作线性化处理,筛选出通过显著相关检验的预报因子,基于多元线性回归方法建立了四种西南地区秋季干旱预测模型,比较了四种模型的估计值与西南地区实际干旱指数值。结果表明,在1961-2004年的模型率定期,四种回归模型均通过了显著性水平为0.01的F检验;非线性模型对2005-2012年验证期的干旱试报成功率高于线性模型,且成功预测了2005、2007、2009年的三场西南地区秋季干旱,预见期长达6个月,对数模型对西南秋季干旱预测效果较其他模型更稳定;欧亚及亚洲经向环流、大西洋及北非副高、北半球极涡和西藏高原位势高度等均为西南致旱的显著强信号。非线性因子处理方法适用于西南地区干旱预测建模,且具有潜在应用前景。

Considering the nonlinear relationship between regional drought and drought factors, based on the standardized precipitation index （SPI） and circulation parameters, linearization was used to process the nonlinear parameters, which met different linetypes. Then, parameters passed the significant correlation test were picked out. Finally, four kinds of autumn droughts forecasting models for Southwest China were established with the multi-linear regression method. The results show that in the calibration period from 1961 to 2004, all models pass the F-test with 0.01 significance level and the simulation effect of the exponential type regression model is the best; nonlinear models successfully forecast the autumn droughts in Southwest China in 2005, 2007, 2009 and they have better performance than linear models in the validation period from 2005 to 2012, with a forecast period of 6 months; meridional circulation of Eurasia and Asia, Atlantic and North Africa subtropical high, Northern Hemisphere polar vortex and geopotential height of Tibet plateau are the strong signal for causing drought in Southwest China.＇ The nonlinear method for predictors is suitable for drought forecasting in Southwest China, which has potential practical application in prospect.