Performance analysis of different kriging interpolation methods based on air quality index in Wuhan

To explore the influence on interpolation results in different models of ordinary kriging interpolation methods and to validate their feasibility in practice, the interpolation experimental study using Spherical, Exponential and Gaussian models were carried out in this paper, based on the monthly average Air Quality Index data of 10 air quality sites in Wuhan, China. The interpolation results were qualitatively analyzed and quantitatively evaluated by the actual render effect and cross-validation method. The experimental results demonstrate that: (1) for the same data, the visual render effect in different models are almost the same in the whole, but the interpolation accuracies are different; (2) the Mean Standardized error and Root-Mean-Square error of the Spherical model are the largest; (3) the Mean error of the Exponential model is minimal, the Root-Mean-Square error is nearest to Average Standard Error, and its Root-Mean-Square Standardized error is closest to 1; (4) the Mean error of the Gaussian model is the largest, but both Mean Standardized error and Root-Mean-Square error are minimal, the Root-Mean-Square error is farthest away from Average Standard error, and also its Root-Mean-Square Standardized error is farthest away from the value 1.