综合分析法在复杂地形气温精细格点化中的应用

Application of A Comprehensive Analysis Method on Hourly Surface Air Temperature Interpolation over A Complex Terrain Region

基于高分辨数值模式预报、数字高程模型和区域自动站逐时气温观测资料,以兼具平原、丘陵、盆地、山地、海岛和海洋等复杂地形的浙江省为研究区域,建立了适用于该地区的综合分析方法,并将该方法与反距离加权插值法、普通克里格插值法和考虑高程的反距离加权插值法进行比较。结果表明:综合分析法能给出最优的格点化温度分析场,考虑高程的反距离加权法其次,最后是反距离加权法和普通克里格法。考虑高程的反距离加权插值法和综合分析法对高海拔无测站区域气温的重建有较为明显的优势。综合分析法获取的温度分析场在山地、海岛和海洋等资料稀少或缺测的复杂地形区域仍旧能保持合理的空间分布状态,对复杂地形资料稀少或缺测地区的插补能力有突出优势。温度资料格点化精度随站点数增加而提高,且当站点数达到一定数值后,精度趋于稳定;综合分析法在站点数达到1000个(站点平均间距约16 km)后就趋于稳定,而其余三种方法在站点数增加到1600个(站点平均间距约13 km)后趋于稳定。

Generating accurate and fine gridded meteorological data becomes one of the final goals in meteoro- logical modernization, which is the key fact in doing researches in meteorology, hydrology, and ecology scienc- es. How to convert the information from local observations into the whole region is getting more and more atten- tion from scientists in meteorology, hydrology, geography, ecology and so on. In practical application, interpo- lation method is usually used to achieve the grid data of the problem. To obtain accurate and fine gridded meteo- rology information, a new interpolation method named Comprehensive Analysis method （IN） is used to repro- duce the gridded 2 m temperature dataset based on the high resolution numerical prediction, digital elevation model, and surface observation data over a complex terrain region. This method divides the model prediction er- ror into the error of modeling the weather system and the error of describing the topography, and then uses the observation data to revise these two parts separately. Four steps are used in the revising： Firstly, the numerical model predicting temperature is used as the first guess field in the reanalysis grid system; Secondly, the first guess value and the corresponding error are obtained on the observation site; Thirdly, this error is interpolated to the reanalysis grid system; Finally, the gridded error is added into the first guess field to obtain the final tempera- ture field. The inter-comparison is mode among the Inverse Distance Weighting （IW）, Ordinary Kriging （KR）, Gradient Inverse Distance Weighting （IG） and IN based on the hourly observed scattered station 2 m temperature data over Zhejiang province and its surround area. The results show that the IN method produces the best result, follows the IG, IW, and KR. The IG and IN have greater abilities on reproduce the temperature over high alti- tude area than the IW and KR. The IN can produce a normal and reasonable temperature field over the no data ar- ea, which shows a stronger ability on no data interpolation. The interpolation accuracy increases with the number of sites increase, and no improvements appear when the number of sites reaches 1000 in IN and 1600 in other three methods.