论文部分内容阅读
利用降水和地形资料，建立了定西县及其邻近的２５个观测站点的栅格数字高程模型（ＤＥＭ）、各测站近３４年的实测降水量及其相关信息空间数据库．采用统计模型、空间插值和综合方法，模拟了研究区３４年的年平均降水量、降水变率、降水保证率、丰水年和欠水年的年均降水量，以及雨季（７～９月）和春小麦生长时期（３月中旬～７月中旬）的降雨量空间分布，用５个观测站点的降水资料对各种模拟方法进行了精度评估．并且，在不同的空间尺度上（定西县各乡、村；典型研究区的小型集水盆地等地域单元），利用上述模拟的降水信息空间数据库计算了３４年的年均降水总量及其变化．研究结果表明，在平均降水量模拟的９种方法中（距离权重法、趋势面法、样条函数法、普通Ｋｒｉｇｉｎｇ法、通用Ｋｒｉｇｉｎｇ方法１、通用Ｋｒｉｇｉｎｇ方法２、泰森多边形法、多元回归法和综合方法），计算精度最高的为综合方法，其最大相对误差为６．８９５％；其次为普通Ｋｒｉｇｉｎｇ法，其最大相对误差为６．７５６％．在７种插值算法中，采用指数模型模拟半异矩的普通Ｋｒｉｇｉｎｇ法计算精度最高．统计模型的误差较大，其最大相对误差为９．０４１％．利用实测降水信息空间数据库和模拟精度最优的降水信息空间分布数?
Based on the data of precipitation and topography, a grid digital elevation model (DEM) of 25 observation stations in Dingxi County and its adjacent areas was established, and the measured precipitation of each station in the past 34 years and its related information spatial database were established. The statistical models, spatial interpolation and comprehensive methods were used to simulate the 34-year annual average precipitation, precipitation variability, precipitation guarantee rate, annual precipitation in wet and dry years, and rainy season (July-September) ) And spring wheat growing period (mid March to mid July). The accuracy of various simulation methods was evaluated by precipitation data from 5 observation stations. In addition, on the different spatial scales (each rural area and village in Dingxi County; small watershed basin in a typical study area), the total annual precipitation and its variation over the past 34 years were calculated using the above-mentioned simulated precipitation information spatial database . The results show that among the 9 methods of average precipitation simulation (distance weight method, trend surface method, spline function method, ordinary Kriging method, general Kriging method 1, general Kriging method 2, Thiessen polygon method, multiple regression method And comprehensive method), the highest accuracy is the comprehensive method, the maximum relative error is 6.895%, followed by the ordinary Kriging method, the maximum relative error is 6.756%. Among the seven kinds of interpolation algorithms, the ordinary Kriging method which uses the exponential model to simulate the semi-moment has the highest computational accuracy. The error of statistical model is larger, the maximum relative error is 9.041%. Using the measured precipitation information spatial database and the simulation accuracy of the optimal spatial distribution of precipitation information?