A framework for national-scale predictions of forage dry mass in Senegal: UAVs as an intermediate step between field measurements and Sentinel-2 images

• Published in: International journal of remote sensing 2024-07, Vol.45 (13), p.4199-4218
• Main Authors: Nungi-Pambu, Maïalicah, Lo, Adama, Fassinou, Cofélas, Tageson, Torbern, Fensholt, Rasmus, Diouf, Abdoul Aziz, Menassol, Jean Baptiste, Assouma, Mohammed Habibou, Toure, Ibra, Taugourdeau, Simon
• Format: Article
• Language: English
• Subjects: Data collection; Earth and Related Environmental Sciences; Geovetenskap och miljövetenskap; Image resolution; Life Sciences; Livestock; Livestock feed; Livestock feeds; Mapping; Monitoring; Natural Sciences; Naturgeografi; Naturvetenskap; Pastoralism; Physical Geography; Remote sensing; Satellite data; Satellite imagery; Satellite observation; Satellites; Spatial resolution; Unmanned aerial vehicles; Vegetation index; Woody plants
• ISSN: 0143-1161; 1366-5901
• DOI: 10.1080/01431161.2023.2290992

Monitoring available feed for livestock is a key factor for developing pastoralism in the Sahel, and satellite images has proven useful in monitoring dry mass on large spatial scales. This approach requires field measurements of dry mass (herbaceous and woody plants) to calibrate such models based on Earth observation data. However, the need for representative field measurements can be a challenge when considering the low spatial resolution of available satellite data. Unmanned Aerial Vehicles (UAV) can produce very high-resolution images, so we tested UAVs as an intermediate step between field measurements and satellite images, to bridge the difference in spatial scale. We used 43 orthomosaics from a red-green-blue (RGB) UAV sensor in combination with field measurements of herbaceous and woody dry biomass at sites located primarily in the northern/central and southernmost parts of Senegal. We developed a dry mass model trained with filed observed measurements to be applied on the UAV orthomosaics. The dry mass information obtained from these UAV maps was subsequently related to vegetation indices derived from Sentinel-2 data to produce a national-scale 10 m spatial resolution baseline map of herbaceous and woody dry mass for Senegal in 2020. We obtained a high correlation between dry mass derived from UAV and Sentinel-2 indices (R² = 0.91), suggesting a robust basis for national-scale mapping. Lastly, our map was compared with a state-of-the-art annual reference map based on satellite remote sensing. This comparison showed a difference of 21 million tons of dry mass at national level. We concluded that bridging the spatial gap between field and satellite observations using spatially representative UAV data collection is a cost-effective approach for accurate mapping of dry mass at national level using freely available Sentinel-2 satellite data.