Department of Environmental Health Sciences, University at Albany, State University of New York, One University Place, Rensselaer, NY 12144, USA.
Joint Global Change Research Institute, Pacific Northwest National Laboratory, 5825 University Research Court, College Park, MD 20740, USA.
Sci Total Environ. 2020 Apr 20;714:136697. doi: 10.1016/j.scitotenv.2020.136697. Epub 2020 Jan 16.
Climate change is exacerbating environmental pollution from crop production. Spatially and temporally explicit estimates of life-cycle environmental impacts are therefore needed for suggesting location and time relevant environmental mitigations strategies. Emission factors and process-based mechanism models are popular approaches used to estimate life-cycle environmental impacts. However, emission factors are often incapable of describing spatial and temporal heterogeneity of agricultural emissions, whereas process-based mechanistic models, capable of capturing the heterogeneity, tend to be very complicated and time-consuming. Efficient prediction of life-cycle environmental impacts from agricultural production is lacking. This study develops a rapid predictive model to quantify life-cycle global warming (GW) and eutrophication (EU) impacts of corn production using a novel machine learning approach. We used the boosted regression tree (BRT) model to estimate future life-cycle environmental impacts of corn production in U.S. Midwest counties under four emissions scenarios for years 2022-2100. Results from BRT models indicate that the cross-validation (R) for predicting life cycle GW and EU impacts ranged from 0.78 to 0.82, respectively. Furthermore, results show that future life-cycle GW and EU impacts of corn production will increase in magnitude under all four emissions scenarios, with the highest environmental impacts shown under the high-emissions scenario. Moreover, this study found that changes in precipitation and temperature played a significant role in influencing the spatial heterogeneity in all life-cycle impacts across Midwest counties. The BRT model results indicate that machine learning can be a useful tool for predicting spatially and temporally explicit future life-cycle environmental impacts associated with corn production under different climate scenarios.
气候变化正在加剧农作物生产带来的环境污染。因此，需要对农作物生产的生命周期环境影响进行时空明确的估算，以便提出具有针对性的环境缓解策略。排放因子和基于过程的机制模型是常用的估算生命周期环境影响的方法。然而，排放因子往往无法描述农业排放的时空异质性，而能够捕捉这种异质性的基于过程的机制模型往往非常复杂且耗时。因此，高效预测农作物生产的生命周期环境影响仍然存在挑战。本研究采用一种新的机器学习方法，开发了一种快速预测模型，用于量化玉米生产的生命周期全球变暖（GW）和富营养化（EU）影响。我们使用了增强回归树（BRT）模型来估算美国中西部各县在 2022 年至 2100 年四个排放情景下玉米生产的未来生命周期环境影响。BRT 模型的交叉验证（R）结果表明，预测生命周期 GW 和 EU 影响的 R 值分别在 0.78 到 0.82 之间。此外，结果表明，在所有四个排放情景下，玉米生产的未来生命周期 GW 和 EU 影响都将增加，在高排放情景下的环境影响最大。此外，本研究发现，降水和温度的变化在影响中西部各县所有生命周期影响的空间异质性方面发挥了重要作用。BRT 模型的结果表明，机器学习可以成为一种有用的工具，用于预测在不同气候情景下与玉米生产相关的时空明确的未来生命周期环境影响。
