NASA Harvest is Innovating Digital Technology to Monitor and Analyze Global Agriculture
The success of local and global food systems, from crop production to supply chains, both impacts and is impacted by a number of factors. Severe weather events; temperature and precipitation extremes; and conflict or other disruptions on the ground can all contribute to the success of a farmer’s crops. Likewise, the success or failure of local crops can have far reaching impacts from local economic conditions to global supply chain disruptions and price instability.
Mitigating issues as they arise requires effective, efficient, and transparent measurement, monitoring, and reporting (MMR).
As our mission at NASA Harvest is to improve methods for agricultural monitoring for the benefit of food supply, crop production, market stability, and human resilience; MMR is vital to our work.
Our international consortium is innovating in digital technologies to help farmers, policymakers, and scientists better measure and understand the conditions of global and local agricultural conditions and make informed decisions.
NASA Harvest Director, Dr. Inbal Becker-Reshef, recently co-authored a Letter in Nature Food on the developments in digital monitoring technologies over the past several years. The letter discusses a number of different technologies that the consortium uses, including remote sensing, big data and machine learning, and smartphones.
Big Earth Data Provides Key Insights
Traditional MMR often relies upon surveys and self reporting, allowing for under-reporting of undesirable practices and the possibility of biases and errors. Additionally, due to the cost of typical human collection methods, data is available for only a portion of farms and at specific times.
Remote sensing, or the collection of Earth observation data from constellations of satellites circling the globe, allows for data to be collected across almost the entire planet every few days.
Given the massive amounts of data produced through remote sensing, developing accurate machine learning models is necessary to properly analyze all of the Earth observation imagery being collected.
Taken together, remote sensing and machine learning are allowing farmers, policymakers, and scientists to see how cropland is expanding or shrinking; the health of crops throughout the growing season; impacts of severe weather events; and the use of specific farm management practices.
NASA Harvest is using remote sensing and machine learning to better understand global agricultural conditions and food supply in a variety of contexts. We’ve examined the impact on Iowan corn fields in the aftermath of a derecho; mapped an entire nation’s cropland in 10 days in response to the COVID-19 pandemic; located unexploded ordnances in conflicted-impacted cropland; and forecast crop yields in areas at war.
Smartphones: Entryway to the Information
Another piece of digital technology the authors discuss are smartphones. The increase in connectivity created by smartphones allows farmers to receive and pass information like never before. Increased access to smartphones with cameras, GPS capabilities, and internet access, equips farmers to easily learn about market prices for crops; incoming weather forecasts; and agricultural extension agent efforts. They can also communicate much easier with their neighbors to coordinate agricultural and market activities. Finally, farmers are also able to share photos, locations, and the conditions of their fields with policymakers and scientists. This allows for rapid understanding of local conditions in the aftermath of targeted events, for example.
NASA Harvest has utilized the availability of smartphones to help in our efforts to collect accurate training data for our models. One instance is our work with geospatial company ESRI, wherein NASA Harvest developed and launched an app that allows farmers to send in photos, GPS coordinates, and the condition of the cropland wherever they are. This data helps us to further improve our modeling efforts and produce more accurate cropland maps and yield forecasts for users.
This is not the only project NASA Harvest has created to improve our data uptake. Helmets Labelling Crops is an innovative project that provides training and GoPro cameras to data collectors in participating regions. These data collectors have GPS-enabled GoPro cameras attached to their helmets as they drive through agricultural areas on motorbikes. The video data is then analyzed with computer vision models to pick out cropland at specific locations and exported as model training data.
The widespread use of digital monitoring presents tremendous opportunities to better understand and analyze food systems at both local and global scales. NASA Harvest is proud to be developing new techniques that are creating a more food secure world.
You can read the letter in Nature Food here.