基于随机森林法的山核桃林地土壤速效养分含量空间分布特征研究  

Spatial distribution characteristics of soil available nutrients in hickory plantation based on random forest method

作　　者：凌晓丹 王罗其 赵科理[1,2] 傅伟军[1,3] 叶正钱[1,2] 丁立忠 LING Xiaodan;WANG Luoqi;ZHAO Keli;FU Weijun;YE Zhengqian;DING Lizhong(College of Environmental and Resource Sciences,Zhejiang Agriculture and Forestry University,Hangzhou 311300,China;Key Laboratory of Soil Contamination Bioremediation of Zhejiang Province,Zhejiang Agriculture and Forestry University,Hangzhou 311300,China;Key Laboratory of Carbon Cycling in Forest Ecosystems and Carbon Sequestration of Zhejiang Province,Zhejiang Agriculture and Forestry University,Hangzhou 311300,China;Agricultural and Forestry Technology Extension Center of Lin′an,Hangzhou 311300,China)

机构地区：[1]浙江农林大学环境与资源学院,杭州311300 [2]浙江农林大学,浙江省土壤污染生物修复重点实验室,杭州311300 [3]浙江农林大学,浙江省森林生态系统碳循环与固碳减排重点实验室,杭州311300 [4]临安区农林技术推广中心,杭州311300

出　　处：《生态学报》2024年第2期662-675,共14页Acta Ecologica Sinica

基　　金：浙江省基础公益研究计划项目(LY20C160004);杭州市临安区《“临安山核桃”亮牌战略三年攻坚行动方案》项目。

摘　　要：土壤氮磷钾是土壤肥力管理的重要元素,是植物生长的必要养分元素。对土壤氮磷钾的空间分布进行特征解译,可为精准管理临安山核桃产区林地土壤肥力,促进山核桃林产业可持续发展提供理论依据。研究以临安山核桃主产区为研究区域,利用随机森林(RF)、普通克里格(OK)和Shapley加性解释(SHAP)方法,结合地形因子、气候因子、土壤因子、遥感因子等环境变量,对山核桃林地土壤碱解氮(AN)、有效磷(AP)、速效钾(AK)的空间分布特征进行分析。研究结果表明:相比于OK模型,基于环境协变量所构建的RF模型对AN、AP和AK含量空间分布预测表现最佳,R^(2)分别为0.68、0.60和0.64,均方根误差(RMSE)分别为20.005、10.287和22.426,平均绝对误差(MAE)分别为15.425、7.709和21.628。RF模型SHAP分析显示,AN和AK含量分布主要受土壤有机质(SOM)的影响,并且SOM与AN和AK存在正相关性;AP主要受pH的影响,其次为色调指数,AP与pH和色调指数均具有负相关性;AK和AP同时受到海拔和坡向的影响。两种模型预测的氮磷钾空间分布趋势总体相似,不同速效养分存在明显的空间异质性。碱解氮高值区域主要分布于研究区东部;有效磷高值区域主要分布于研究区西部,但分散度高;速效钾高值区域则主要分布于研究区中部。总体而言,基于随机森林模型可以高精度模拟山核桃林地土壤氮磷钾含量空间分布特征,并依据主要环境协变量对土壤氮磷钾的影响关系,提出相应改良措施。在有效磷含量低值区域可以施用石灰来缓解土壤酸化,同时补追磷肥;碱解氮含量高值区域可以合理减少氮肥施用;速效钾含量低值区域合理施加钾肥;对于海拔较高及迎风坡多降雨的区域,可以构建林下高效水土保持植被,减轻水土流失;在林地施用有机肥料,改善土壤理化性质,增加土壤养分含量。Soil nitrogen,phosphorus,and potassium are important elements for soil fertility management as well as vital nutrient elements for plant development.Understanding the geographical distribution features of soil nitrogen,phosphorus,and potassium can offer a theoretical foundation for properly controlling the soil fertility of forest land in the Lin′an hickory plantation and boosting the hictory forest industry′s sustainable growth.In this study,the main production area of hickory plantation Lin′an was taken as the study area,and the random forest(RF),ordinary kriging(OK),and shapley additive interpretation(SHAP)methods were used to analyze the spatial distribution characteristics of soil alkali-hydrolyzable nitrogen(AN),available phosphorus(AP),and available potassium(AK)in hickory forest land in combination with environmental variables such as terrain factors,climate factors,soil factors,and remote sensing factors.According to the findings,the environmental covariates-based RF model outperformed the OK model in terms of accurately predicting the geographic distribution of AN,AP,and AK content.And theR^(2) of AN,AP,and AK in the RF model were 0.68,0.60,and 0.64,respectively,as well as the root mean square error(RMSE)of 20.005,10.287,and 22.426,and the mean absolute error(MAE)of 15.425,7.709,and 21.628.SHAP analysis of the RF model showed that the content distribution of AN and AK was mainly affected by soil organic matter(SOM),while SOM was positively correlated with AN and AK.AP was mainly affected by pH,followed by hue index.AP was negatively correlated with pH and hue index,and AK and AP were affected by elevation and aspect at the same time.The spatial distribution trends of AN,AK,and AP predicted by the two models were generally similar,but the spatial patterns of different available nutrients were different.The high value of AN was mainly distributed in the eastern part of the study area.The region with high AP value was mainly distributed in the west of the study area,but the dispersion was high.The hi

关 键 词：山核桃林地 随机森林 Shapley加性解释 速效养分 

分 类 号：S66[农业科学—果树学]