|
Spatial variability of soil nutrients under different parent materials (New Alluvium, Shaximiaozu, and Suiningzu) was conducted in the main grain production areas of Liangping, Chongqing. Two indicators, namely, Mean Prediction Error (MPE) and Root Mean Squared Prediction Error (RMSPE), were used to evaluate the accuracy of spatial interpolation of soil nutrients. The main results were as follows:(1) On average, the order of coefficients of variation of soil nutrients was soil P> soil N> soil K> organic matter (OM). The spatial variability structure of soil nutrients was explored using GS+. The range of OM was 50940m, followed by soil N (21480m), K (2910m) and P (1710m). Soil P had lowest nugget value while OM had largest one. In terms of nugget to sill ratio, OM, soil N, and K showed moderate spatial autocorrelation suggesting that their spatial variability were affected by natural factors as well as human activities, whereas soil P presented weak spatial autocorrelation indicating that P was mainly associated with human activities.(2) The variability of soil nutrients under different parent materials.①New Alluvium region. The coefficients of variation of soil nutrients originated from New Alluvium showed in the order of soil P> soil K>soil N>OM. Soil K had the largest range of 57240m followed by OM (3240m), P (1520m), and soil N (1320m). In terms of nugget to sill ratio, soil P presented weak spatial autocorrelation indicating that P was mainly associated with human activities, whereas OM, soil N, and K showed moderate spatial autocorrelation suggesting that their spatial variability were affected by natural factors as well as human activities.②Shaximiaozu region. The coefficients of variation of soil nutrients developed from Shaximiaozu hold the rank of soil P> soil K>soil N>OM. Organic matter had the largest range of 4560m followed by soil P (1080m), K (850m), and N (730m). In terms of nugget to sill ratio, soil N presented weak spatial autocorrelation indicating that N was mainly associated with human activities, whereas OM, soil P, and K showed moderate spatial autocorrelation suggesting that their spatial variability were affected by natural factors as well as human activities.③Suiningzu region. The coefficients of variation of soil nutrients originated from Suiningzu were in the order of soil P>soil K>soil N>OM. Organic matter had the largest range of 27660m followed by soil N (2370m), K (1430m), and P (1100m). In terms of nugget to sill ratio, soil N, P, and K presented weak spatial autocorrelation indicating that their spatial variability were mainly associated with human activities, whereas OM showed moderate spatial autocorrelation suggesting that OM was affected by natural factors as well as human activities.(3) Analysis of prediction accuracy. In the case of eliminating the effect of parent materials, MPEs were 0.001424,0.149,0.1701 and 0.19968 and RMSPEs were 0.9108,0.8054,0.8603 and 0.8811 for OM, soil N, P and K, respectively.Nevertheless, MPE was lower and RMSPE was much closer to 1 with respect to the effect of parent material. In New Alluvium region, MPEs were-0.000941,0.01799,-0.1466 and-0.1712 and RMSPEs were 0.987,0.9712,1.048 and 0.9343 for OM, soil N, P and K, respectively. In the area of Shaximiaozu, MPEs were 0.0009544,-0.02857,-0.1682 and-0.1741, RMSPEs were 1.057,0.9168, 0.9837 and 1.028 for OM, soil N, P and K, respectively. In Suiningzu region, MPEs were 0.001404, 0.02087,-0.1533 and-0.1735, and RMSPEs were 1.037,1.004,1.114 and 1.023, respectively. Consequently, it was of great necessity to take into account the effects of parent materials to study the spatial interpolation of soil nutrients.(4) Through spatial interpolate and splice the soil nutrients under three parent materials, we could come to the conclusion that OM ranged between 0.6-1.3%, the content of OM was deficient in the study area. Content of soil N was below the critical value (50mg/L), changed between 10.9-33.9%. Soil P ranged between 2.9-19.0%, there were approximately 94.7% of study area in which the content of soil P were below the critical value (12mg/L). Soil K changed between 37.2-90.6%, and about 95.5% of study area where the content were below the critical value (78.2mg/L).
|