Spatial distribution of soil organic matter in tillage layers in a southern China basin using classifications and spatial interpolation algorithms

Soil organic matter (SOM) in tillage layers is a very important component of soil nutrients. A critical issue for effective management of agricultural fields is proper analysis and interpolating SOM point sampling data onto a surface. An optimal approach needs to be developed to investigate the concentration of SOM in tillage layers. In this study, a total of 2341 soil samples were collected. The landforms (Valley plains, Low hilly areas and High hilly areas), land-use patterns (Paddy field and Citrus orchard) and soil types (including Paddy soils and Red earths), which could affect the concentration of SOM in tillage layers, were taken into account as impact factors for the data classification. Five different spatial interpolation algorithms-inverse distance weighting (IDW), radial basis function (RBF), and kriging (including simple kriging (SK), ordinary kriging (OK), and universal kriging (UK)) - were used to analyze the spatial variations of SOM. Cross-validation was applied to evaluate the accuracy of these methods, and subsequent, mean error (ME) and root mean square error (RMSE) were used to compare and assess the performance of these interpolation methods. The results showed that in each classification of soil samples, the spatial variation of SOM in tillage layers was moderate to strong and the ratio of sill to nugget falls varied by 10%-50%. The study also revealed that RBF had the lowest RMSE value in most classification groups. With overall evaluation, the RBF was the optimal algorithm for interpolating SOM value in the study region, followed by SK, OK, UK and IDW. Finally, one classification group using RBF method was found to be the optimal method for predicting SOM in tillage layers in our study area, and a prediction map for the concentration of SOM was produced using this method.