缺资料地区降水系列的插补及验证

姬世保 杜军凯 仇亚琴 刘欢 吕向林

摘 要：针对部分地区降水资料缺乏的问题，以德清县为研究对象，提出了日尺度降水资料的插补方法，该方法通过对已有降水数据进行时间尺度转换和空间插值的方式插补出缺资料地区日降水数据。结果表明：①德清站插补结果中日降水强度、强降水量及连续有雨天数的误差均小于3%，与直接用周边站点的逐日数据对数据缺失站进行插值相比，各站月降水数据与实测值的相关系数平均提高0.11，均方根误差降低了42.3%;②将降水插补结果作为分布式水文模型的降水输入时，径流模拟效果得到了有效改善，1960—2018年模拟径流量和实测径流量系列的纳什效率系数由0.62提高到0.87，相对误差由-4.2%降至-2.3%;③日尺度降水插补结果的相关系数、5 d最大降水量及强降水量对研究区分布式水文模拟效果影响较大。

关键词：降水插补;缺资料地区;降水极端事件;交叉验证;WEP-L模型;分布式水文模型

中图分类号：TV214 文献标志码：A

doi：10.3969/j.issn.1000-1379.2021.05.008

Abstract： Aiming at areas which have no sufficient precipitation data， an interpolation method of daily scale precipitation data was proposed. This method could obtain the daily precipitation in some deficient data areas by means of conversion of time scale and space transplantation of existing data. The results show that a） the error between the simple daily intensity， very wet days and consecutive wet days of the interpolation result of Deqing Station and the measured series is less than 3%， compared with the daily data of interpolation result which gained by surrounding stations directly， the correlation coefficient between the monthly precipitation of each station and the measured value is increased by 0.11 on average， and the RMSE is reduced by 42.3%; b） when the interpolation results are used as the input in distributed hydrological simulation， the runoff simulation effect has been effectively improved， the Nash coefficient is increased from 0.62 to 0.87， and the relative error is reduced from -4.2% to -2.3% during the period of 1960-2018 and; c） the correlation coefficient， the 5-day maximum precipitation and very wet days of daily interpolation result have a great influence to the distributed hydrological simulation in the study area.

Key words： precipitation interpolation; data-deficient regions; precipitation extreme events; cross-validation; WEP-L model; distributed hydrological simulation

1 前 言

降水数据对水涵养评价、灾害风险管理、水循环模拟、植被分布和生态演变等研究具有重要價值[1-3]。现阶段，我国气象局管理的国家级气象站基本保持在相对稳定的2 200～2 400个[4-5]，平均密度为4 000～4 363 km2/站。胡庆芳[6]在赣江流域证实，当气象站网密度低于1 300 km2/站时，降水空间估计精度随站网密度变化而急剧变化，因此相对于全国来讲，测站不足、分布不均的问题仍然存在。另外，受自然条件及人为影响，水文站及雨量站停测、缺测、漏测现象时有发生[7]。……