基于累积分布函数的统计降尺度模型校验方法适用性研究

Applied Research of Calibration Method for SDSM Model Based on Cumulative Distribution Function

为更精确的为区域气候模拟和预估研究提供参考,开展了基于累积分布函数的统计降尺度模型校验,在传统统计降尺度模型的基础上,使用基于累积分布函数的校验方法校正了SDSM模型预估的A2和B2情景下中国265个站点1961~2099年逐日温度数据,校正后A2情景下,观测值和模拟值R2达到0.9999以上的比例占到85%,达到1的占6%;B2情景下,观测值和模拟值R2达到0.9999以上的比例占到87%,达到1的占19%;斜率值接近1的站点分别增加了57%和51%;截距接近0的站点分别增加了31%和16%。校正后的模型能更好地预估出未来逐年稳定通过0℃的日期,也即生长季开始的日期。

For more accurate regional climate prediction to provide the reference,Carried out statistical downscaling model validation based on the cumulative distribution functions（ CDFs）,on the basis of traditional statistical downscaling model,using the SDSM based on the cumulative distribution functions model calibration predicted the daily temperature at 265 sites under the A2 and B2 scenarios from 1961 to 2099 in China. The result showed that the proportion of observation and simulation value R2 was above 0.9999 accounted for 85% under A2 scenarios after correction,R2 was 1 accounted for 6%; the proportion of observation and simulation value R2 was above 0.9999 accounted for 87% under B2 scenarios after correction,R2 was 1 accounted for 19%;The sites of the slope value close to 1 increased by 57% and 51% respectively; The sites of the intercept is close to 0 increased by31% and 16% respectively. The model after correction could better predict steadily through the date of the 0 ℃ in future（ the accurate date of the beginning of the growing season）.