首頁 - 生態(tài)環(huán)境監(jiān)測 - 應(yīng)用 - 文章推薦 | EARTH-SCI REV：量化田間農(nóng)業(yè)碳排放結(jié)果的可擴(kuò)展框架 A scalable framework for quantifying field-level agricultural carbon outcomes
（《量化田間農(nóng)業(yè)碳排放結(jié)果的可擴(kuò)展框架》）
期刊：Earth-Science Reviews
2021年影響因子：12.038
5年影響因子：14.424
在線發(fā)表時(shí)間：2023.05.29
第一作者：管開宇副教授
第一單位：美國伊利諾伊大學(xué)厄巴納-香檳分校/可持續(xù)發(fā)展、能源和環(huán)境研究所/農(nóng)業(yè)生態(tài)系統(tǒng)可持續(xù)發(fā)展中心
研究背景Research background
農(nóng)業(yè)貢獻(xiàn)了全球近四分之一的溫室氣體排放（GHG），這激發(fā)了人們對采用某些有可能減少溫室氣體排放或在土壤中固碳的農(nóng)業(yè)實(shí)踐方法的興趣。相關(guān)的GHG排放（包括N2O和CH4）和土壤碳儲(chǔ)量的變化在這里被定義為“農(nóng)業(yè)碳結(jié)果（agricultural carbon outcomes）”。農(nóng)業(yè)碳結(jié)果的準(zhǔn)確量化是實(shí)現(xiàn)農(nóng)業(yè)減排的基礎(chǔ)，但現(xiàn)有的碳結(jié)果測量方法（包括直接測量、排放因子、基于過程的建模）未能達(dá)到支持可信（credible）、可驗(yàn)證（verifiable）和成本效益高的（cost-effective）碳結(jié)果的測量和改進(jìn)所需的準(zhǔn)確性和可擴(kuò)展性（scalability）。
Illustration of the “additionality" concept for agricultural carbon credit, using ahypothetical corn-soybean rotation field in the U.S. Midwest as an example,assuming cover cropping is newly adopted in 2021 with a ten-year commitment.(a) Annual change in the SOC stock (i.e. ASOC) since 2015, with hypotheticalscenarios from 2021 to 2030.(b) Generated annual carbon credit from 2021 to2030.(c) Change in SOC stock over time.
研究內(nèi)容Research contents
在這里，作者提出了一個(gè)基礎(chǔ)和可擴(kuò)展的框架，以量化農(nóng)田的田間碳結(jié)果，該框架基于農(nóng)業(yè)生態(tài)系統(tǒng)的整體碳平衡：農(nóng)業(yè)生態(tài)系統(tǒng)碳結(jié)果?=?環(huán)境（E）?×?管理（M）?×?作物（C）。
在全面審閱了與現(xiàn)有方法相關(guān)的科學(xué)挑戰(zhàn)以及它們在成本和準(zhǔn)確性之間的權(quán)衡之后，作者提出，量化農(nóng)田田間碳結(jié)果的最可行途徑是有效整合各種方法（例如，不同的觀測、傳感器/現(xiàn)場數(shù)據(jù)、建模），定義為“系統(tǒng)體系（system-of-systems）”解決方案。
這種“系統(tǒng)體系”解決方案應(yīng)同時(shí)包括以下組成部分：（1）可擴(kuò)展地（scaleable）收集地面實(shí)況數(shù)據(jù)，并在當(dāng)?shù)貙?shí)地對環(huán)境變量（E）、管理做法（M）和作物狀況（C）進(jìn)行跨尺度傳感；（2） 具有必要過程的高級(jí)建模，以支持碳結(jié)果的量化；（3）系統(tǒng)的模型數(shù)據(jù)融合（Model-Data Fusion，MDF），即在每個(gè)地方農(nóng)田級(jí)別集成傳感數(shù)據(jù)和模型的穩(wěn)健有效的方法；（4）高計(jì)算效率和人工智能（AI），以低成本擴(kuò)展到數(shù)百萬個(gè)單獨(dú)的實(shí)地農(nóng)田；以及（5）穩(wěn)健的多層驗(yàn)證系統(tǒng)和基礎(chǔ)設(shè)施，以確保解決方案的準(zhǔn)確性（fidelity）和真正的可擴(kuò)展性，即解決方案在所有目標(biāo)領(lǐng)域以公認(rèn)的準(zhǔn)確性穩(wěn)健執(zhí)行的能力。
在這方面，作者在本文中提供了詳細(xì)的科學(xué)原理、當(dāng)前進(jìn)展和未來研發(fā)（R&D）優(yōu)先事項(xiàng)，以實(shí)現(xiàn)“系統(tǒng)體系”解決方案的不同組成部分，從而實(shí)現(xiàn)環(huán)境×管理×作物框架，以量化田間級(jí)農(nóng)業(yè)碳結(jié)果。
Fig. 1. Conceptual diagram of quantifying agroecosystem carbon outcomes at the fieldlevel for agroecosystems.(a) Agricultural carbon outcome is determined by three factors.i.e. environment condition (E), management practices (M), and crop condition (C), as welas their interactions.(b) Accuracy of the quantification methods improves significantly asmore information is constrained at the field level. The example shown here focuses onquantifying net ecosystem exchange (NEE), which is the net CO exchange between landand atmosphere that can be measured directly with the eddy-covariance flux tower sitesin the U.S. Midwest (Zhou et al, 2021); the three scenarios refer to: (left) only using Einformation (i.e. weather and soil) as input in the carbon outcome quantification,center) using both M (i.e. field-level management practices) and E information for thecarbon outcome quantification, and (right) using C (i.e. photosynthesis, yield, leaf areaindex), M, E information together to drive or constrain the model.
Fig. 2. The holistic carbon and nitrogen balance and its linkage with greenhouse gasemissions over annual row cropping farmland(a) and a mass balance based approach toquantify the change of soil organic carbon (SOC)(b). GPP: gross primary productivity;Ra:autotrophic respiration; Rh: heterotrophic respiration; NEE: net ecosystem exchange:DOM: dissolved organic matter; POM: particulate organic matter; MAOM: mineral-associated organic matter.
Fig. 3. Soil sampling accuracy (i.e. minimum detectable change, in terms of relativechange in the SOC stock) as a function of the number of soil samples and field sizeswhich is much larger than the annual change of SOC stock in reality (Maillard et al.,2017).
Fig. 4. Qualitative illustration of how different technological solutions for quantifyingfield-level carbon outcomes fit in the accuracy and cost diagram. The relative positions ofdifferent solutions are mainly based on the technical review by ARPA-E (DOE ARPA-E:DE-FOA-0002250, 2020) and literature (e.g. Paustian et al., 2019; Smith et al., 2020).
Fig. 5. Illustration of a "System-of-Systems" solution for quantifying field-level carbonoutcome, including above and belowground processes. The "System-of-Systems" solutionincludes sensing, monitoring, modeling, and model-data fusion, targeting to assure field-level accuracy, scalability, and cost-effectiveness. ξ represents carbon loss from varioussources, which is usually very small (<0.5%) and thus can be neglected in most cases.
Fig. 6. Cross-scale sensing to generate photosynthesis information at the field level. (Top)The cross-scale sensing from leaf to canopy, and to regional levels for estimatingphotosynthesis.(Bottom) A snapshot of field-level estimation of photosynthesis on 07-10-2020, derived from the large-scale SLOPE photosynthesis data at daily frequency(Jiang et al., 2021), showing field-level Champaign County pattern and a field-level dailytime series of photosynthesis.
Fig. 7. Cross-scale sensing to quantify regional high-resolution tillage intensity (Wang etal., 2023a, Wang et al., 2023b). Ground truth of crop residue cover was obtained throughground photos and computer vision-aided image segmentation. Then, airbornehyperspectral imaging along with machine learning was applied to upscale ground pointmeasurements to landscape scale. Finally, massive airborne hyperspectral survey derivedresidue cover data were integrated with multi-source satellite fusion data to deriveregional information of residue cover and tillage intensity.
Fig. 8. GPP is closely linked with soil carbon input, i.e. litter and root exudates, and thusdirectly affect change of SOC(Grant, 2001).This site-level sensitivity analysis isconducted over three sites with different environmental and soil conditions in northern(Site 1), central(Site 2) and southern (Site 3) illinois using the ecosys model (Zhou et al.2022a, Zhou et al. 2022b).
Fig. 9. N2O and CO2 fluxes estimated by two KGML models.(a) KGML outperforms theprocess-based model and pure ML model in simulating N2O fluxes. (b) KGML predictednet ecosystem exchange (NEE) demonstrated great agreement with observations from 1eddy-covariance flux towers in the U.S. Midwest. Validations of KGML were carried outon a dataset that was not used in its training process.
Fig. 10. Example of the Tier 1 super site: using the ARPA-E SMARTFARM Phase 1 site atChampaign, Illinois, managed by University of illinois Urbana Champaign.
原文鏈接：
https://www.sciencedirect.com/science/article/pii/S0012825223001514
電話：