University of Twente Student Theses
Integrating environmental data with satellite imagery for large-scale crop grain nutrient mapping
Ofori-Karikari, Kwasi Appiah-Gyimah (2024) Integrating environmental data with satellite imagery for large-scale crop grain nutrient mapping.
PDF
5MB |
Abstract: | Food Composition Tables (FCTs), which are widely used in food security analysis, offer country-level snapshots of food nutritional quality. FCTs are highly uncertain because crop nutrient concentrations vary over space and time. Little progress has been made to exploit satellite data for assessing the nutritional status of crop yield. Sentinel-1 and Sentinel-2 have emerged as invaluable data sources for agricultural monitoring from space, because they track crop growth and development at high (10-20m) spatial resolution through time. This study integrates Sentinel-1, Sentinel-2, and other geospatial information across Ethiopia in 2018 into a Random Forest to predict nutrient concentrations in grain yield for important global staple crops. The model was trained and evaluated with data from field measurements collected over the same period. Results were promising: Cu in Sorghum (R² = 0.73), Mn in Teff (R² = 0.70), Mg in Teff (R² = 0.64), Cu in Maize (R² = 0.60), Mg in Wheat (R² = 0.58), Mn in Wheat (R² = 0.58), Mg in Sorghum (R² = 0.57), and Fe in Sorghum (R² = 0.54). The findings underscore the effectiveness of Sentinel-2 narrowband vegetation indices and soil properties for nutrient analysis. The results obtained in this study have the potential to improve the availability of crop nutrient data across large spatial areas. |
Item Type: | Essay (Master) |
Faculty: | ITC: Faculty of Geo-information Science and Earth Observation |
Subject: | 43 environmental science |
Programme: | Geoinformation Science and Earth Observation MSc (75014) |
Link to this item: | https://purl.utwente.nl/essays/102250 |
Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Repository Staff Only: item control page