Background Accurate description of spatial heterogeneity of hydrophysical properties of forest soil is of great significance for understanding hydrological processes such as stand infiltration and runoff, as well as ecological processes such as tree growth and material cycle. And the accurate acquisition of the optimal sampling size is conducive to improving the measurement accuracy of soil hydrophysical properties, and thus providing scientific basis for forest soil investigation and sampling design in the study area.

Methods We selected a Larix principis-rupprechtii plantation in the semi-humid area of Liupan Mountains, Ningxia province as the study area. The spatial variability of surface (i.e. 0-20 cm soil depth) soil mass moisture content, bulk density and saturated hydraulic conductivity was studied using the methods of classical statistics and geostatistics. And the optimal sampling quantity of each index at 90% and 95% confidence level was quantified using Monte Carlo simulation resampling method.

Results 1) The surface soil hydrophysical properties had obvious spatial heterogeneity, for example, the coefficient of variation of soil mass moisture content and bulk density and saturated hydraulic conductivity were 0.23, 0.11 and 0.47, respectively, which belonged to medium variation and had obvious spatial structure. And their ratio of nugget (C₀) and sill (C₀+C) were 0.043, 0.119 and 0.075, respectively, showing a strong spatial correlation and small spatial autocorrelation range. 2) Under the confidence interval of 90% or 95%, if the measurement error was controlled within 10%, it was at least 13 or 18 sampling points of soil mass moisture content, 4 or 5 sampling points of soil bulk density and 34 or 45 sampling points of soil saturated hydraulic conductivity should be arranged in 0-20 cm soil depth of the sample plot of 30 m×30 m. If the measured error was controlled within 5%, at least 35 or 45 sampling points of soil mass moisture content, 13 or 15 sampling points of soil bulk density and 70 or 75 sampling points of soil saturated hydraulic conductivity should be arranged.

Conclusions Surface soil mass moisture content, bulk density and saturated hydraulic conductivity of the L. principis-rupprechtii plantation in our study site have significant spatial heterogeneity, and the optimal sample size of each index representing the average level of sample plot vary largely. At present, the sampling size of forest soil hydrophysical properties is relatively small, and it is necessary to increase the sampling size in the future research to improve the accuracy of estimation. Besides, the change of sampling quantity of soil hydrophysical properties caused by their spatial heterogeneity should be fully considered in the investigation of forest soil.