利用GIS和多变量分析估算青藏高原月降水  

Mapping monthly precipitation for Tibetan Plateau with GIS and multivariate analysis based on DEM data

作　　者：何红艳[1] 郭志华[1] 肖文发[1] 郭泉水[1] 

机构地区：[1]中国林科院森林生态环境与保护研究所,北京100091

出　　处：《生态学报》2005年第11期2933-2938,共6页Acta Ecologica Sinica

基　　金：国家自然科学基金重大研究计划资助项目(90211006);国家重点基础研究发展规划资助项目(973项目)(2002CB412508);国家自然科学基金面上资助项目(30371141);国家林业局重点试验室开放基金资助项目~~

摘　　要：随着空间降水信息需求的日益增加,空间降水插值已被广泛应用。降水区域不同,插值方法不同;时间尺度不同,插值方法也不相同。适合于所有地区的通用降水插值模型是不存在的。青藏高原自然地理特征独特,分析高原降水的时空格局意义重要。以青藏高原及其周边地区140个气象站点的月降水信息及其该地区的数字高程数据(DEM)为基础,利用G IS工具,对比分析了五种插值方法在青藏高原不同降水年份(以1998年、1997年分别代表丰水及欠水年份)的干湿季(1998年的干湿季分别以12月份和8月份为代表,1997年的干湿季分别以1月份和7月份为代表)月降水插值中的应用,并对整个高原地区的干季和湿季的月降水进行制图。这5种插值方法分别是:克里金插值法、反距离加权法、样条法、混合插值法Ⅰ和混合插值法Ⅱ,前3种插值方法未考虑海拔高度对降水的影响,而混合插值法则将高程作为降水的重要影响因子。结果表明:①在干季,无论是丰水还是欠水年份,月降水量都比较少,高程对降水量的影响较小,在月降水插值时可不考虑高程的影响,克里金法的月降水插值精度最高。②在湿季,月降水量较多,高程的影响较大,混合插值法比局部插值法及克里金插值法的精度高,尤以混合插值法Ⅱ(多元回归和样条法的综合)的精度最高。③干季,整个高原的月降水很少,西部和北部降水最少,东部和南部相对较多;湿季,高原的月降水较多,空间格局表现为由东南向西北递减。Spatial precipitation interpolation is of interest because it depends on many environmental variables, and there is no universal interpolation model applicable to all the terrains. The spatial availability of interpolation is problematic because the precipitation data was recorded at distributed weather forecast stations. As a result, values at any other point in the terrain must be interpolated from the neighboring stations. Using multiple linear regression and Geographic Information System （GIS）, the spatial distributions of monthly precipitation for both wet and dry years in the Tibetan Plateau were modeled. Based on the precipitation data collected by 140 stations for both wet year 1998 and dry year 1997 （Dec. and Aug. for dry and wet months in 1998, Jan. and Jul. for dry and wet months in 1997） and the DEM data of the Tibetan Plateau, a monthly precipitation map of the Tibetan Plateau was drawn. Five interpolation methods were compared in this study and they were Kriging, Inverse Distance Weighting （IDW）, Splines, and two mixed methods. One mixed method is the combination of Multiple Regression and Inverse Distance Weighting, while the other is the combination of Multiple Regression and Splines. Both of the mixed methods take the elevation as an important factor during interpolation, while the former other three methods do not put into consideration of elevation influence. The mixed methods can be summarized into the following formula ： P = f （B, L, H） ＋ R; where P refers to precipitation, B/L/H represents respectively longitude, latitude and elevation, and R refers to the residual. The validity of the monthly rainfall maps was checked through 10 independent experiment weather stations. The results show ： ① In dry seasons, monthly precipitation was low regardless of wet years or dry years; and the best results for dry monthly precipitation mapping were obtained by using the Kriging interpolation. ② In wet seasons, monthly precipitation was highly affected by the factor of altitude,

关 键 词：月降水量 局部插值 混合插值 青藏高原 

分 类 号：P468.024[天文地球—大气科学及气象学]