中国农业科学 ›› 2025, Vol. 58 ›› Issue (13): 2538-2551.doi: 10.3864/j.issn.0578-1752.2025.13.004

• 耕作栽培·生理生化·农业信息技术 • 上一篇    下一篇

基于DNDC和NSGA-Ⅲ耦合模型的旱地春小麦稳产减排多目标优化

曹景文1(), 聂志刚1,2(), 李广2,3, 杨洁4   

  1. 1 甘肃农业大学信息科学技术学院,兰州 730070
    2 甘肃农业大学省部共建干旱生境作物学国家重点实验室,兰州 730070
    3 河西学院,甘肃张掖 734000
    4 甘肃农业大学草业学院,兰州 730070
  • 收稿日期:2024-12-31 接受日期:2025-05-30 出版日期:2025-07-01 发布日期:2025-07-05
  • 通信作者:
    聂志刚,E-mail:
  • 联系方式: 曹景文,E-mail:1519436076@qq.com。
  • 基金资助:
    国家自然科学基金(32160416); 国家自然科学基金(32360438); 甘肃省拔尖领军人才项目(GSBJLJ-2023-09); 2025年甘肃省高校研究生“创新之星”项目(2025CXZX-797)

Multi-Objective Optimization of Stable Yield and Emission Reduction of Dryland Spring Wheat Based on DNDC and NSGA-Ⅲ. Coupling Model

CAO JingWen1(), NIE ZhiGang1,2(), LI Guang2,3, YANG Jie4   

  1. 1 School of Information Science and Technology, Gansu Agricultural University, Lanzhou 730070
    2 Gansu Agricultural University State Key Laboratory of Aridland Crop Science, Lanzhou 730070
    3 HeXi University, Zhangye 734000, Gansu
    4 College of Pratacultural Science of Gansu Agricultural University, Lanzhou 730070
  • Received:2024-12-31 Accepted:2025-05-30 Published:2025-07-01 Online:2025-07-05

摘要:

【目的】 为应对日益增长的粮食需求和生态可持续性要求,探讨灌溉与施肥综合调控对西北地区旱地春小麦产量、土壤CO2及N2O排放通量的综合影响,以明确最优的灌溉量及施肥策略,实现农业生产与环境效益的协调发展。【方法】 基于DNDC农业生态系统模拟模型,立足于2021—2023年甘肃省定西市安定区凤翔镇安家坡村的田间试验数据,对模型进行参数校准和验证,构建不同施肥水平(0—400 kg·hm-2)与灌溉水平(0—300 mm)的管理情景,探索不同灌溉量与施肥量管理措施下的小麦生长动态和土壤温室气体(CO2、N2O)排放通量的响应规律,结合NSGA-Ⅲ多目标优化算法构建多目标优化框架,以“最大化作物产量”“最小化土壤CO2排放通量”“最小化土壤N2O排放通量”为三目标函数,实现小麦产量提升与土壤温室气体减排的协同优化,确定可以兼顾产量和环境效益的最优管理方案。【结果】 DNDC模型能够较好地模拟春小麦产量及土壤温室气体排放通量。在4种施肥梯度处理下,3年间产量、土壤CO2和N2O排放通量的归一化均方根误差(NRMSE)分别为17.4%—18.8%、7.62%—11.41%和9.19%—12.47%;在2种灌溉量处理下,3年间产量的归一化均方根误差(NRMSE)为13.3%—17.2%。优化后的灌溉量与施肥量表明,当施肥量控制在150—180 kg·hm-2、灌溉量110—150 mm时,小麦产量可提升至2 088.48 kg·hm-2,同时土壤CO2排放通量控制在每年4 998.87—5 011.50 kg·hm-2,土壤N2O排放通量控制在每年4.06—4.14 kg·hm-2。【结论】 耦合DNDC模型与NSGA-Ⅲ算法可实现旱地春小麦产量与土壤温室气体排放通量的协同优化,当灌溉量为110—150 mm、施氮量为150—180 kg·hm-2时,可在保障产量稳定的同时有效控制土壤CO2及N2O排放通量,为陇中旱地春小麦稳产减排提供科学依据。

关键词: 春小麦, 产量, DNDC模型, NSGA-Ⅲ, 温室气体排放, 多目标优化算法

Abstract:

【Objective】 In response to growing food demand and ecological sustainability requirements, this study explored the comprehensive impact of integrated irrigation and fertilization management on spring wheat yield, soil CO2 and N2O emissions fluxes in arid areas of Northwest China, with the aim of identifying optimal irrigation and fertilization strategies to achieve coordinated development of agricultural production and environmental benefits. 【Method】 Based on the DNDC agricultural ecosystem simulation model, using field trial data from Anjiapo Village, Fengxiang Town, Dingxi City, Gansu Province, from 2021 to 2023, the model was calibrated and validated. Different fertilization levels (0-400 kg·hm-2) and irrigation levels (0-300 mm). The model simulates the response patterns of wheat growth dynamics and soil greenhouse gas (CO2 and N2O) emission fluxes under different irrigation and fertilization management measures. Combining the NSGA-III multi-objective optimization algorithm, a multi-objective optimization framework was established with three objective functions: “maximizing crop yield”, “minimizing soil CO2 emissions flux”, and “minimizing soil N2O emissions flux” as the three objective functions, achieving synergistic optimization between wheat yield enhancement and soil greenhouse gas emission reduction, and determining the optimal management scheme that balances yield and environmental benefits. 【Result】 The DNDC model effectively simulates spring wheat yield and soil greenhouse gas emission fluxes. Under four fertilization gradient treatments, the normalized root mean square error (NRMSE) for yield, soil CO2 emissions, and N2O emissions over three years was 17.4%-18.8%, 7.62%-11.41%, and 9.19%-12.47%, respectively. Under two irrigation treatments, the normalized root mean square error NRMSE for yield over three years was 13.3%-17.2%. The optimized irrigation and fertilization rates indicate that when fertilization is controlled at 150-180 kg·hm-2 and irrigation volume is 110-150 mm, wheat yield can be increased to 2 088.48 kg·hm-2, while soil CO2 emissions flux is controlled at 4 998.87-5 011.5 kg·hm-2 per year, and soil N2O emission flux is controlled at 4.06-4.14 kg·hm-2 per year. 【Conclusion】 Coupling the DNDC model with the NSGA-III algorithm enables the simultaneous optimization of spring wheat yield and soil greenhouse gas emissions fluxes in dryland areas. When the irrigation amount is set between 110-150 mm and nitrogen application rate between 150-180 kg·hm-2, it is possible to maintain stable yields while effectively controlling soil CO2 and N2O emission fluxes. This provides a scientific basis for achieving both yield stability and emission reduction in dryland spring wheat systems in central Gansu.

Key words: spring wheat, yield, DNDC model, NSGA-Ⅲ, greenhouse gas emissions, multi-objective optimization algorithm