Main content area

A new spatial precipitation interpolation method based on the information diffusion principle

Huang, Huaping, Liang, Zhongmin, Li, Binquan, Wang, Dong
Stochastic environmental research and risk assessment 2019 v.33 no.3 pp. 765-777
algorithms, data collection, diffusivity, geostatistics, kriging, rain, uncertainty, China
In this paper, a spatial precipitation interpolation method based on the information diffusion principle is presented to reduce data or information uncertainties from point observations. To estimate the core parameters (i.e., diffusion coefficients) in the diffusion function, the empirical diffusion coefficient (EDC) and optimal diffusion coefficient (ODC) based on the particle swarm optimization (PSO) algorithm are introduced. The methodology was compared with other common interpolators including the inverse distance weighting (IDW) and three geostatistical interpolators. Three measures of interpolation quality were used to assess the accuracy of the results for all methods via the cross-validation procedure and a range of comparative split-sampling tests. The dataset chosen in the study comprises annual and seasonal precipitation means (2013–2017) measured at 80 rain gages in Linan City (East China). The findings of this study could be concluded as follows: (a) in general, the interpolation methods provided similar spatial distributions for annual precipitation, and the differences among the methods were small when using a dense observation network; (b) cokriging (CoOK), detrended universal kriging (DUK) and the ODC had the smallest errors in the cross-validation analysis; and (c) the EDC presented nearly the same accuracy as the ODC with low density of observation network, while the ODC was a better interpolator as the number of observations increased.