Comparison of digital mapping methods of regional soil quality
Zhang Shiwen,Zhang Liping,Ye Huichun,Hu Youbiao,Huang Yuanfang
DOI: https://doi.org/10.3969/j.issn.1002-6819.2013.15.031
2013-01-01
Abstract: Studies on soil quality cover almost all areas of soil studies, and soil quality cartographic theory and method is an important research subject of soil quality research. Based on an established minimum data set of soil quality assessment and soil quality index calculated by the index sum method, absorbing geostatistics research, the paper tried to explore the methods of digital mapping of soil quality in the geological model support. The study designed five methods of regional digital mapping of soil quality, which included the method of digital mapping based on spatial interpolation results on single indicators (M1), the method of digital mapping based on calculated SQI and inverse distance weighting (M2), the method of digital mapping based on SQI for samples and ordinary kriging method (M3), the method of digital mapping based on calculated samples SIQ and regression kriging (M4), and the method of digital mapping based on calculated SQI and indicators interpolation results (M5), respectively, and compared spatial mapping accuracies of the different methods. We established a minimum data set of soil quality assessment using six steps including Pearson correlation analysis, principal component analysis, the calculation of the vector norm values, the relationship analysis between environmental factors and soil quality, linear transformation and parameters score calculation, and sort packet. The results showed that RMSE value for the method for soil quality digital cartography based on spatial interpolation of the results of the participating indicators (RMSE=0.03831) is the largest, so the accuracy is the lowest, where RMSE value is minimum for the method based on calculated SQI and regression kriging (RMSE=0.01897), so the accuracy is the highest. The size relationship of RMSE values for the five methods:M1>M2>M3>M4>M5. The precision accuracy of the M1 method widely used is the minimum, the process is more cumbersome, and cannot reflect the characteristics of the highly heterogeneous landscape of the study area. For the method, the degree affected by the different participating indicators is relatively large, often showing a similar distribution pattern and some indicators, compared with the measured value of samples, prediction results are generally too large. Based on the soil quality index calculated, soil quality digital mapping method by means of geostatistical methods was relatively more scientific and reasonable, and predicted effect based on the soil quality index calculated and the regression kriging method was the best, and the relative increase in accuracy rate reached 50% more with respect to the method based on spatial interpolation results of the participating indicators. Considering the spatial mapping accuracy, the degree of sophistication of the process, the method based on the soil quality index calculated and regression kriging is optimal among the five methods of the study design, which uses a linear combination of the environment variables as an external drift trend to separate the residuals and it can eliminate smoothly, not only solve the larger problem on regression residuals, but also avoid the interpolation limitations of the highly heterogeneous landscape, and the predicted results was most consistent with the actual situation.