摘要
以三峡库区草堂河流域为研究区,利用网格布点,共采集102个土壤样点,分析测定土壤的pH值,结合成土母质和地形等10个环境因子,以样点总数的85%作为训练集进行预测模型构建,15%作为验证集检验模型精度,利用随机森林(Random Forest, RF)模型对研究区土壤pH进行空间分布预测并制图。结果表明:土壤pH与谷深、坡长呈显著正相关,与海拔、距河网垂直距离、坡高呈显著负相关。三叠系大冶组灰岩发育的土壤pH值高于三叠系须家河组石英砂岩发育的土壤pH值。基于环境因子的RF预测模型,平均绝对误差(MAE)为0.47、均方根误差(RMSE)为0.59、决定系数(R^2)为0.85,能解释研究区土壤pH值85%的空间变异。对土壤pH值产生主要影响的环境因子为成土母质和海拔。可见,基于环境因子的RF预测模型,预测精度高,可以作为土壤pH空间分布预测的有效方法,能为流域尺度下其他土壤属性的空间分布预测提供依据和借鉴。
A total of 102 samples were collected from the topsoil at a depth of 20 cm to predict and map the spatial distribution of soil pH over the Caotang River Basin in the Three Gorges Reservoir Area. The samples were divided into calibration(85%) and validation(15%) sets. Random Forest(RF) method was applied to predict the spatial distribution of soil pH based on parent materials and terrain indicators(Elevation, Slope, Aspect, Slope Height, Valley depth, Topographical wetness index, Vertical Distance to Channel Network, Multi-resolution index of valley bottom flatness, Slope Length). The major influencing environmental factors on soil pH spatial variability were investigated by the RF model. The results showed that soil pH was significantly positively correlated to Valley depth and Slope Length, while significantly negatively correlated to Elevation, Vertical Distance to Channel Network and Slope Height. Soils developed from Limestone of Triassic Daye formation had higher values of pH than that developed from Sandstone of Triassic Xujiahe Formation. The RF model had a good performance with the mean absolute error(MAE), the root mean square error(RMSE) and the determination coefficient(R^2) of 0.47, 0.59 and 0.85, respectively. The model could explain 85% variation of soil pH in the study area. The major factors to soil pH variations were soil parent material and elevation. Therefore, RF model can serve as an effective method to predict the spatial distribution of soil pH, and can provide the basis and reference for other soil properties prediction at watershed scale.
作者
马冉
刘洪斌
武伟
MA Ran;LIU Hong-bin;WU Wei(College of Resources and Environment,Southwest University,Chongqing 400716,China;College of Computer and Information Science,Southwest University,Chongqing 400716,China;Chongqing Key Laboratory of Digital Agriculture,Chongqing 400716,China)
出处
《长江流域资源与环境》
CAS
CSSCI
CSCD
北大核心
2019年第3期691-699,共9页
Resources and Environment in the Yangtze Basin
基金
国家科技支撑计划课题(2008BADA4B10)
中央高校基本科研业务费专项(XDJK2016D041)
关键词
环境因子
随机森林
土壤PH
空间分布预测
environmental factors
RF model
soil pH
prediction of spatial distribution
作者简介
马冉(1996~),女,硕士研究生,主要从事土地信息系统研究.E-mail:mr5159@126.com;通讯作者:刘洪斌,E-mail:lhbin@swu.edu.cn.
