喀斯特地区月均降水协克里金插值方法研究——以贵州省为例  

On the association of Co-Kriging interpolation method research based on GIS:A case study in Karst area of Guizhou Province

作　　者：闫星光[1] 吴琳娜[2] 周涌[2] 宋具兰 邓仕雄 

机构地区：[1]贵州大学矿业学院,贵州贵阳550025 [2]贵州大学资源与环境工程学院,贵州贵阳550025

出　　处：《云南大学学报（自然科学版）》2017年第3期432-439,共8页Journal of Yunnan University(Natural Sciences Edition)

基　　金：贵州大学研究生教育创新基地建设项目(贵大研CXJD[2014]002)

摘　　要：喀斯特山区地形复杂,地势起伏大,降水量时空分布不均匀,尤其是丰水期降水量的分布直接影响当地经济作物的生长,也是地质灾害发生的诱因.以贵州省77个气象站点30 a(1981—2010年)丰水期月均降雨量为基础数据,分析了地形因素(海拔、坡度和坡向)和气象因素(站点压强及相对湿度)与贵州省降水的相关性,并对4种协克里金插值模型方法进行了对比研究.结果表明:采用Pearson相关性分析得出坡向与研究区降水相关性最强,相关系数为0.998.综合对比不同协克里金半变异函数模型(稳定模型、指数模型、球面模型和高斯模型)预测值和实测值的结果表明球面模型的偏差均值最小(MAE=-0.000 4),一致性系数最优(RMSE=0.864).采用球面模型的协克里金插值是进行贵州省降水插值的最好方法,这为更有效地识别出喀斯特地区丰水期降水空间分布提供基础.The geographical condition of Karst mountainous areas is complicated, which, together with the in- fluence of obvious peak clusters ,causes an unbalance of the spatial and temporal distribution of rainfalls.In parti- cular, the distribution of precipitation in wet season has a direct impact on local economic crop growth and the oc- currence of geological disasters.Based on the 77 weather stations in Guizhou within 30 years （ 1981--2010）, and on the data of monthly precipitation in analysis of the terrain factors （ elevation, gradient, slope direction） and meteorological factors （ site pressure, relative humidity, sea level pressure） and the correlation of rainfall in Karst region, Guizhou Province, four Kriging interpolation methods have been explored in light of a comparative study. The results show that firstly, when using Pearson correlation analysis correlation between precipitation and various factors,respectively, the slope is 0.998, and the slope is of strongest correlation with precipitation. Secondly, through the association and the different haft the variation function model （ stable model, index model, spherical model and gaussian model） contrast, it is found that the mean difference between spherical model is minimum （MAE =-0.000 4）, and the consistency coefficient of optimal is best（ RMSE = 0.864）.Results from comprehen- sive comparisons of different model predicted values and measured values show that the best way to carry out in- terpolation of precipitation in Guizhou Karst area is applying Co-Kriging interpolation with semi variation functio- ning as the spherical model, which can help enhance the efficiency in identifying the spatial distribution of preoc- cupation in Karst area during wet season. ysis

关 键 词：喀斯特山区 丰水期 协克里金插值 Pearson相关性分析 

分 类 号：P332.1[天文地球—水文科学]