Comparison of Three Spatial Interpolation Methods for Climate Variables in China

  • Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China

Received date: 2001-06-04

  Revised date: 2001-09-26

  Online published: 2002-01-25

Supported by

National Natural Science Foundation of China, No.49890330; Project of Institute of Geographic Sciences and Natural Resources Research, CAS, No.CX10G-C00-05-01; Special Funds for Major State Basic Research Project, No. G2000077905


Spatial interpolation of climate data is frequently required to provide input for regional plant growth models. As no single method among so many available ones to spatial interpolation of climate variables is optimal for all regions and all variables, it is very important to compare the results obtained using alternative methods applied to each set of data. For estimating 30-year of 10-day mean air temperature and monthly photosynthetic active radiation (PAR) fluxes at specific sites in China, we examined ordinary Kriging (OK) and other two relatively simple methods, one is inverse distance squared (IDS) technique and the other is gradient plus inverse-distance-squared (GIDS) technique. Based on the mean absolute errors from cross-validation tests, the methods were ranked as GIDS>OK>IDS for interpolating 10-day mean temperature, and IDS>GIDS>OK for interpolating monthly mean PAR fluxes. GIDS gives the lowest errors which averaged 1.3 oC for 10-day mean temperature. IDS gives the lowest errors averaged 0.46 MJ/(m2·d) for monthly PAR fluxes. Although OK errors were more consistent for temperatures of different seasons, yet GIDS had lower errors during crop growth seasons. Compared with OK, GIDS was simple to apply for interpolating 10-day mean temperature. Since PAR was affected by so many factors, the performance of GIDS and OK were not as good as we expected. It also implied that the complex methods could not ensure the best results. The multiple linear regressions carried out for GIDS revealed strong gradients in temperature and PAR fluxes varied by month in a fairly consistent way. For all 10-days, r was above 0.94 for temperature, and temperatures decreased with increasing latitude, longitude and elevation. The elevation coefficient for temperature during the year ranged from -0.0033~-0.0048 oC/m. For PAR, r was lower than that for temperature. The coefficients of PAR for latitude, longitude and elevation ranged from positive values to negative values during the year. Using a 1 km Digital Elevation Model (DEM) for China, the first 10-day mean temperature and monthly mean PAR flux of July were estimated for each pixel to assess the performance of the three interpolation techniques. From temperature maps, we can see that GIDS gives more reasonable estimates in valleys and mountainous area than the other two techniques. The temperature map generated by GIDS shows the elevation effects of temperature in West China especially in valleys and Tibet Plateau. The strong temperature trends towards the northeast in East China and PAR trends towards the northwest were also shown in the maps.

Cite this article

LIN Zhong-hui, MO Xing-guo, LI Hong-xuan, LI Hai-bin . Comparison of Three Spatial Interpolation Methods for Climate Variables in China[J]. Acta Geographica Sinica, 2002 , 57(1) : 47 -56 . DOI: 10.11821/xb200201006


[1] Ashraf M, Loftis J C, Hubbard K G. Application of geostatistics to evaluate partial station networks. Agric. For. Meteorol., 1997, 84: 225-271.

[2] Gottschalk L, Batchvarova E, Gryning S E et al. Scale aggregation-comparison of flux estimates from NOPEX. Agri. For. Meteorol., 1999, 98-99: 103-119.

[3] Nalder I A, Wein R W. Spatial interpolation of climate normals: test of a new method in the Canadian boreal forest. Agric. For. Meteorol., 1998, 92: 211-225.

[4] Price D T, Mckenney D W, Nalder I A et al. A comparison of two statistical methods for spatial interpolation of Canadian monthly mean climate data. Agric. For. Meteorol., 2000, 101: 81-94.

[5] Li Xing, Cheng Guodong, Lu Ling. Comparison of spatial interpolation methods. Advances in Earth Sciences, 2000, 15(3): 260-265.
[李新, 程国栋, 卢玲. 空间内插方法比较. 地球科学进展, 2000, 15(3): 260-265.]

[6] Zhou Yunhua. Distributions of energy and quantum fluxes of photosynthetically active radiation in China. In: Beijing Agro-ecosystems Experimental Station of Chinese Academy of Sciences (Eds). Experimental studies on the Eenvironment of Field Crops. Beijing: Meteorological Press, 1990. 15-39.
[周允华. 中国地区光合有效辐射能量和光量子通量的时空分布. 见: 中国科学院北京农业生态系统试验站. 农田作物环境实验研究. 北京: 气象出版社, 1990. 15-39.]

[7] Clark, I. Practical Geostatistics. London: Elseivier Applied Science, 1982. 129.

[8] Wackernagel H. Multivariate Geostatistics: An Introduction with Applications. Berlin Heidelberg Springer-Verlag. 1995.

[9] Hou Jingru. Practical Geostatistics. Beijing: Geological Press, 1998. 31-50.
[侯景儒. 实用地质统计学. 北京: 地质出版社, 1998. 31-50.]

[10] Sun Hongquan, Kang Yongshang, Du Huizhi. Practical Program Library of Geostatistics. Beijing: Geological Press. 1997.
[孙洪泉, 康永尚, 杜惠芝. 实用地质统计学程序集. 北京: 地质出版社, 1997.]

[11] Holdaway M R. Spatial modeling and interpolation of monthly temperature using Kriging. Clim. Res., 1996, 24:1835-1845.


[13] Hogg T H. Climate and the southern limit of the western Canadian boreal forest. Can. J. For. Res., 1994. 24:1835-1845.

[14] Lennon J J, Turner J R. Predicting the spatial distribution of climate: temperature in Great Britain. J. Anim. Ecol., 1995, 64: 370-392.