UAVs and Data Fusion Concepts for Livestock and Fodder Monitoring - Supporting Resilience in Arid Landscapes in Northern Kenya
The topic is open and also suitable for GEM students in track 3 – GEM for Ecosystems & Natural Resources.
Drones or UAVs are revolutionizing conservation and agricultural practices, offering cost-effective and safe methods for data collection over large, challenging areas. In the arid and semi-arid lands (ASALs) of northern Kenya, drones can provide a unique opportunity to monitor livestock, wildlife, and fodder resources with unprecedented detail. These insights are critical for supporting pastoralist livelihoods, conserving wildlife, and ensuring sustainable land use under the increasing pressures of climate change. Utilizing UAVs to supplement or replace existing crewed aerial surveys, this research focuses on enhancing the detection of livestock, and fodder resources.
Fixed-wing UAVs/drones with long-range capabilities will be deployed to fly transects over extensive areas, capturing high-resolution imagery using RGB cameras. These systems may optionally include thermal or multispectral sensors to enrich data collection. The primary objective is to use the collected data to train machine learning models capable of automatically detecting livestock or assessing fodder resources. These models can efficiently analyse large datasets, offering population estimates and resource assessments.
To scale these findings spatially and temporally, data fusion techniques can be employed by integrating UAV data with satellite imagery. High-resolution satellite imagery, such as PlanetScope can complement UAV data for large-scale analysis. Thermal EO data from systems like ECOSTRESS or the upcoming LSTM mission can be utilized for detecting large herd enclosures (bomas) and be particularly useful for understanding vegetation health and fodder availability. Spectral information from frequent fine-resolution acquisitions allows for the detection of enclosure locations and enables fodder monitoring by identifying vegetation types, estimating biomass, and tracking changes in pasture quality over time.
For enhanced applications, statistical models can spatially extrapolate UAV data across larger regions, leveraging satellite data to account for areas not directly surveyed. By combining UAV data with remote sensing platforms through data fusion approaches, it becomes possible to scale insights efficiently and cost-effectively for broader monitoring applications.
This MSc project is embedded in the RANGE(Resilient Approaches in Natural ranGeland Ecosystems) Project , which supports resilience in rangeland ecosystems. The research will be conducted in close collaboration with local partners, integrating on-the-ground knowledge and aligning with ongoing initiatives to build a comprehensive understanding of ASAL dynamics.
This project aims to harness UAV technology and data fusion concepts to address critical challenges in ASAL regions. The objectives and corresponding methods are outlined below:
- Evaluate UAV capabilities for monitoring livestock, wildlife, or fodder resources.
This includes a literature review on the use of UAVs/drones for livestock detection or fodder resource assessment and identification of knowledge gaps in monitoring methods and potential improvements.
- Develop and validate machine learning models for analysing UAV imagery.
Train machine learning models to detect livestock and bomas, or fodder from high-resolution UAV images and test and refine the models based on ground-truth data.
- Scale insights for broader applications.
Use data fusion or statistical models to extrapolate livestock population densities and integrate UAV data based findings with satellite data to map fodder resources at larger scales.
Temuulen Ts. Sankey, Jackson M. Leonard, Margaret M. Moore, Unmanned Aerial Vehicle−Based Rangeland Monitoring: Examining a Century of Vegetation Changes, Rangeland Ecology & Management, Volume 72, Issue 5, 2019, Pages 858-863, ISSN 1550-7424, https://doi.org/10.1016/j.rama.2019.04.002.
Anton Vrieling, Francesco Fava, Sonja Leitner, Lutz Merbold, Yan Cheng, Teopista Nakalema, Thomas Groen, Klaus Butterbach-Bahl, Identification of temporary livestock enclosures in Kenya from multi-temporal PlanetScope imagery, Remote Sensing of Environment, Volume 279, 2022, 113110, ISSN 0034-4257, https://doi.org/10.1016/j.rse.2022.113110.
Emilien Alvarez-Vanhard, Thomas Corpetti, Thomas Houet, UAV & satellite synergies for optical remote sensing applications: A literature review, Science of Remote Sensing, Volume 3, 2021, 100019, ISSN 2666-0172, https://doi.org/10.1016/j.srs.2021.100019.
Ample scope for adaptation exists. This topic encompasses collecting, processing, and analysing spatial data from UAVs, while integrating it with remote sensing, satellite data and geostatistical approaches. Depending on your interests, there is room to focus more on biophysical (fodder resources) or ecological (livestock distribution) aspects. Collaboration with local partners ensures the research has practical and policy-relevant outcomes.