山东农业大学学报（自然科学版）2025，Vol.56Issue(6)：974-985,12.DOI:10.3969/j.issn.1000-2324.2025.06.008
基于随机森林算法的苹果种植适宜性评价
Evaluation of Apple Cultivation Suitability Based on Random Forest Algorithm
摘要
Abstract
Scientific evaluation of apple planting suitability plays a vital role in the rational utilization of land resources,scientific spatial planning,and sustainable development of the apple industry.This study takes Zhaotong City,the largest high-quality apple production region in southern China,as the research area,and considers entire land resources throughout Zhaotong City as the research subject.Employing the Random Forest algorithm,a total of 16 feature factors across four dimensions—topography,climate,soil,and spatial accessibility—selected as input variables,while apple suitability probability values are used as the output variable,this study conducts an evaluation of the apple cultivation suitability in Zhaotong City.The results show that the prediction model for apple cultivation suitability based on the Random Forest algorithm achieves a training accuracy of 88.73%,and the area under the Receiver Operating Characteristic curve,namely the AUC,reaches 0.96,indicating high prediction accuracy.In Zhaotong City,the areas classified as highly suitable,suitable,sub-suitable and unsuitable for apple cultivation are 1 052.76 km2,2 350.8 km2,4 639.73 km2,and 14 956.71 km2,accounting for 4.58%,10.22%,20.17%and 65.03%of the total land area of Zhaotong City,respectively.The highly suitable areas are mainly concentrated in Zhaoyang District and Ludian County;the suitable and sub-suitable areas are mostly located in Yiliang County and Zhenxiong County;and the unsuitable planting areas are primarily distributed in Shuifu City,Yanjin County,Weixin County and Suijiang County.The key factors affecting apple cultivation suitability are identified as soil clay content,topographic relief,average annual temperature,and average annual relative humidity.The application of the Random Forest algorithm avoids the heterogeneity caused by subjective manual weighting and multi-source data,thereby improving the objectivity of the evaluation results.This study provides theoretical support and technical methods for the multi-factor suitability evaluation of apple cultivation and the scientific planning of the apple industry in Zhaotong City.关键词
随机森林算法/苹果种植/适宜性评价/昭通市Key words
Random forest/apple planting/suitability evaluation/Zhaotong City分类
农业科技引用本文复制引用
胡丹,刘绥华,宁飞,徐青松,崔文刚..基于随机森林算法的苹果种植适宜性评价[J].山东农业大学学报（自然科学版）,2025,56(6):974-985,12.基金项目
国家自然科学基金项目(42161029) （42161029）
