南京林业大学学报(自然科学版) ›› 2022, Vol. 46 ›› Issue (3): 177-184.doi: 10.12302/j.issn.1000-2006.202103028

• 研究论文 • 上一篇    下一篇

基于空间自相关的县域尺度土壤侵蚀抽样方法研究

陈超1(), 齐斐1, 徐雁南1,*(), 黎家作2, 赵传普2, 苏新宇2   

  1. 1.南京林业大学林学院,江苏 南京 210037
    2.水利部淮河水利委员会淮河流域水土保持监测中心站,安徽 蚌埠 233001
  • 收稿日期:2021-03-11 接受日期:2021-05-06 出版日期:2022-05-30 发布日期:2022-06-10
  • 通讯作者: 徐雁南
  • 基金资助:
    国家自然科学基金项目(32071840);国家自然科学基金项目(31070627);水利部重点科技项目(SBJ2018010)

Sampling methods of soil erosion at county scale based on spatial autocorrelation

CHEN Chao1(), QI Fei1, XU Yannan1,*(), LI Jiazuo2, ZHAO Chuanpu2, SU Xinyu2   

  1. 1. College of Forestry, Nanjing Forestry University, Nanjing 210037, China
    2. Monitoring Center Station of Soil and Water Conservation, Huaihe River Commission, Ministry of Water Resources, Bengbu 233001, China
  • Received:2021-03-11 Accepted:2021-05-06 Online:2022-05-30 Published:2022-06-10
  • Contact: XU Yannan

摘要:

【目的】探讨不同空间抽样方法和插值方法对土壤侵蚀预测精度的影响,为县域土壤侵蚀快速评价提供技术支撑。【方法】以沂蒙山区蒙阴县为研究区,GF-1号和GF-6号卫星影像、1∶1万地形图等数据为基础数据源,基于空间自相关理论,采用不同的空间抽样方法和空间插值方法,对研究区土壤侵蚀状况进行快速抽样预测研究。【结果】①根据全局Moran’s I指数、Z score和平均图斑类型,选取400 m×400 m方格为抽样单元最优尺寸。②与空间随机抽样、空间系统抽样方法相比,空间分层抽样样本容量最少(150个,其他样本容量最均超过250个),抽样精度最高(96.69%),抽样效率最高。③空间分层抽样下,普通克里金法和协同克里金法的计算结果与栅格计算法差异较小,相对差异分别为8.25%和9.85%(另两种方法的相对误差大于25%),但在空间分布上,受坡度影响的协同克里金法土壤侵蚀强度分布更为详细。【结论】综合考虑抽样调查的精度和工作量,在沂蒙山区县域尺度开展土壤侵蚀调查工作时,可采用空间分层抽样和协同克里金法或普通克里金法的方法进行土壤侵蚀快速预测,从而为区域生态环境保护提供重要数据支撑。

关键词: 土壤侵蚀, 空间自相关, 空间抽样方法, 空间插值, 抽样单元尺寸, 沂蒙山区

Abstract:

【Objective】Soil erosion is one of the main ecological and environmental problems in China. To take effective measures to control soil erosion, field investigations must be undertaken. It is important to maintain the ecological balance and promote sustainable development of the social economy. 【Method】In China, the method of soil erosion investigation combining sampling surveys and a Chinese soil loss equation (CSLE) experience model is one of the most important methods to obtain the soil erosion status at different scales. Stratified sampling with unequal probability and the CSLE experience model were used in the First National Water Census from 2010 to 2012. To further explore the influence of different spatial sampling methods and spatial interpolations on estimation precision in soil erosion areas, and to provide technological support for fast regional evaluation of soil erosion, this paper uses Mengyin County of Yimeng Mountain as the study area. Based on the spatial autocorrelation theory, using GF-1 and GF-6 satellite images, a CSLE model, 1∶10 000 topographic map and other data, three spatial sampling methods of spatial random sampling, spatial systematic sampling, spatial stratified sampling and four spatial interpolation methods of inverse distance weight, spline function method, ordinary Kriging, and co-Kriging was applied to predict the soil erosion area in Mengyin County in the Yimeng Mountain area. 【Result】(1) According to the global Moran’ I index, Z score and average pattern type, 400 m × 400 m square is the optimal size for sampling units. (2)The efficiency and accuracy of the sampling survey were affected by the sampling methods. Compared with spatial random samples and spatial systematic sampling, which had more than 250 units, spatial stratified sampling by slope had the highest sampling efficiency with the minimum sample size (150 units) and the highest sampling accuracy (96.69%). (3) Using spatial stratified sampling, the percentage of the soil erosion area for both ordinary Kriging and co-Kriging was similar to the grid calculation with a relative error of 8.25% and 9.85%, while that of the inverse distance weight and the spline function method were quite different from the grid calculation with a relative error of more than 25%. The spatial distribution of soil erosion-based co-Kriging was more detailed and is mainly influenced by the slope, but there were some differences between the spatial interpolation and the grid calculation. 【Conclusion】In view of the precision and workload in soil erosion surveys at the county scale in Yimeng Mountain area, stratified sampling, ordinary Kriging or co-Kriging are advisable for rapid prediction of soil erosion to provide an important data support for regional ecological environment protection.

Key words: soil erosion, spatial autocorrelation, spatial sampling method, spatial interpolation, sampling unit size, Yimeng Mountain area

中图分类号: