Simon Wachira Muthee
- Water resources (management)
- Land use and food security
- Environmental and climate change
- Biodiversity
- Environmental Impact Analysis
- Agriculture, land use, climate change
- Kenya
Geopatial Engineering data colection and analysis, Environmental and Hydrological modelling, Programming
Developing a Novel Soil Moisture Content Index under the Influence of Land Use Using Deep Learning Techniques
Soil moisture content (SMC) is crucial for sustaining plant, animal, and microbial life in the soil. It governs nutrient distribution and surface processes such as erosion. SMC supports decisions on sustainable soil water management and droughts monitoring. Most developing economies face numerous challenges such as poverty, food insecurity, and changing climatic and Land use and Land Cover (LULC) conditions have worsened the situations leading to frequent and prolonged droughts. Thus, drastically reducing SMC and threatening food security. This study aims to develop a novel SMC which is a geospatial tool that simulates SMC variability as influenced by land use and provides short-term predictions for sustainable soil water management. The study will utilise the Muringato Basin, located in the Upper Tana River Basin, in Kenya, as the test site. The major data inputs will include remotely sensed (RS) datasets, such as Digital Elevation Models (DEMs), satellite imagery, and hydrometeorological data (precipitation, temperature, evapotranspiration, solar irradiance), ground-referenced data from SMC testing, LULC sample points, weather stations, and community social surveys. The study will employ the Extreme Gradient Boosting (XGBoost), an ensemble Machine Learning model to evaluate relationships between multiple factors affecting SMC and integrate its findings to the analysis of the social survey data. The integration will provide a social context to the model outputs and insights on community's practices toward sustainable soil water management. Additionally, the research will utilise the Recurrent Neural Network-Long Short Term Memory (RNN-LSTM), a deep learning algorithm to model and develop the novel index via integration the various datasets. The new index will be calibrated and validated with ground sampled SMC data to ensure its reliability. The index will not only provide an accurate tool for real-time SMC monitoring in Muringato, but also serve as a prototype for other basins in similar ecosystems and facing similar challenges. Thus, enhancing food security, guide policy formulation, and facilitate long-term resilience against climate change. The integration of artificial intelligence with local community knowledge is significant in promoting sustainable soil water management practices and support drought monitoring.
DAAD
doctoral work
There are no publications of this person.
and Downloads
