引用本文:江叶枫,孙凯,郭熙,叶英聪,饶磊,李伟峰,等.基于环境因子和邻近信息的土壤属性空间分布预测[J].环境科学研究,2017,30(7):1059-1068.
JIANG Yefeng,SUN Kai,GUO Xi,YE Yingcong,RAO Lei,LI Weifeng,et al.Prediction of Spatial Distribution of Soil Properties Based on Environmental Factors and Neighbor Information[J].Reserrch of Environmental Science,2017,30(7):1059-1068.]
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 182次   下载 298 本文二维码信息
码上扫一扫!
分享到: 微信 更多
基于环境因子和邻近信息的土壤属性空间分布预测
江叶枫1,2, 孙 凯1,2, 郭 熙1,2, 叶英聪2, 饶 磊1,2, 李伟峰1,2
1.江西农业大学国土资源与环境学院, 江西 南昌 330045 ;2.江西省鄱阳湖流域农业资源与生态重点实验室, 江西 南昌 330045
摘要:
为探索乡镇尺度上土壤属性空间分布预测的最佳方法,以江西省万年县齐埠镇为例,借助四方位搜索法、地统计学和遥感影像分析技术提取环境因子(地形因子和植被覆盖指数)和邻近信息〔w(有机质)与w(速效钾)〕,构建OK法(普通克里金法)、RK1法(仅基于环境因子的回归克里金法)以及RK2法(基于环境因子和邻近信息的回归克里金法)对齐埠镇耕地表层(0~20 cm)土壤w(有机质)、w(速效钾)空间分布进行预测. 结果表明:齐埠镇土壤w(有机质)平均值为35.03 g/kg,w(速效钾)平均值为96.73 mg/kg,均为中等空间变异性. 对62个样点进行建模,16个测试样点进行独立验证的误差分析表明,RK2法对土壤w(有机质)、w(速效钾)预测结果的均方根误差、平均绝对误差和平均相对误差较OK法分别降低了18.05%、18.01%、21.77%和7.25%、9.49%、9.84%;较RK1法分别降低了22.48%、20.91%、22.02%和9.27%、12.61%、13.52%. 研究显示,RK2法明显提高了土壤w(有机质)、w(速效钾)空间分布模拟精度,并且存在改进和提高的空间.
关键词:  土壤属性  四方位搜索法  回归克里金法  空间预测
DOI:
分类号:
基金项目:国家自然科学基金项目(41361049);江西省自然科学基金项目(20122BAB204012);江西省赣鄱英才“555”领军人才项目(201295)
Prediction of Spatial Distribution of Soil Properties Based on Environmental Factors and Neighbor Information
JIANG Yefeng1,2, SUN Kai1,2, GUO Xi1,2, YE Yingcong2, RAO Lei1,2, LI Weifeng1,2
1.Academy of Land Resource and Environment, Jiangxi Agricultural University, Nanchang 330045, China ;2.Key Laboratory of Poyang Lake Watershed Agricultural Resources and Ecology of Jiangxi Province, Nanchang 330045, China
Abstract:
Abstract: As the most important determinants of soil quality, soil properties significantly influence farmland use and soil environmental protection. In order to find the best interpolation method of soil properties, including soil organic matter (SOM) and Soil Available Potassium (AK) in Qibu Town in Wannian County, Jiangxi Province, Ordinary Kriging (OK), Regression Kriging 1 (RK1, based on environmental factors), Regression Kriging 2 (RK2, based on environmental factors and neighbor information) were adopted to predict the distribution of soil properties. Environmental factors were extracted by digital terrain and remote sensing image analysis technique. The four-direction search method was applied to get the neighbor information. To establish and validate the three methods, 78 points of surface soil samples (0-20 cm) were collected and randomly divided into two groups, as modeling points (62) and validation points (16). The results showed that SOM content ranged from 17.30-53.58 g/kg, with an average value of 35.03 g/kg, a moderate variability. The nugget/sill ratio was 0.59, indicating a moderate spatial dependence for SOM. AK content ranged from 50.98 to 152.13 mg/kg, with an average value of 96.73 mg/kg, a moderate variability. The nugget/sill ratio was 0.66, indicating a moderate spatial dependence for AK. The prediction map obtained by RK2 model was more consistent with the true geographical information than OK and RK1. Moreover, RK2 model reduced the prediction errors. Compared to RK1 model, the root mean square errors, the mean absolute errors and the mean relative errors of RK2 were 18.05%, 18.01%, and 21.77% (for SOM), and 7.25%, 9.49%, and 9.84% (for AK), smaller than those of OK. The same numbers for the OK model with the validation points were 22.48%, 20.91%, and 22.02% (for SOM), and 9.27%, 12.61%, and 13.52% (for AK), smaller than those of RK1. The results suggested that it is helpful for improving the prediction accuracy to employ neighbor information and environmental factors in spatial prediction of soil properties. Therefore, RK2 could be recognized as the best interpolation method, but could be improved in the future.
Key words:  soil properties  four-direction search method  Regression Kriging method  spatial prediction