Image
agricultural field
Projects

Scale Sustainable Agriculture

Creating open source tools to depict sustainable agriculture’s economic and environmental benefits and developing a public database for studying crop yield and stability

The Problem


Feeding the world’s population requires massive amounts of agricultural resources. These same resources also substantially contribute to climate change and biodiversity loss. Pesticides, herbicides, and fertilizers (along with other industrial practices) can negatively impact water pollution and soil erosion, and increase greenhouse gas emissions and threats to extreme climate events. Sustainable food production is an alternative to industrial agriculture and is essential to a future with worsening growing conditions due to climate change.

 

The Opportunity


Sustainable agricultural practices include increasing crop diversity, crop rotation, and cover cropping. These practices reduce greenhouse gas emissions, support habitat and species diversity, prevent soil erosion, improve water quality, and make farmlands more resilient to climate change. Despite these advantages, sustainable agriculture is not currently implemented at scale, often due to systemic regulatory issues and financial hurdles. 

 

Key Highlights

  • Creating an AI tool for determining climate impacts to crop yield and yield stability.
  • Uncovered compelling federal policy opportunities to incentivize sustainable agriculture.
  • Developing a public database for studying crop yield, stability, and other key agricultural applications.

 

Our Impact

 

Climate Stability Tools


DSE is developing an open source tool for crop insurance and agricultural finance agents to directly quantify and better understand sustainable agriculture’s benefits, so that they can support farmers in adopting these practices. 

 

The tool uses AI to simulate many thousands of possible agricultural outcomes under climate change for the next three decades. We find that farmers’ and insurance companies’ financial stabilities are further threatened as crop losses become increasingly frequent and devastating from climate change. Without action, the claims rate of the examined crop insurance program will unsustainably double by 2050.

 

We depict our projections under expected further global warming (intermediate warming scenario - IPCC SSP245) versus a “baseline,” which assumes today’s growing conditions continue into the future unchanged
Above: We depict our projections under expected further global warming (intermediate warming scenario - IPCC SSP245) versus a “baseline,” which assumes today’s growing conditions continue into the future unchanged (Pottinger et al 2024 preprint).



 

Our simulations also uncover that specific adjustments to the Farm Bill could further enable the United States Department of Agriculture’s Risk Management Agency (which manages the federal crop insurance program) to optionally recognize the benefits of sustainable agricultural practices for US food system.

 

Crop Yield & Stability Tools

 

We are using data from NASA's Landsat satellites to develop a public database tracking over 14,000 points in corn and soy fields in the midwestern United States. The data will be useful for studying and monitoring crop yield, yield-stability, soil health, cover-cropping, and other sustainable agricultural practices. We hope that the data will help empower and accelerate research and action in the agricultural field more broadly.  We are releasing an open-source codebase so that researchers can quickly generate new databases for their own locations and metrics of interest (project description available here).

 

Schmidt DSE’s Spectral Trends Database monitors over 14,000 corn and soy fields in the midwestern United States from 2000 to present day
Schmidt DSE’s Spectral Trends Database monitors over 14,000 corn and soy fields in the midwestern United States from 2000 to present day. The above chart overlays biomass yield and the “Specific Leaf Area Vegetation Index” (SLAVI) for a field from 2000 to 2012. SLAVI is one of 36 vegetation indices tracked by our system. The bottom row shows True-Color and SLAVI Landsat imagery from both before (the spikes in the chart) and after harvest.

DSE Contributors

  • Image
    Maya Weltman-Fahs

    Maya Weltman-Fahs

    Senior Program Manager
    Eric and Wendy Schmidt Center for Data Science & Environment at Berkeley
  • Image
    Sam Pottinger

    Sam Pottinger

    Senior Research Data Scientist / Software Engineer
    Eric and Wendy Schmidt Center for Data Science & Environment at Berkeley