县域土壤质量数字制图方法比较

土壤质量研究几乎涵盖土壤研究的所有领域，土壤质量制图理论与方法是土壤质量研究的一项重要研究内容。该研究以北京市密云县为研究区，基于土壤质量评价最小数据集和指数和法计算的土壤质量指数，探究了在地学模型支持下区域土壤质量数字制图方法。研究设计了5种区域土壤质量数字制图方法，并比较了不同方法的空间数字制图精度。结果显示，目前广泛使用的基于参评指标空间插值结果的土壤质量数字制图方法精度最低、工序较繁琐，且无法反映研究区景观高度异质的特点；而基于计算后的土壤质量指数（soil quality index，SQI），借助于地统计学方法的土壤质量数字制图方法相对比较科学合理，其中又以基于计算后的SQI和回归克里格法预测效果最好，均方根误差最小，仅为0.01897，相对于基于参评指标空间插值结果的土壤质量数字制图方法，精度相对提高率最大，达到50%以上。综合考虑空间制图精度、工序的繁简程度，在该研究设计的5种方法中基于计算的SQI和回归克里格法最佳，该法避免了地统计插值在景观高度异质区的应用局限性，预测结果与实际最为相符。

Comparison of digital mapping methods of regional soil quality

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.