Search

Accurate and cost-effective quantification of the carbon cycle for agroecosystems at decision-relevant scales is critical to mitigating climate change and ensuring sustainable food production. However, conventional process-based or data-driven modeling approaches alone have large prediction uncertainties due to the complex biogeochemical processes to model and the lack of observations to constrain many key state and flux variables. Here we propose a Knowledge-Guided Machine Learning (KGML) framework that addresses the above challenges by integrating knowledge embedded in a process-based model, high-resolution remote sensing observations, and machine learning (ML) techniques. Using the U.S. Corn Belt as a testbed, we demonstrate that KGML can outperform conventional process-based and black-box ML models in quantifying carbon cycle dynamics. Our high-resolution approach quantitatively reveals 86% more spatial detail of soil organic carbon changes than conventional coarse-resolution approaches. Moreover, we outline a protocol for improving KGML via various paths, which can be generalized to develop hybrid models to better predict complex earth system dynamics.

Agriculture contributes nearly a quarter of global greenhouse gas (GHG) emissions, which is motivating interest in adopting certain farming practices that have the potential to reduce GHG emissions or sequester carbon in soil. The related GHG emission (including N2O and CH4) and changes in soil carbon stock are defined here as “agricultural carbon outcomes”. Accurate quantification of agricultural carbon outcomes is the basis for achieving emission reductions for agriculture, but existing approaches for measuring carbon outcomes (including direct measurements, emission factors, and process-based modeling) fall short of achieving the required accuracy and scalability necessary to support credible, verifiable, and cost-effective measurement and improvement of these carbon outcomes. Here we propose a foundational and scalable framework to quantify field-level carbon outcomes for farmland, which is based on the holistic carbon balance of the agroecosystem: Agroecosystem Carbon Outcomes = Environment (E) × Management (M) × Crop (C). Following a comprehensive review of the scientific challenges associated with existing approaches, as well as their tradeoffs between cost and accuracy, we propose that the most viable path for the quantification of field-level carbon outcomes in agricultural land is through an effective integration of various approaches (e.g. diverse observations, sensor/in-situ data, and modeling), defined as the “System-of-Systems” solution. Such a “System-of-Systems” solution should simultaneously comprise the following components: (1) scalable collection of ground truth data and cross-scale sensing of environment variables (E), management practices (M), and crop conditions (C) at the local field level; (2) advanced modeling with necessary processes to support the quantification of carbon outcomes; (3) systematic Model-Data Fusion (MDF), i.e. robust and efficient methods to integrate sensing data and models at each local farmland level; (4) high computation efficiency and artificial intelligence (AI) to scale to millions of individual fields with low cost; and (5) robust and multi-tier validation systems and infrastructures to ensure solution fidelity and true scalability, i.e. the ability of a solution to perform robustly with accepted accuracy on all targeted fields. In this regard, we provide here the detailed scientific rationale, current progress, and future research and development (R&D) priorities to achieve different components of the “System-of-Systems” solution, thus accomplishing the Environment×Management×Crop framework to quantify field-level agricultural carbon outcomes.

Cropland carbon budget depicts the amount of carbon flowing in and out of agroecosystems and the changes in carbon stocks of soil and living biomass during the same period. Soil carbon credit is the additional change in soil carbon stock under certain farming practices compared with the business-as-usual practices. Accurately calculating cropland carbon budget and soil carbon credit is critical to assessing climate change mitigation potential in agroecosystems. The calculation of cropland carbon budget and soil carbon credit is sensitive to local soil and climatic conditions, especially initial soil organic carbon (SOC) stock, which is determined by both SOC concentration (SOC%) and bulk density (Bulk_Density). SOC stock data are either from soil sampling or gridded public survey data. In agroecosystem models, SOC stock data are a key model input for quantifying cropland carbon budget and soil carbon credit. However, various types and degrees of uncertainties exist in SOC stock datasets, which propagate to the quantification of SOC stock change. In particular, a large discrepancy is found in two widely used SOC stock datasets — Rapid Carbon Assessment dataset (RaCA) and Gridded Soil Survey Geographic Database (gSSURGO) — in the U.S. Midwest, with a relative difference (quantified using Normalized Root Mean Square Error, NRMSE) of 48.0% for 0–30 cm SOC stock between the two datasets. It remains largely unclear how uncertainty in SOC stocks affects the calculation of cropland carbon budget and soil carbon credit. To address this question, we used a well-validated process-based agroecosystem model, ecosys, to assess the impacts of SOC stock uncertainty on carbon budget and soil carbon credit calculation in the U.S. Midwestern corn-soybean rotation systems. Our results reveal the following findings: (1) A sizable discrepancy exists in simulated cropland carbon budget between using gSSURGO and using RaCA for their SOC% and Bulk_Density as model inputs, with a Pearson correlation coefficient (r) of only 0.4 for simulated change of SOC stock (ΔSOC) using these two different soil datasets. (2) Simulated cropland carbon budget components were more sensitive to initial SOC% than to Bulk_Density. For example, the upper and lower quartiles of multi-year averaged ΔSOC were −29.8 and 4.8 gC/m2/year for the selected counties respectively, with an uncertainty of 13.7 and 0.7 gC/m2/year induced by uncertainties in initial SOC% and Bulk_Density, respectively. (3) Both simulated ΔSOC and its uncertainty were negatively correlated with initial SOC%, whereas ΔSOC was negatively correlated with air temperature, and ΔSOC uncertainty was positively correlated with air temperature. (4) The uncertainty of calculated soil carbon credits was much smaller compared with the uncertainty of calculated absolute carbon budgets assuming the same SOC stock uncertainty level in the inputs. Specifically, in our assessment comparing planting cover crops vs no cover crop, the uncertainty of calculated soil carbon credits induced by initial SOC% uncertainty was less than 4% (relative to the quantified value of the soil carbon credits) for 90% of the cases. Our analysis highlights that high accuracy measurement of SOC% as inputs is needed for the calculation of cropland carbon budgets; however, soil carbon credit quantification is much less sensitive to the initial SOC% inputs, and the current publicly available soil datasets (e.g., gSSURGO) are largely suitable for the calculation of soil carbon credits.

Nature-based Climate Solutions (NbCS) are managed alterations to ecosystems designed to increase carbon sequestration or reduce greenhouse gas emissions. While they have growing public and private support, the realizable benefits and unintended consequences of NbCS are not well understood. At regional scales where policy decisions are often made, NbCS benefits are estimated from soil and tree survey data that can miss important carbon sources and sinks within an ecosystem, and do not reveal the biophysical impacts of NbCS for local water and energy cycles. The only direct observations of ecosystem-scale carbon fluxes, for example, by eddy covariance flux towers, have not yet been systematically assessed for what they can tell us about NbCS potentials, and state-of-the-art remote sensing products and land-surface models are not yet being widely used to inform NbCS policymaking or implementation. As a result, there is a critical mismatch between the point- and tree-scale data most often used to assess NbCS benefits and impacts, the ecosystem and landscape scales where NbCS projects are implemented, and the regional to continental scales most relevant to policymaking. Here, we propose a research agenda to confront these gaps using data and tools that have long been used to understand the mechanisms driving ecosystem carbon and energy cycling, but have not yet been widely applied to NbCS. We outline steps for creating robust NbCS assessments at both local to regional scales that are informed by ecosystem-scale observations, and which consider concurrent biophysical impacts, future climate feedbacks, and the need for equitable and inclusive NbCS implementation strategies. We contend that these research goals can largely be accomplished by shifting the scales at which pre-existing tools are applied and blended together, although we also highlight some opportunities for more radical shifts in approach.

Recent advances in satellite observations of solar-induced chlorophyll fluorescence (SIF) provide a new opportunity to constrain the simulation of terrestrial gross primary productivity (GPP). Accurate representation of the processes driving SIF emission and its radiative transfer to remote sensing sensors is an essential prerequisite for data assimilation. Recently, SIF simulations have been incorporated into several land surface models, but the scaling of SIF from leaf-level to canopy-level is usually not well-represented. Here, we incorporate the simulation of far-red SIF observed at nadir into the Community Land Model version 5 (CLM5). Leaf-level fluorescence yield was simulated by a parametric simplification of the Soil Canopy-Observation of Photosynthesis and Energy fluxes model (SCOPE). And an efficient and accurate method based on escape probability is developed to scale SIF from leaf-level to top-of-canopy while taking clumping and the radiative transfer processes into account. SIF simulated by CLM5 and SCOPE agreed well at sites except one in needleleaf forest (R 2 > 0.91, root-mean-square error  0.68). At the global scale, simulated SIF generally captured the spatial and seasonal patterns of satellite-observed SIF. Factors including the fluorescence emission model, clumping, bidirectional effect, and leaf optical properties had considerable impacts on SIF simulation, and the discrepancies between simulate d and observed SIF varied with plant functional type. By improving the representation of radiative transfer for SIF simulation, our model allows better comparisons between simulated and observed SIF toward constraining GPP simulations.

Agricultural nitrous oxide (N2O) emission accounts for a non-trivial fraction of global greenhouse gas (GHG) budget. To date, estimating N2O fluxes from cropland remains a challenging task because the related microbial processes (e.g., nitrification and denitrification) are controlled by complex interactions among climate, soil, plant and human activities. Existing approaches such as process-based (PB) models have well-known limitations due to insufficient representations of the processes or uncertainties of model parameters, and due to leverage recent advances in machine learning (ML) a new method is needed to unlock the "black box"to overcome its limitations such as low interpretability, out-of-sample failure and massive data demand. In this study, we developed a first-of-its-kind knowledge-guided machine learning model for agroecosystems (KGML-ag) by incorporating biogeophysical and chemical domain knowledge from an advanced PB model, ecosys, and tested it by comparing simulating daily N2O fluxes with real observed data from mesocosm experiments. The gated recurrent unit (GRU) was used as the basis to build the model structure. To optimize the model performance, we have investigated a range of ideas, including (1) using initial values of intermediate variables (IMVs) instead of time series as model input to reduce data demand; (2) building hierarchical structures to explicitly estimate IMVs for further N2O prediction; (3) using multi-task learning to balance the simultaneous training on multiple variables; and (4) pre-training with millions of synthetic data generated from ecosys and fine-tuning with mesocosm observations. Six other pure ML models were developed using the same mesocosm data to serve as the benchmark for the KGML-ag model. Results show that KGML-ag did an excellent job in reproducing the mesocosm N2O fluxes (overall r2Combining double low line0.81, and RMSECombining double low line3.6g€¯mgNm-2d-1 from cross validation). Importantly, KGML-ag always outperforms the PB model and ML models in predicting N2O fluxes, especially for complex temporal dynamics and emission peaks. Besides, KGML-ag goes beyond the pure ML models by providing more interpretable predictions as well as pinpointing desired new knowledge and data to further empower the current KGML-ag. We believe the KGML-ag development in this study will stimulate a new body of research on interpretable ML for biogeochemistry and other related geoscience processes.

As one of the major agricultural production areas in the world, the United States (U.S.) Midwest plays a vital role in the global food supply and agricultural ecosystem services. Although significant efforts have been made in modeling the carbon cycle dynamics over this area, large uncertainty still exists in the previous simulations in terms of reproducing individual components of the carbon cycle and their responses to environmental variability. Here we evaluated the performance of an advanced agroecosystem model, ecosys, in simulating carbon budgets over the U.S. Midwest, considering both the magnitude of carbon flux/yield and its response to environmental (climate and soil) variability. We conducted model simulations and evaluations at 7 cropland eddy-covariance sites as well as over 293 counties of Illinois, Indiana, and Iowa in the U.S. Midwest. The site-level simulations showed that ecosys captured both the magnitude and seasonal patterns of carbon fluxes (i.e., net ecosystem carbon exchange (NEE), ecosystem gross primary production (GPP), and ecosystem respiration (Reco)), leaf area index (LAI), and dynamic plant carbon allocation processes, with R2 equal to 0.92, 0.87, 0.87, and 0.78 for GPP, NEE, Reco, and LAI, respectively across all the sites compared with the observations. For regional scale simulations, ecosys reproduced the spatial distribution and interannual variability of corn and soybean yields with the constraints of observed yields and a new remotely sensed GPP product, with R2 of multi-year averaged simulated and observed yield equal 0.83 and 0.80 for corn and soybean, respectively. The simulated responses of carbon cycle dynamics to environmental variability were consistent with that from the empirical observations at both site and regional scales. Our results demonstrated the applicability of ecosys in simulating the carbon cycle and soil carbon dynamics of the U.S. Midwestern agroecosystems under different climate and soil conditions.

Data from thousands of past and present eddy covariance flux stations are available across the globe, while multiple hundreds actively operating as individual process-level studies, small flux networks dedicated to specific research goals, and larger national and continental networks with broad ecological and environmental foci.Many flux stations have weather and soil data to help clean, analyze and interpret the fluxes but most do not have optical proximal sensors, do not allow straightforward coupling with remote sensing (drone, aircraft, satellite, etc.) data, and cannot easily be used for validation of remotely sensed products, ecosystem modeling, or upscaling from field to regional levels. The flux source areas themselves (e.g., flux footprints) are typically not defined in the flux datasets, and the time stamps of the fluxes come in a large number of outdated non-trackable formats. Finally, the past ways of the flux data quality control, analysis and interpretation require a participation of micrometeorological expert (or an entire network) with their own custom codes or exceptional skills in using existing software such as MatLab or VB Tools in Excel. These are the key issues effectively preventing a larger environmental research community and remote sensing community from fully utilizing eddy covariance flux data.In 2016-2020, a set of new tools to collect, process, analyze, time- and space- allocate and share time-synchronized flux data from multiple flux stations were developed and deployed globally. These new tools can be effective in solving most or all of the key issues listed above. The fully automated FluxSuite system combines hardware, software and web services, and does not require an expert to run it. It can be incorporated into a new flux station or added to a present station, using a weatherized remotely-accessible microcomputer, SmartFlux3 which utilizes EddyPro software to calculate fully-processed fluxes in near-real-time, alongside biomet data and flux footprints. All data are merged into a single quality-controlled file timed using PTP time protocol. Remote sensing researchers and modelers without actual physical stations can form “virtual networks” of actual stations by collaborating with tower PIs from different physical networks and flux databases.The very latest development in this overall approach is the flux data analysis software, Tovi, designed to seamlessly ingest the data from the flux stations and to allow a non-micrometeorologist to quality control, analyze and interpret the flux data. It allows rapid execution of the QC/QA and data analysis steps which have been time-consuming and complicated in the past, and other data analysis steps virtually not doable in the past, all using interactive and intuitive GUI, including advanced footprint calculations and flux apportioning necessary for remote sensing community; NEE flux partitioning; automated generation specific lists of references for each workflow; etc.This presentation will show how combinations of these new tools are used by major networks, and describe how this approach can be utilized for matching remote sensing and tower data for ground truthing, improve scientific interactions, and promote a better utilization of the eddy covariance flux data by a wider environmental research community.

Data from thousands of past and present eddy covariance flux stations are available across the globe, while multiple hundreds actively operating as individual process-level studies, small flux networks dedicated to specific research goals, and larger national and continental networks with broad ecological and environmental foci.Many flux stations have weather and soil data to help clean, analyze and interpret the fluxes but most do not have optical proximal sensors, do not allow straightforward coupling with remote sensing (drone, aircraft, satellite, etc.) data, and cannot easily be used for validation of remotely sensed products, ecosystem modeling, or upscaling from field to regional levels. The flux source areas themselves (e.g., flux footprints) are typically not defined in the flux datasets, and the time stamps of the fluxes come in a large number of outdated non-trackable formats. Finally, the past ways of the flux data quality control, analysis and interpretation require a participation of micrometeorological expert (or an entire network) with their own custom codes or exceptional skills in using existing software such as MatLab or VB Tools in Excel. These are the key issues effectively preventing a larger environmental research community and remote sensing community from fully utilizing eddy covariance flux data.In 2016-2020, a set of new tools to collect, process, analyze, time- and space- allocate and share time-synchronized flux data from multiple flux stations were developed and deployed globally. These new tools can be effective in solving most or all of the key issues listed above. The fully automated FluxSuite system combines hardware, software and web services, and does not require an expert to run it. It can be incorporated into a new flux station or added to a present station, using a weatherized remotely-accessible microcomputer, SmartFlux3 which utilizes EddyPro software to calculate fully-processed fluxes in near-real-time, alongside biomet data and flux footprints. All data are merged into a single quality-controlled file timed using PTP time protocol. Remote sensing researchers and modelers without actual physical stations can form “virtual networks” of actual stations by collaborating with tower PIs from different physical networks and flux databases.The very latest development in this overall approach is the flux data analysis software, Tovi, designed to seamlessly ingest the data from the flux stations and to allow a non-micrometeorologist to quality control, analyze and interpret the flux data. It allows rapid execution of the QC/QA and data analysis steps which have been time-consuming and complicated in the past, and other data analysis steps virtually not doable in the past, all using interactive and intuitive GUI, including advanced footprint calculations and flux apportioning necessary for remote sensing community; NEE flux partitioning; automated generation specific lists of references for each workflow; etc.This presentation will show how combinations of these new tools are used by major networks, and describe how this approach can be utilized for matching remote sensing and tower data for ground truthing, improve scientific interactions, and promote a better utilization of the eddy covariance flux data by a wider environmental research community.

To address the global challenge of increasing the efficiency of food production in an era of climate change, integrated research frameworks based on multiple performance criteria are required to promote synergies while minimizing tradeoffs. In this study, we investigated spatial and temporal trends in yield gaps, resource use efficiencies for nitrogen and water, and yield stability at the county-level in the U.S. Corn Belt, the world's largest maize production region. Our results provide an integrated assessment of regions with higher vs. lower relative performance across indicators, while also revealing high spatial variability in relationships between yield gaps and nitrogen and water use efficiencies. Despite overall positive trends in maize production, there was evidence of increasing disparity over time, with more improvements occurring in highly productive areas. This study provides a holistic framework for balancing efforts to increase crop production, resource use efficiencies, and yield stability at the regional scale.

An exponential rise in the atmospheric vapour pressure deficit (VPD) is among the most consequential impacts of climate change in terrestrial ecosystems. Rising VPD has negative and cascading effects on nearly all aspects of plant function including photosynthesis, water status, growth and survival. These responses are exacerbated by land–atmosphere interactions that couple VPD to soil water and govern the evolution of drought, affecting a range of ecosystem services including carbon uptake, biodiversity, the provisioning of water resources and crop yields. However, despite the global nature of this phenomenon, research on how to incorporate these impacts into resilient management regimes is largely in its infancy, due in part to the entanglement of VPD trends with those of other co‐evolving climate drivers. Here, we review the mechanistic bases of VPD impacts at a range of spatial scales, paying particular attention to the independent and interactive influence of VPD in the context of other environmental changes. We then evaluate the consequences of these impacts within key management contexts, including water resources, croplands, wildfire risk mitigation and management of natural grasslands and forests. We conclude with recommendations describing how management regimes could be altered to mitigate the otherwise highly deleterious consequences of rising VPD. Summary statement: Rising atmospheric vapour pressure deficit (or VPD) is one of the most widespread and significant consequences of climate warming for terrestrial ecosystems. This article reviews the mechanistic bases of these usually deleterious impacts and synthesises that information into a set of management recommendations to mitigate them. [ABSTRACT FROM AUTHOR]

Accurate estimation of terrestrial photosynthesis has broad scientific and societal impacts. Measurements of photosynthesis can be used to assess plant health, quantify crop yield, and determine the largest CO₂ flux in the carbon cycle. Long-term and continuous monitoring of vegetation optical properties can provide valuable information about plant physiology. Recent developments of the remote sensing of solar-induced chlorophyll fluorescence (SIF) and vegetation spectroscopy have shown promising results in using this information to quantify plant photosynthetic activities and stresses at the ecosystem scale. However, there are few automated systems that allow for unattended observations over months to years. Here we present FluoSpec 2, an automated system for collecting irradiance and canopy radiance that has been deployed in various ecosystems in the past years. The instrument design, calibration, and tests are recorded in detail. We discuss the future directions of this field spectroscopy system. A network of SIF sensors, FluoNet, is established to measure the diurnal and seasonal variations of SIF in several ecosystems. Automated systems such as FluoSpec 2 can provide unique information on ecosystem functioning and provide important support to the satellite remote sensing of canopy photosynthesis.