一种基于GEP的地面气温观测资料的质量控制方法

A GEP-BASED METHOD FOR QUALITY CONTROL OF SURFACE TEMPERATURE OBSERVATIONS

地面观测资料的单站质量控制是气象业务中不可或缺的环节,特别在台站密度较低或新增台站的地区,单站质量控制是观测资料有效性的基本保障。在研究传统单站质量控制方法的基础上,利用气温与相对湿度之间的耦合关系,提出一种基于基因表达式编程(GEP)算法的地面逐时气温资料单站质量控制方法,实现相对湿度对气温的质量控制。并利用该方法对江苏省六市2005—2007年地面观测的逐时气温资料进行质量控制,以评估其适用性,并将控制效果与传统方法进行比较。结果表明,GEP质量控制方法能有效标记出可疑数据,具有可控性强、识别精度高等优点,尤其在每年5—9月的检验效果优于传统的单站质量控制方法。

Quality control （QC） of single-station surface observations is an indispensable step in routine meteorological work, especially in areas where observation density is small or there are newly setup stations, as it is a basic guarantee to the validity of meteorological observations. By studying traditional QC methods, a new QC technique, called Gene Expression Programming, is introduced in detail. Based on the coupling relationship between temperature and relative humidity, this method runs QC of temperature with relative humidity. In order to assess the suitability of the method, surface temperature observations from six cities of Jiangsu province from 2005-2007 are used. According to the test, this method can flag suspicious data effectively and it also has the advantages of strong controllability and high accuracy. Specifically, this method is superior to traditional methods from May to September.