摘要
以闽南地区的厦门市、漳州市和泉州市为研究区,首先利用灰色关联模型探究初选因子对1982—2018年耕地土壤有机碳变化的驱动作用,再基于筛选出的主要驱动因子构建BP神经网络模型模拟预测2019—2055年碳源/汇的变化趋势.研究结果表明:1)土壤本身性质(砂粒含量、黏粒含量、初始有机碳密度等)、外在环境条件(道路通达度、年均温、年降水量等)和农业管理措施(秸秆还田量、耕作强度、有机肥和氮肥施用量等)是过去30多年耕地有机碳变化的主要驱动因子,碳源/汇与这些因素的关联度系数均达0.83以上;2)基于主要驱动因子构建的碳模型能较好地拟合研究区碳源/汇变化,模拟值相对误差和模拟效率分别为1.11%和0.48,均通过验证要求;3)未来30多年研究区耕地有机碳密度均值上升0.15 kg·m^(-2),总体表现为碳汇,但不同区域土壤具有“源”与“汇”双重效应,其中“碳源”土壤面积比例为30.33%,主要分布于平原台地区或为初始有机碳密度较高的耕地,“碳汇”土壤以有机碳密度上升幅度≤1 kg·m^(-2)为主,占“碳汇”土壤总面积的97.10%.因此,未来应重点关注“碳源”耕地,并针对性制定减排措施以期转“源”为“汇”,同时还需进一步提升土壤固碳增汇能力,防止高碳土壤发生碳损失而产生碳源效应.
Three cities located in Southern Fujian,namely Xiamen,Zhangzhou,and Quanzhou,were selected as the study area.The grey correlation model was employed to identify the driving effects of preliminary factors on the changes in cultivated soil organic carbon from 1982 to 2018.Based on the main driving factors identified,a BP neural network model was constructed to simulate and predict the trends in carbon sources/sinks from 2019 to 2055.The results indicate that:1)soil properties(sand content,clay content,initial soil organic carbon density,etc.),external environmental conditions(road accessibility,annual average temperature,annual precipitation,etc.),and agricultural management measures(straw returning amount,tillage intensity,organic or nitrogen fertilizer application amount,etc.)are the main driving factors of cultivated soil organic carbon changes over the past 30 years,with correlation coefficients of carbon source/sink to these factors all above 0.83.2)The carbon model established based on the main drivers fits the carbon source/sink changes in the study area quite well,with a relative error of the simulation values of 1.11%and a simulation efficiency of 0.48,both meeting the verification.3)Over the next 30 years,the average soil organic carbon density in the study area’s cultivated land is projected to increase by 0.15 kg·m^(−2),generally exhibiting a carbon sink behavior.However,soils in different regions have dual functions of carbon source and sink.The proportion of“carbon source”soils accounts for 30.33%,primarily located in plain and terrace areas or in cultivated lands with higher initial organic carbon densities.“Carbon sink”soils,characterized by an increase in organic carbon density of≤1 kg·m^(−2),make up 97.10%of the total area of“carbon sink”soils.Therefore,priority should be given to the soils with carbon source effect,and specific emission reduction measures should be developed to transform“sources”into“sinks”.Besides,it is also recommended to further improve the carbon sequestration and sink capacity of soils to prevent high-carbon soils from experiencing carbon loss,which could lead to a carbon source effect.
作者
龙军
黄倩
朱丽霞
LONG Jun;HUANG Qian;ZHU Lixia(School of Biological Science and Biotechnology,Minnan Normal University,Zhangzhou,Fujian 363000,China;University Key Lab of Soil Ecosystem Health and Regulation in Fujian,Fuzhou,Fujian 350002,China)
出处
《闽南师范大学学报(自然科学版)》
2024年第3期102-113,共12页
Journal of Minnan Normal University:Natural Science
基金
教育部基金(21YJC630090)
福建省自然科学基金(2021J05197)
福建省社会科学基金(FJ2021X002)
闽南师范大学校长基金(KJ2020025)。
关键词
有机碳变化
驱动因子
模型模拟
BP神经网络模型
耕地土壤
organic carbon change
driving factors
model simulation
BP neural network model
cultivated soil