基于随机森林的耕地质量评价智能模型及其应用研究
作者:
作者单位:

华中师范大学城市与环境科学学院/地理过程分析与模拟湖北省重点实验室

作者简介:

通讯作者:

中图分类号:

基金项目:

国家自然科学基金项目(42171061)、国家科技基础资源调查专项(2021FY100505)和湖北省农业农村厅重点项目(2018-05-20,2018-10-15)资助


An Intelligent Model of Cultivated Land Quality Evaluation Based on Random Forest and Its Application
Author:
Affiliation:

Central China Normal University

Fund Project:

Supported by the National Natural Science Foundation of China (No. 42171061), Special Foundation for National Science and Technology Basic Research Program of China (No. 2021FY100505) and Hubei Provincial Department of Agriculture and Rural Affairs Key Project (Nos. 2018-05-20, 2018-10-15)

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
    摘要:

    为了更精确地表达耕地质量的系统性、非线性、差异性等特点,本研究旨在探索一种新的智能化耕地质量评价方法,提高耕地质量评价精度。以襄州区为研究区域,从地形地貌、土壤条件、社会经济、生态安全4个方面构建耕地质量综合评价指标体系,选用熵权法(EW)、后向传播神经网络(BPNN)、随机森林(RF)三种模型进行训练,比较三种模型的评价精度,并分析襄州区2018年耕地质量等级分布规律。结果表明:(1)襄州区耕地质量整体较好,以二、三等级为主,累积面积占比达到54.63%,呈现出明显的地域分异规律,高质量耕地主要分布在中北部,低质量耕地主要聚集在南部,且各乡镇耕地质量等级分布也具有明显差异;(2)耕地质量RF评价模型能够较为精确地模拟指标之间的复杂关系,科学定量分析各指标对耕地质量的贡献;(3)耕地质量平均指数比较,RF>BPNN>EW,RF与BPNN的评价结果具有相似的空间分布,且均与EW的差异较为显著;(4)相比于BPNN和EW,RF具有更高的数据挖掘能力和训练精度,其评价结果最为理想,决定系数R2为0.814 5,MAE为0.009,MSE为0.012,RF能有效运用于耕地质量评价研究。本研究丰富和完善了县域尺度耕地质量评价指标体系及方法,为襄州区耕地资源数量、质量、生态“三位一体”的管护提供理论依据,同时为其他类似地区耕地质量评价提供借鉴与参考。

    Abstract:

    【Objective】Constructing a scientific and quantitative quality-assessment model for cultivated land is important for understanding cultivated land quality, and can provide a theoretical basis and technical support for formulating rational and effective management policies and realizing the sustainable use of cultivated land resources. To accurately reflect on the systematic, complex, and differential characteristics of cultivated land quality, this study aimed to explore an intelligent cultivated land quality assessment method that avoids the subjectivity of determining indicator weights while improving assessment accuracy. 【Method】In this study, taking Xiangzhou in Hubei Province of China as the study area, 14 indicators were selected from four dimensions—terrain, soil conditions, socioeconomics, and the ecological environment—to build a comprehensive assessment index system for cultivated land quality. A total of 1,590 representative cultivated land quality samples in Xiangzhou were selected, of which 1,110 were used as training samples, 320 as test samples, and 160 as validation samples. Three models of entropy weight (EW), back propagation neural network (BPNN), and random forest (RF) were selected for training, and the assessment results of cultivated land quality were output through simulations to compare the assessment accuracy of the three methods to verify the reliability and superiority of the RF model. In addition, the distribution pattern of cultivated land quality grades in Xiangzhou in 2018 was also analyzed in this study. 【Result】 The results are summarized as follows: (1) The overall quality of cultivated land in Xiangzhou was better, with a larger area of second- and third-grade farmland, accounting for 54.63%, and the grades conformed to a positive distribution trend. From the distribution point of view, the spatial distribution of cultivated land quality in Xiangzhou was unbalanced, influenced by the topography and socioeconomic development level and showing an obvious geographical differentiation pattern, with overall characteristics of high in the north-central area and low in the southern area. The distribution of cultivated land quality grades also varied widely among towns. (2) The RF model for cultivated land quality assessment required fewer parameters and could simulate the complex relationships between indicators more accurately and analyze each indicator’s contribution to cultivated land quality scientifically. (3) In terms of the average quality index of farmland, RF > BPNN > EW. The spatial patterns of the quality index from RF and BPNN were similar, and both were significantly different from EW. (4) Compared to BPNN and EW, RF had a higher data mining ability and training accuracy, and its assessment result was the best. The coefficient of determination (R2) was 0.8145, the mean absolute error (MAE) was 0.009, and the mean squared error (MSE) was 0.012. 【Conclusion】 The findings in this study showed that RF was more suitable for the quality assessment of cultivated land with complex nonlinear characteristics. This study enriches and improves the index system and methodological research of cultivated land quality assessment at the county scale, and provides a theoretical basis for achieving a threefold production pattern of cultivated land quantity, quality, and ecology in Xiangzhou, while also serving as a reference for the evaluation of cultivated land quality in similar regions.

    参考文献
    相似文献
    引证文献
引用本文

王 丽,周 勇,李 晴,徐 涛,左 岍,吴正祥,刘婧仪.基于随机森林的耕地质量评价智能模型及其应用研究[J].土壤学报,2022,59(5). DOI:10.11766/trxb202012030670 WANG Li, ZHOU Yong, LI Qing, XU Tao, ZUO Qian, WU Zhengxiang, LIU Jingyi. An Intelligent Model of Cultivated Land Quality Evaluation Based on Random Forest and Its Application[J]. Acta Pedologica Sinica,2022,59(5).

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数: