Logo Lanfrica
  • Home
  • Atlas
  • Insights
  • Docs
  • Sign in

© 2026 Lanfrica. All rights reserved. All copyrights of the resources shown on the Lanfrica website belong to the original copyright holders, unless explicitly stated otherwise.

billletitio/offgrid-pv-soiling-forecast-subsaharan-africa

Domain:

environment and energy

Record type:

software
Creator:
bil
Host:
Soiling forecast toolkit for off-grid PV systems — fetches NASA POWER & MERRA-2 aerosol reanalysis data, engineers climate features, and trains an XGBoost baseline model. Developed and validated in Kenya; designed for broader applicability across sub-Saharan Africa. # offgrid-pv-soiling-forecast-subsaharan-africa > **Developed and validated in Kenya; designed for broader applicability across sub-Saharan Africa.** Soiling forecast toolkit for off-grid PV systems — fetches NASA POWER & MERRA-2 aerosol reanalysis data, engineers climate features, and trains an XGBoost baseline model. Developed and validated in Kenya; designed for broader applicability across sub-Saharan Africa. --- ## Research Objective Soiling — the accumulation of dust, aerosols, and particulate matter on photovoltaic panels — is one of the least-monitored sources of energy yield loss at off-grid community solar sites across sub-Saharan Africa. Physical soiling measurements require sensors and maintenance infrastructure that are rarely available in remote, data-scarce deployments. **This study investigates whether aerosol optical depth (AOD) and meteorological reanalysis data can serve as reliable proxies for PV soiling rate forecasting in environments where no ground-truth soiling measurements exist.** **Research question:** Can an XGBoost model trained on NASA POWER and MERRA-2 reanalysis features predict soiling-induced clearness index degradation with sufficient accuracy to support maintenance scheduling decisions at off-grid community solar sites in semi-arid Kenya? **Hypothesis:** Sites with higher aerosol optical depth (AOD > 0.3), lower precipitation ( =2.31.0 pandas>=2.0.0 numpy>=1.24.0 python-dateutil>=2.8.2 tqdm>=4.65.0 xgboost>=2.0.0 scikit-learn>=1.3.0 # Phase 2 — uncomment after NASA Earthdata registration: # requests-netrc>=0.1.2 ``` --- ## Community Development Context This research is conducted in partnership with **SACDEP (Sustainable Agriculture Community Development Programme)** and affiliated partners. Both field sites — Kabati (Kitui) and Esoit (Kajiado) — are active off-grid community solar installations serving rural communities with no grid access. The soiling forecast work directly supports long-term energy yield planning and …

Visit

github.com

Licenses

MIT