FY-2C云分类资料估算日照百分率

日照百分率是地球辐射研究、太阳能资源评估等工作中使用的一个重要指标量。为探讨FY-2C静止气象卫星资料估算日照百分率的可能性,以江西省为例,采用了FY-2C静止气象卫星2007年整年白天逐小时云检测资料和同期江西省87个国家气象站逐小时日照时数资料,建立了北京时间8:00-17:00逐时多元线性回归和权重系数两种估算模型,估算得到了覆盖江西省范围的5 km空间分辨率的逐时日照百分率格点资料。另外,用反距离加权(IDW)和克里金法(Kriging)模型对全省8:00-17:00逐时的日照百分率进行5 km分辨率空间插值,以进行多种方法对比分析。将各模型的估算结果与未参与模型构建的17站实测资料进行对比分析后发现,基于FY-2C云检测资料的两种估算模型的平均绝对误差(MAE)和相对误差均比IDW和Kriging的小50%以上,且其均方根误差(RMSIE)均明显小于后两者,误差离散性亦更小。基于FY-2C云检测资料的日照百分率的两个估算模型中多元回归模型的平均绝对误差(MAE)为3.40%,平均相对误差为8.62%,略小于权重系数模型的3.47%和8.75%,两者的均方根误差(RMSIE)则相差无几,误差空间分布也较为一致;各时次两种模型的估算值多大于实测值,仅17:00的估算结果小于实测值,8:00和17:00的MAE、RMSIE、相对误差均明显大于其他时次。对试验结果进行综合分析得出结论:基于FY-2C云检测资料的两种日照百分率估算模型均明显优于传统的IDW和Kriging方法;就2007年资料而言,多元回归模型优于权重系数模型。

Estimating of sunshine percentage using the cloud classification data from FY-2C

The sunshine percentage is an important index in the radiation research and the solar energy assessment. Taking Jiangxi Province for example, we discussed the possibility to use the data from geostationary meteorological satellite to stimulate the sunshine hour distribution. First, we collected the cloud of FY-2C of 2007 and 87 meteorological stations sunshine hour data of Jiangxi Province. Second, we used multiple liner regression and weight coefficient methods to create two inversion models. Third, we acquired 5 km×5 km spatial resolution hourly percentage sunshine grid data of Jiangxi Province. By comparing the inversion results with observations data from 17 stations which were not involved in creating models, we found that the relative error and the Mean Absolute Error (MAE) of the two estimation models is smaller than Inverse Distance Weighted (IDW) and Kriging for more than 50%. The MAE and the relative error of multiple liners method were 3.40% and 8.62% smaller than weight coefficient method 3.47% and 8.75%. Both RMSE and error distributions were the same. The inversion data of both methods were bigger than the validation data except for the 17:00 sunshine hour data. The MAE, RMSIE and relative error of 8:00 and 17:00 are much higher than the other hours. Results based on the comprehensive analysis show that geostationary meteorological satellite data based on the FY-2C can be used in sunshine hour inversion, and the MAE of the two estimation models are much better than the traditional IDW and Kriging methods. According to data in 2007, the multiple liner model is better than weight coefficient model.