Priority selection rating of sampling density and interpolation method for detecting the spatial variability of soil organic carbon in China

Abstract

Soil sampling density and spatial interpolation method both have effects on interpreting the spatial variability of regional soil organic carbon (SOC). However, there are few comparisons of the effects between the two factors. Based on three soil sampling designs in Yujiang County, Jiangxi Province of China, the SOC spatial distributions in a specific area, imposed on three grid sampling densities of 2 × 2 km (G2×2), 1 × 1 km (G1×1), and 0.5 × 0.5 km (G0.5×0.5), were predicted via two interpolation methods: Ordinary Kriging (OK) and Kriging combined with land use information (LUK). Prediction accuracies from OK and LUK at three sampling densities were compared on the basis of 65 validation samples in the area. The results demonstrated that the correlation coefficients (r) between the measured and predicted values of validation locations obtained from OK (r = 0.212, 0.491 and 0.512) and LUK (r = 0.602, 0.776 and 0.875) increased with decreased grid size, and the root mean square errors (RMSE) from OK (RMSE = 6.79, 5.33, and 5.19 g kg−1) and LUK (RMSE = 4.74, 3.60, and 3.14 g kg−1) all decreased as expected with the sampling density increasing from G2×2 to G0.5×0.5. The rs from LUK were all higher and RMSEs were all lower than those from OK at three densities, respectively. More interesting, the prediction accuracy of LUK from G2×2 was not only lower than that of OK at same density, but also lower than those of OK at G1×1 and G0.5×0.5. This indicates that LUK can use several times fewer soil samples than OK to predict SOC spatial variability with same accuracy. The conclusion is that the efficient interpolation method not only makes sense to obtain high-precision SOC distribution information, but also can save lots of sampling points and research costs. Therefore, research on efficient interpolation method is a key step and should be paid more attention than increasing sampling points for revealing SOC spatial variability in the hilly red soil region of China, even in the regions with similar complex terrain.