CLIGEN生成干旱半干旱地区降水相关参数的验证

选用黄河上中游地区无定河流域为中心的15个气象站1959~1999年的降水日值资料,对随机天气发生器CLIGEN在干旱半干旱地区再现降水的能力进行了验证。结果表明:CLIGEN模型较好地模拟了该区域的降水发生概率;很好地再现了年、月、日降水总量平均值,平均相对偏差分别为2.4%、2.4%和2.1%;CLIGEN再现了96.4%的日降水变率、95.9%的月降水变率和84.1%的年降水变率。对年降水变率估计稍差,表明CLIGEN在模拟降水变率方面还有改进的必要。从降水极值看,年降水最大值的平均相对偏差为11.1%,偏差最大的是干旱区的临河站（39.1%）;年降水最小值的平均相对偏差为20.5%,偏差最大的是临河站（-30.7%）;月最大降水量除两站稍低外,其它站平均偏高20.2%;日降水最大值除临河站偏低3.4%外,其余各站平均偏高43.2%。总体上讲,CLIGEN在干旱地区的模拟能力比在半干旱区稍差。鉴于CLI-GEN模拟的极大值绝大部分都偏高,因此利用CLIGEN模型生成的降水资料运行径流和土壤侵蚀模型有高估径流量和土壤侵蚀量的可能,需要进一步利用自计雨量计的资料对CLIGEN生成次降水的参数进行验证,以确保径流和土壤侵蚀预测的精度。

Evaluation of CLIGEN Precipitation Parameters in the Semiarid and Arid Regions of the Yellow River Basin

Precipitation records from 15 weather stations in the arid and semiarid regions in the Yellow River Basin of China were used to validate the CLIGEN weather generator.Daily records of 41 years from 15 stations were used to evaluate the generator.Generally,the performance of the CLIGEN generator is better in semiarid regions than in the arid regions.Results show that the generator was successful in modeling the means of the total of the annual,monthly and daily precipitation,the monthly probabilities of wet and dry days,and the variability of daily,monthly and annual precipitation.Mean absolute relative errors for simulating daily,monthly and annual precipitation across 15 stations were 2.1%,2.4% and 2.4% for the means and 3.6%,4.1% and 15.9% for the standard deviations,respectively.The relative error for the standard deviation of annual precipitation was relative high.Thus,the improvements of precipitation occurrence are expected.Mean absolute relative errors for the all-time maxima of daily,monthly and yearly precipitation were 5.2%,17.3% and 11.1%,respectively.Most of the maxima values across these stations were overestimated.It may lead to overestimation of runoff and sediment yield by the soil erosion models such as WEPP using the precipitation patterns generated by the CLIGEN model.The validation of the parameters of storm events such as storm duration and peak rain density are needed to conduct by using the pluviograph data.