效率系数和一致性指数及其在干旱预测精度评价中的应用

基于关中平原2003—2014年Aqua-MODIS数据的条件植被温度指数的干旱监测结果,引入效率系数和一致性指数对自回归求和移动平均(ARIMA)模型和季节性ARIMA模型的预测精度进行了评价,并分析了其在干旱预测精度评价中的适用性。两种干旱预测模型的精度评价结果为ARIMA模型与季节性ARIMA模型的效率系数分别为-0.04与-4.27,一致性指数分别为0.40与0.37,表明ARIMA模型对干旱的预测精度高于季节性ARIMA模型,这与均方根误差、皮尔森相关系数及Kappa系数等的评价结果一致,且效率系数对不同干旱预测模型的预测性能的区分效果尤为显著,对干旱预测模型的精度评价比一致性指数更为合理。因此,效率系数更适合用于遥感干旱预测模型的精度评价。

Application of the coefficient of efficiency and index of agreement on accuracy assessment of drought forecasting models

The coefficient of efficiency and index of agreement are usually used to evaluate the accuracy of hydrological models. Based on the drought monitoring results from the vegetation temperature index using Aqua-MODIS data from 2003 to 2014 in the Guanzhong Plain, China, the coefficient of efficiency and index of agreement were employed to assess the precision accuracy of two drought forecasting models, the Autoregressive Integrated Moving Average (ARIMA) and Seasonal ARIMA (SARIMA) models, and to analyze the performance of these models as well. The results showed that the coefficients of efficiency of the two models were -0.04 and -4.27, and the values of the index of agreement of the two models were 0.40 and 0.37, respectively, indicating that the performance of the ARIMA model was better than that of the SARIMA model. These results were in good agreement with the ones based on root mean square error, Pearson''s correlation coefficient and Kappa coefficient. Particularly, the assessments of the coefficient of efficiency were reasonable and remarkable than those of the index of agreement. Therefore, the coefficient of efficiency was suitable for the accuracy assessment of the drought forecasting models.