重庆山地区域气象要素空间插值方法对比

利用重庆地区1999年和2018年气象数据, 分别采用薄盘光滑样条、协同克里金、普通克里金、反距离加权4种方法, 从年和月两种尺度对气温、降水、太阳总辐射三个要素进行空间插值; 采取交叉验证方法, 用MAE、MRE、RMSE评估插值精度, 确定各要素最优插值方法。结果表明: 气温和太阳总辐射最优插值方法为薄盘光滑样条, 降水为反距离加权; 插值精度上气温、太阳总辐射高值月份优于低值月份, 降水则相反, 但三个要素均表现出年尺度优于月尺度。MRE检验表明, 插值精度为气温>太阳总辐射>降水, 1999年年尺度插值精度分别为1.86%、4.60%、6.87%, 月尺度插值精度分别为2.79%、5.82%、17.42%;2018年太阳总辐射年、月尺度插值精度分别为3.03%、4.88%, 区域站加密后气温、降水年尺度插值精度分别为2.03%、11.20%, 月尺度对应插值精度分别为3.20%、23.14%。

Comparison of spatial interpolation methods of meteorological elements over Chongqing mountainous region

Mountainous terrain is one typical and complex land surface, and the accurate simulation and acquisition of meteorological elements over mountainous regions are facing challenges due to a limited number of meteorological stations. Based on the meteorological data over the studied area in 1999 and 2018, we used four methods of thin-plate smoothing splines (ANUS), Co-Kriging (CK), ordinary Kriging (OK), and inverse distance weighting (IDW) to spatially interpolate air temperature, precipitation, and total solar radiation on annual and monthly scales. Using a cross-validation method, mean absolute error (MAE), magnitude of relative error (MRE), and root mean square error (RMSE) are used to evaluate the interpolation accuracy and determine the optimal interpolation method for each meteorological element. The results showed that ANUS is the optimal interpolation method for air temperature and total solar radiation, while IDW is the optimal interpolation method for precipitation. The interpolation accuracy for air temperature and total solar radiation is better during the months with high air temperature and total solar radiation than that during months with their low values, and the trend is opposite for the precipitation. The interpolation accuracy for the three elements is better on an annual scale than on a monthly scale. The MRE values showed that the interpolation accuracy for the three elements is in order of air temperature > total solar radiation > precipitation, being 1.86%, 4.6%, and 6.87% on an annual scale in 1999 and being 2.79%, 5.82%, and 17.42% on a monthly scale, respectively. In 2018, the interpolation accuracy for total solar radiation is 3.03% and 4.88% on the annual and monthly scales, respectively. After using data at regional encryption stations, the interpolation accuracy for air temperature and precipitation can reach 2.03% and 11.2% on the annual scale and 3.2% and 23.14% on the monthly scale, respectively. Our research can provide scientific reference and a basis for the spatialization of meteorological elements in similar complex terrain areas.