热带印度洋-太平洋地区海表温度异常的线性转置预报模型

Linear Inverse Modeling for Predicting the SSTA of Tropical Indian-Pacific

采用 COADS的月平均海平面温度距平 ( SSTA)资料 ,建立了预报热带印度洋-太平洋 SSTA的线性转置模型 ( L IM)。经检验 ,对于非独立和独立样本 ,预报的均方根误差分别在 1 2个月和 1 0个月预报时效内小于 SSTA的均方差 ,相对误差在 5个月预报时效内都小于 5 0 %。在 El Nino和 La Nina时段效果优于其他时段 ,其中La Nina时段又更好些。对同一地区的 SSTA,L IM预报优于自回归模型预报、奇异谱 -最大熵预报、气候预报和持续性预报。

Based on monthly COADS SSTA dataset,linear inverse modeling (LIM) is developed to predict SSTA over tropical Indian Pacific Oceans.Prediction results of independent and dependent samples show that the RMS errors of SSTA predicted in 10~12 months ahead are smaller than SSTA mean square deviation and that the relative errors in 5 months ahead are less than 50 %.The prediction accuracy is btter during El Nino and La Nina periods than during other periods,especially during La Nina periods.The prediction accuracy of LIM is better than that of auto regression model,SSA MEM model,climate value prediction and persistence prediction.