南方丘陵区土壤有机质空间插值模型及采样点密度对农用地分等精度的影响——以福建省龙海市为例

The Influence of Spatial Interpolation Model and Sampling Density of Soil Organic Matter in the Farmland Quality Evaluation Accuracy in Hilly Region of South China:A Case Study of Longhai City, Fujian Province

研究目的:分析南方丘陵区土壤有机质的采样点密度和空间插值模型对农用地分等精度的影响。研究方法:以福建龙海市为研究区,对设计的8种格网密度和6种结合不同类型信息的插值模型所得农用地分等结果进行比较研究。研究结果:(1)结合不同类型信息的克里格空间插值模型对于土壤有机质含量及农用地分等成果(自然质量等)具有显著差异,土壤有机质含量和农用地分等成果的精度与格网密度呈正相关。其中结合地貌和土壤信息的空间插值方法(KDMTR)对于农用地分等成果有最好的预测效果;(2)如果仅需考虑获取较高精度的土壤有机质含量信息时,按2 km×2 km的样点密度并结合KDMTR法进行空间插值,为最高效的样点布设和数据处理方式;(3)在开展县级农用地分等时,如果仅需考虑获取农用地分等结果时,土壤采样点密度对农用地分等精度影响较小,但结合不同类型信息的空间插值方法对农用地分等成果精度影响显著。采用KDMTR法并按3.5 km×3.5 km的格网密度布设土壤调查样点,为最高效的样点布设和空间插值模型。研究结论:南方丘陵区在开展县级农用地分等工作时,采用的空间插值模型对农用地分等成果的精度产生显著影响而土壤采样点布设的格网密度对农用地分等成果的精度影响较小。

The purpose of this paper is to research spatial interpolation model and sampling density＇s effects on quality evaluation accuracy in hilly region of south China. The paper compares different evaluation results according to eight kinds of grid density based on grid sampling of soil organic matter and six classification methods in Longhai City of Fujian Province. The results show that： 1）Combination of different information of kriging interpolation models have significant differences on soil organic matter and the result of farmland classification（natural quality）, soil organic matter and the result of farmland classification show positively correlated with grid density of soil sample. Spatial interpolation method which combined with topography and soil information（KDMTR）has the best prediction for farmland quality evaluation. 2）If accurate information of soil organic matter content should be in consideration, the efficient sampling point layout is grid sampling based on the topography and soil types, and the best grid size is about 2 km×2 km. 3）If only farmland quality evaluation result at the county level should be in consideration, the density of soil sampling points has small effects on the accuracy of farmland quality evaluation result, but the combination of different information space interpolation method has significant effects on the accuracy of farmland quality evaluation result. The best spatial interpolation model is Kriging interpolation based on the topography and soil types and the most efficient sampling point layout is grid sampling with grid size of 3.5 km×3.5 km. The paper concludes that obviously effect can be noticed by spatial interpolation model to evaluate farmland quality, yet the density of sampling has small effect at the county level in hilly region of south China.