Geostatistical Assessment of Soil Nutrient Variability in Cultivated Land of Wuwei, Arid Hexi Corridor, China

Abstract

Regional assessment of soil fertility requires separating concentration dispersion from spatial dependence while recognizing the limits imposed by coarse regional surveys. We quantified soil organic matter (SOM), total nitrogen (TN), available phosphorus (AP), and available potassium (AK) in 638 topsoil samples (0–20 cm) from cultivated land in Wuwei, northwestern China. Descriptive statistics, Pearson correlations, distributional diagnostics, isotropic semivariograms, and additional sensitivity analyses of drift, directionality, and spatial validation were used. Pooled coefficients of variation ranged from 57.8% to 97.3%, with AK showing the greatest dispersion. SOM and TN were strongly correlated (r = 0.954, p < 0.01). The strongest in-sample semivariogram fits were exponential for SOM, spherical for TN, and Gaussian for both AP and AK; nugget-to-sill ratios [C0/(C0 + C)] were 0.266–0.441. In a separate 373-record georeferenced sensitivity analysis, detrended directional practical-range ratios (larger/smaller) ranged from 1.17 to 1.92, and spatially blocked five-fold R2 values were markedly lower than random-fold values, especially for AP. Thus, directional structure and geographic transfer remain uncertain, and the empirical variogram fit does not establish predictive performance. The results provide a regional baseline for sampling prioritization, not field-scale fertilizer prescription.