基于辅助变量和神经网络模型的土壤有机质空间分布模拟

发布时间：2018-04-11 13:17

本文选题：土壤有机质 + 辅助变量　；参考：《长江流域资源与环境》2017年08期

【摘要】：为快速准确获取省域尺度下土壤有机质的空间分布状况。以江西省2012年测土配方施肥项目采集的16 582个耕地表层(0~20 cm)土壤样点数据,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和邻近信息作为辅助变量,构建基于地理坐标与辅助变量的BP神经网络模型和普通克里金法结合的方法(BPNN_OK)、基于地理坐标与辅助变量的RBF神经网络模型和普通克里金法结合的方法(RBFNN_OK)和普通克里金法(OK法)3种方法,模拟省域尺度下耕地表层(0~20 cm)土壤有机质的空间分布。对2 416个验证样点进行独立验证的研究结果显示:基于辅助变量的神经网络模型较普通克里金法有较大提升。BPNN_OK法对土壤有机质预测结果的均方根误差、平均绝对误差、平均相对误差较OK法分别降低了2.76 g/kg、2.34 g/kg、9.83%,RBFNN_OK法较OK法分别降低了2.70 g/kg、2.29 g/kg、9.61%。研究显示,基于辅助变量的神经网络模型与OK法结合的方法明显地提高了土壤有机质空间分布模拟精度,并且存在改进和提高的空间。
[Abstract]:In order to obtain the spatial distribution of soil organic matter at provincial scale quickly and accurately.Based on the data of 16 582 topsoil samples collected from Jiangxi Province in 2012, environmental factors and adjacent information were extracted by means of four directions search method, geostatistics and remote sensing image analysis techniques.The method of combining BP neural network model based on geographical coordinates and auxiliary variables and ordinary Kriging method to build RBF neural network model based on geographical coordinates and auxiliary variables and ordinary Kerkin methodLijin method and OK method,The spatial distribution of soil organic matter was simulated on a provincial scale.The results of independent verification of 2 416 verification samples show that the neural network model based on auxiliary variables has a larger RMS error and an average absolute error than the ordinary Kriging method in predicting soil organic matter by the BPNNNOK method.The average relative error was 2.76 g / kg / kg 2.34 g / kg / kg 9.83% lower than that of OK method, and 2.70 g / kg / kg 2.29 g / kg / kg 9.61% lower than that of OK method respectively.The results show that the combination of neural network model based on auxiliary variables and OK method can obviously improve the simulation accuracy of soil organic matter spatial distribution, and there is room for improvement and improvement.
【作者单位】：江西农业大学江西省鄱阳湖流域农业资源与生态重点实验室/国土资源与环境学院;南方粮油作物协同创新中心;东华理工大学马克思主义学院;
【基金】：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)~~
【分类号】：S153.6

【相似文献】