Evaluating Global and Regional Weather Models for Solar Energy Forecasting in West Africa: A Case Study in Burkina Faso

The potential for photovoltaic energy in West Africa is high, and the use of this resource is expected to grow in the future. Due to the variability of solar energy, accurate weather forecasts are essential to ensure the smooth operation of the electricity network. In this region, the primary weather prediction challenges include the West African monsoon and the advection of dust from the nearby Sahara Desert.

Currently, SteadySun – a company specializing in power and weather forecasts for renewable energies – relies on low to medium-resolution global models to predict GHI (Global Horizontal Irradiance) for West Africa. It has been previously shown that weather models with higher horizontal resolution provide a more realistic representation of small-scale weather phenomena such as convective clouds. Most global models only take into account aerosols concentration through a monthly climatology which does not give information on AOD (Aerosol Optical Depth) variations on small temporal scales. Given the specific characteristics of the West African climate, employing high-resolution models that account for hourly dust concentration is likely to enhance the forecasting system.

This study aims to assess the benefits, limitations, and differences in GHI predictions from five global models and one high-resolution regional model over Burkina Faso. The five global weather models include IFS (ECMWF), GFS (NOAA), and ICON (DWD), which provide simulations with hourly outputs, as well as ARPEGE (Météo-France) and GDPS (CMC), which provide simulations with 3-hourly outputs. The regional weather model used is an augmentation of the weather model WRF for solar energy forecasting: WRF-Solar (NCAR). It features a spatial resolution of 3 km, outputs data every 15 minutes and integrates hourly aerosol optical depth forecasting data from the global atmospheric composition forecast production system CAMS (ECMWF).

The WRF-Solar forecast is expected to deliver improved accuracy during dust advection events and more realistic variability during cloud passages, potentially benefiting the planned forecasting system. However, the absence of data assimilation in WRF-Solar could result in misplaced convection cells, among other inaccuracies. On the other hand, the global models, with their varied physics and resolutions, may offer some advantages under specific weather conditions. This initial evaluation could also identify models that are less suitable for integration into the planned forecasting system.  To perform this assessment, GHI data with a spatial resolution of 3 km and temporal resolution of 15 minutes, derived from MSG (EUMETSAT) satellite imagery, will be used. Two assessment periods have been defined: the first during the monsoon season (July to September 2023) and the second during the dry season (January to March 2024), when dust advections from the Sahara Desert are common.