Abstract: Kriging, inverse distance weighting (IDW) and spline are three classical methods for interpolating soil properties. High accuracy surface modeling (HASM) has been developed in recent years, which is successfully used in geographic information system and ecological modeling. In order to assess its feasibility of soil property interpolation, HASM based on multi-grid (HASM-MG) was used to interpolate soil pH for a red soil hilly region of Jiangxi Province, China. A total of 150 samples of topsoil (0-20 cm) were collected in Ji'an municipal district, Ji'an County and Taihe County in October, 2007. Three different data sets were randomly created for simulation and validation to comparatively simulate performance of HASM, kriging, IDW and spline. A total of 30, 60 and 90 samples were removed as validation points, thus 120, 90 and 60 data points were selected as simulation points, respectively. The results show that HASM simulation is the closest to the measured data among the four methods. It has much smaller mean absolute error, root mean square error and much higher correlation coefficient of simulated value and measured value than the three classical methods. Soil pH values in the central and southeastern parts of the study area dominated by paddy soils are much higher than those in the red soil area. Therefore, HASM-MG can be considered as an accurate method for interpolating soil properties on the premise of improving the calculation speed, and it advances the practicability of HASM applied in soil property interpolation.

Key words: soil pH, high accuracy surface modeling based on multi-grid, interpolation accuracy, the red soil hilly region in South China