利用基于地统计学的简单克里格、普通克里格、协克里格及综合方法4种插值方法对中国西部地区的气温估计结果进行比较研究.气温的空间分布格局和交叉验证结果表明在考虑高程信息后协克里格和综合方法比简单克里格和普通克里格的效果好.相比之下,协克里格比综合方法有更多奇异值出现；其他指标也表明综合方法优于其他3种插值方法.%More attention had been paid to interpolation methods based on geostatistics for climate variables. Four interpolation methods for air temperature in western China, namely, simple kriging, ordinary kriging, co-kriging and combined approach, were compared with their temperature distribution and cross-validated. The temperature distribution provided by cokriging and combined approach considering altitude information was better than simple krging and ordinary kriging. The details were well reflected. But more extreme singular values appeared in co-kriging than in combined approach. Further analysis showed that mean errors of simple kriging, ordinary kriging and co-kriging method were less than the combined approach, while their root mean squared error and correlation coefficient were lower than the combined approach. Although co-kriging method could show more details than simple kriging and ordinary kriging, the other indicators showed that there were many restraints to use in western China. In contrast, the performance of the combined approach was optimal.
展开▼
