Kylen Solvik Receives NASA FINESST Award
The first stage of the project is to use NASA/USGS Landsat satellite data and AI computer vision methods to automatically identify human-made reservoirs and distinguish them from natural water bodies. Small reservoirs are ubiquitous in Brazilian agriculture, but these will be the first comprehensive maps of very small reservoirs (<0.5 ha), helping analyze this under-studied impact of agriculture on water in the region. The spatial distribution of reservoirs over time will be modeled over time in relation to land-use/land-cover change using existing data from MapBiomas Brasil.
The machine learning and remote sensing part of the research will be paired with qualitative interviews with farmers and ranchers in eastern Mato Grosso, Brazil about water use and management, particularly in response to climate change. Due to climate change, the rainy season that much of Brazil's agriculture depends on is becoming shorter and less predictable, driving farmers to rely more on reservoirs to store and access water. The potential impacts of these changes are not well understood. By combining both remote sensing and qualitative field interviews, this project will help researchers, policy-makers, and water users more fully understand the past, present, and future impacts of agriculture on surface water in Brazil.