Performing Department
(N/A)
Non Technical Summary
Rationale and SignificanceCOVID-19 underscored the need to understand relationships between resiliency and sustainability. Existing models are insufficient, but we can enhance the existing Food System Supply-chain Sustainability (FoodS3) model developed by Co-Is Schmitt and Pelton to make it sufficient. Some of these updates will come from cutting edge data science/ML. Then with our new model we look at both resilience and sustainability. The new modeling framework will ultimately achieve multiple critical objectives: deep insight on APS structural elements that lead to more resilience and improved system performance in response to future shocks (COVID-variants, Cyber-attacks, extreme weather, animal disease outbreaks, etc.); better policy (national, state, local); investment strategies for major supply chain companies; innovation opportunities for start-up's; enhanced and more equitable consumer access to high quality protein, with nutrition and public health benefits; and producer income stability.Our project integrates Data Science with environmental LCA through enhanced understanding (via Foods3 modeling) of vulnerabilities and responses of the animal protein system to exogenous shocks.Design, validate and implement new algorithms and methods for depicting and leveraging massive data.Develop data-integration and data-quality algorithms to improve analytic capability.Create new methodologies and frameworks for tracking and processing data.Develop decision-support tools that use diverse data sources and Big Data analytics modeling of short-term impacts of various factors to create best value to the U. S. agricultural enterprise.Our long-term goal is to develop an innovative data-based modeling framework for evaluation of interventions intended to mitigate tradeoffs between resilience and sustainability (defined broadly to include socioeconomic factors). Stakeholder-input on important problem definition issues, including the identification of relevant data sources at appropriate spatial and temporal scales, must be gathered, in order to ensure that the resultant modeling framework is measuring APS performance and tradeoffs in a meaningful way. As detailed below, this input will be gathered as the primary deliverable of Objective 1. Under Objective 2, we will develop new advances in data science that will make it possible to enhance the capabilities of FoodS3 to model APS supply chains and to simulate supply and demand shocks. Objective 3 is focused on the advancement of data science capabilities to measure and monitor farm management practices deployed in the APS, including from crop and livestock production, and again implement these within FoodS3 to improve spatially explicit environmental impact models. Objective 4 involves expansion of FoodS3 capabilities to report on regional and demographic patterns in socioeconomic response to the COVID shock. Lastly, under Objective 5, the new modeling framework will be used to simulate APS response to interventions that are proposed by stakeholders for mitigating resilience and sustainability tradeoffs in the APS
Animal Health Component
0%
Research Effort Categories
Basic
(N/A)
Applied
100%
Developmental
(N/A)
Goals / Objectives
Our long-term goal is to develop an innovative data-based modeling framework for evaluation of interventions intended to mitigate tradeoffs between resilience and sustainability (defined broadly to include socioeconomic factors) as animal protein food systems continue to reduce their environmental impacts, which appears to be having the unintended consequence of enhancing certain system vulnerabilities to shocks. We have defined five objectives in support of this goal:Collect stakeholder input on data sources and metrics appropriate for quantifying the resilience and sustainability of the US animal protein food system (APS)Make enhancements (using AI & ML as appropriate) to the input data streams of the current Food System Supply-chain Sustainability (FoodS3) model such that it has sufficient temporal resolution (monthly) and scope of APS food types to effectively characterize the resilience and sustainability behavior of the APS.Advance data science capabilities to measure and monitor APS environmental impacts, adding into Foods3 where possible.Analyze regional and demographic patterns in the response of data-based APS socioeconomic metrics to the COVID shock, and prioritize these metrics for future addition to FoodS3.Demonstrate use of the new modeling framework by simulating APS response to interventions that are proposed by stakeholders for mitigating resilience and sustainability tradeoffs in the APS.Achieving our long-term goal will make it possible for policy-makers and others to quantitatively explore and evaluate the benefits of interventions intended to mitigate the tradeoffs between resilience and sustainability in the APS, which is clearly responsive to the overall remit of AFRI. In addition, the project will provide notable advances in understanding of the APS, arguably the most important (in terms of economic, environmental, nutrition, and societal impact) of all of the components of the US Food System - and it will also significantly advance the field of data science, by virtue of its integrated application of remote sensing, AI, and ML to characterization of APS animal-handling facilities.
Project Methods
Stakeholder engagement:We will convene a stakeholder event about mid-way through the first year of the project where we will gather broad input on data sources and metrics relevant to the performance of the APS from both a resilience and a sustainability perspective. As noted above, our working definition of resilience is well-aligned with the one proposed previously (Schipanski et al., 2016) and simply refers to the capacity for continued production and equitable access to nutritious food over time and space in the face of shocks. For sustainability, we will propose, for stakeholder feedback a broad definition that encompasses not only environmental aspects (e.g., carbon, land, and water footprints) but also socioeconomic metrics, such as jobs, income, equitable access to nutritious food, etc. The facilitated discussions will include level-setting plenary presentations and ample use of breakout sessions in order to ensure that all perspectives are heard and recorded. Outcomes will be summarized, shared with event participants, and directly incorporated into the subsequent modeling efforts within the project.We will have follow-up stakeholder meetings during the course of the project to ensure relevance of the project direction (vis. interventions and tradeoffs).Near the conclusion of the project, we will convene a final stakeholder event to present the project findings and gather feedback on the results.Following the final stakeholder engagement event on project findings, the team will implement a mechanism for sustained stakeholder interaction, most likely via the current PRICE Initiative, where one of our team members serves on the Scientific Committee.Enhance Supply Chain Modeling Capabilities of FoodS3Objectives, Advances in Data Science, and OutcomesObjective 2: Supply Chain ModelAdvances in Data Science: Pairing high spatial and temporal resolution (3m; daily) optical satellites with new machine learning algorithms to track and monitor facility-scale protein processing outputs and bottlenecksOutcome: Refined estimates of 1000 processing facility weekly production, and creation of real-time monitoring systemAdvances in Data Science: Advanced geospatial regression models trained on census, survey, and cropland raster dataOutcome: Monthly livestock supply and demand estimates at county and processing facility nodes of supply chainObjective 3: Environmental Impact ModelAdvances in Data Science: ML algorithms (e.g., convolutional neural nets and reinforcement learning) integrating high resolution optical satellite data and methane sensor satellites with process based LCA model outputs for manure management classificationOutcome: Spatial maps of CAFOs and the respective manure management systems for improved spatial specificity of livestock production environmental modelsAdvances in Data Science: Self-supervised ML algorithms using optical sensors detecting soil reflectance tuned to detect limestone depositsOutcome: County maps of lime application rates for improved spatial specificity of crop production environmental modelsObjective 4: Socioeconomic metricsAdvances in Data Science: Sub-national structural path analysis algorithms for informing spatially explicit supply chain models and socioeconomic impactsOutcome: Spatially explicit supply node-specific estimates of income and job creation outcomes from shock and intervention scenarios