设为首页  |   加入收藏
引用本文:
【打印本页】   【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 112次   下载 251 本文二维码信息
码上扫一扫!
分享到: 微信 更多
基于投影寻踪的土壤养分综合评价及影响因素研究
张 彬1, 杨联安2, 杨粉莉3, 王 辉4, 谢贤健1, 陈卫军5
1.内江师范学院地理与资源科学学院;2.西北大学陕西省地表系统与环境承载力重点实验室;3.咸阳市农业科学研究院;4.咸阳市土壤肥料工作站;5.旬邑县土壤肥料工作站
摘要:
客观、精准研究土壤综合养分及影响因素,可为农作物科学施肥提供参考依据。以陕西省旬邑县苹果种植区为研究区,选取有机质、碱解氮、速效钾和有效磷为评价指标,将投影寻踪综合评价法运用到土壤养分评价中,根据最佳投影方向,计算土壤养分综合投影指数(comprehensive projection index, CPI),并依据经验等级对应投影值,从而评价土壤养分等级,基于普通克里格和GIS绘制土壤养分综合指数等级分布图,并利用随机森林的importance()函数分析影响因素对养分含量的重要性。结果表明:研究区CPI的变幅为0.653 ~ 1.516,平均值1.022,变异系数为14.746%,呈中等程度变异,同时CPI属于 Ⅲ 级和 Ⅳ 级;基于80% 训练集和20% 验证集的精度交叉验证,球面函数为最优理论模型;土壤综合养分的总体空间分布格局:除土桥、职田和湫坡头镇东部的养分等级高,土壤养分等级从西部向东部递减,从高到低的面积比例为:14.415:28.522:35.450:20.115:1.498,土壤综合养分属于中等及偏下水平,且主要分布在东部;高程、气温和坡度等因子是影响区域土壤养分的主导因子,且每项影响因子对土壤养分含量的影响度差异大。研究结果与当地实际相吻合,可为当地果园提高土壤肥力提供科学依据,也为区域土壤养分评价提供了新思路。
关键词:  土壤养分  影响因素  综合评价  投影寻踪  随机森林
DOI:10.13758/j.cnki.tr.2020.06.019
分类号:S158.2
基金项目:陕西省农业科技攻关项目(2011K02-11)、教育部人文社会科学研究规划项目(10YJA910010)、内江师范学院科研资助项目(17JC03)和西安市科技计划农业技术研发项目(NC1402,NC150201)。
Study on Comprehensive Evaluation and Influencing Factors of Soil Nutrients Based on Projection Pursuit
ZHANG Bin1, YANG Lianan2, YANG Fenli3, WANG Hui4, XIE Xianjian1, CHEN Weijun5
1.School of Geography and Resources Science, Neijiang Normal University;2.Shaanxi Key Laboratory of Earth Surface System and Environmental Carrying Capacity, Northwest University;3.Academy of Agriculture Sciences of Xianyang;4.Xianyang Station of Soil and Fertilizer;5.Xunyi Station of Soil and Fertilizer
Abstract:
The objective and accurate study of comprehensive soil nutrients and their influential factors can provide reference for scientific fertilization of crops. The apple-planting area in Xunyi County of Shaanxi Province was selected as the study object. The comprehensive evaluation of projection pursuit was applied to comprehensively assess soil nutrients including organic matter, alkali hydrolysable N, available K and available P. The comprehensive projection index of soil nutrients was calculated according to the optimum projection direction. Soil nutrient level was evaluated according to the level of experience corresponding projection value. The spatial distribution of comprehensive soil nutrients was mapped by ordinary Kriging method and GIS, and the importance function of random forests was used to analyze the importance of influencing factors on nutrient contents. The results showed that the CPI variation in the study area was 0.653-1.516 with an average of 1.022, the variation coefficient was 14.746%, which was moderate. Moreover, CPI were in grades of III and IV. The spherical function was the optimal theoretical model by the accuracy of the cross-validation based on 80% training set and 20% verification set. The spatial distribution pattern of soil nutrients was as follows: except the high grades in Tuqiao, Zhitian, and the eastern part of Qiupotou towns, the integrated soil nutrient grade decreased from the west to the east in the study area, and the area ratio of comprehensive levels of soil nutrients from high to low was: 14.415 : 28.522 : 35.450 : 20.115 : 1.498. The comprehensive levels of soil nutrients were medium or low, and were mainly distributed in the eastern part. Soil nutrients were mainly affected by elevation, temperature and slope, and the influence degree of each factor varied greatly. The above results are consistent with the actual conditions in the area, thus can provide scientific bases for improving soil fertility of the local orchards and new ideas for comprehensive evaluation of soil nutrients.
Key words:  Soil nutrients  Influencing factors  Comprehensive evaluation  Projection pursuit  Random forest

您是第3617379位访问者
版权所有 © 《土壤》编辑部
本系统由北京勤云科技发展有限公司设计   京ICP备09084417号