This project investigated the food desert (FD) phenomena in a way that integrates human and natural systems, with a focus on the supply and demand of fresh vegetables in North Carolina (NC). A FD is a geographic area characterized by both low income and low access to healthy food. We employed various geospatial techniques and agent-based modeling approaches. Multiple spatial and temporal databases were geo-coded for analysis and modeling including the Neilson Database (consumer panel data), InfoUSA, NC OneMap, U.S Census, NC DOT road network, NLCD and CDL, NCCPI, NOAA climate data and USGS topography and soil data.

We conducted a systematic review of the literature which identified the driving factors and potential policies to mitigate the impact of FDs. While socioeconomic factors make up most of the list, some natural factors such as soil health, water quality, and weather events can exacerbate the occurrences of FDs. The social factors are primarily due to land use choices that have historically and systematically been racially motivated.

We used GIS tools to assess alternative techniques for estimating the driving time and distance from residential locations to destinations (e.g., grocery stores). An analysis in Guilford County, NC, showed that alternative approximations of more than 200,000 residential locations did not affect the finding that the FD status, driving distance, and driving time do not contribute to fresh vegetable consumption meaningfully.

Land use analysis using the geo-coded data showed that fresh vegetable production in three study areas in NC is limited and concentrated in only a few spots. The estimated first-order Markov chain models pointed that neither vegetable nor fallow land use have a high probability of remaining the same cover the following year. Out of all the land parcels that had vegetables in any given consecutive six years, the overwhelming majority had vegetables planted for no more than three years.

Land suitability analysis was conducted at regional scales in three FD-prone regions of NC combined geospatial analysis with a multi-criteria evaluation. We showed that the outcome provides useful information on building community food capacity from network perspectives and land use configuration. For example, the analysis partitioned Guilford County in highly suitable (1%), suitable (25%), moderately suitable (64%), and unsuitable (10%).

The analysis of consumption patterns in alternative food access environments revealed that fresh vegetable demand in the FDs is statistically significantly less than the demand in the non-FDs. Fresh vegetable consumption is statistically significantly associated with socioeconomic statuses such as household income, age, and race.

An agent-based model (ABM) was developed using factors contributing to FDs, including natural systems (i.e., land use and land-cover change) and human systems (i.e., production and consumption). The purpose of the ABM is to explore and simulate the role of main FD indicators and to provide the tool for the use by individuals, communities, policymakers, and other agencies. The fresh produce and vegetables production ABM component simulates the farmers? or households? decisions to make land use changes based on the criteria such as rainfall, land size, soil types, production seasons, and income. The consumption ABM component simulates the role of main factors that limit access to fresh fruits and vegetables, including proximity to supermarkets or food stores, income, unemployment rate, vehicle ownership, and SNAP aid. Using NC as a case study, we developed an approach to integrate the production and consumption. Modeling this linkage is essential to understanding how consumer and producer interactions led to sufficient healthy food access to overcome food insecurity issues.

The outreach activities included more than 20 training workshops organized by the Center for Environmental Farming Systems (CEFS); a symposium at the North Carolina A&T State University with a team of PIs, student researchers, and others including members of the advisory board; exhibition booth at the NC Minority Farm and Landowners Conference; and interdisciplinary grant writing workshop at an international applied economics conference. The project provided an interdisciplinary platform to foster a collaborative working relationship between the PIs, and an opportunity to train future professionals including 2 postdocs, 7 graduate students (3 Ph.D. and 4 MS), and 6 undergraduate researchers, the majority of which are from underrepresented backgrounds. The activities resulted in at least 3 MS theses and a few partial dissertation chapters. The results were presented at AAEA, AGU, and ASPRS, in the form of oral and poster presentations. The project resulted in 21 presentations, 4 conference proceedings, one book chapter, and 7 high quality peer-reviewed journals publications, with a few under preparation. The results outcomes were also utilized in classroom teaching including a seminar course, an advanced GIS course, and an applied geostatistics course. The efforts resulted in several grant awards totaling over $14.8 million in eight federal grants including USDA, NSF and NASA. More information can be found in annual reports, and on the project website: https://sites.google.com/view/ncfdmodeling

Last Modified: 05/03/2023
Modified by: Manoj K Jha