Monitoring and forecast of daytime fog for applications in solar PV systems

Solar photovoltaic (PV) power production is strictly dependent on the global irradiance that reaches the PV panels plane. Solar power production thus depend heavily on the different meteorological conditions. In particular, under fog, the balance between direct and diffuse components of solar irradiance varies depending on the thickness of the fog layer. This modifies the amount of power generated by the solar panels. Information about the presence of fog can improve solar PV power output simulation, thus reducing the associated uncertainty and related costs for the grid operators. Research in satellite meteorology has shown that satellite imagery can be useful in detecting and forecasting fog (Gultepe et al., 2007). Different fog monitoring algorithms have already been developed using the narrowband visible and infrared channels of Meteosat Second Generation (MSG) Spinning Enhanced Visible and Infrared Imager (SEVIRI) (Cermak et al., 2008; Bendix, 2000). In this study fog daytime detection (physical and statistical tests) has been implemented using the High Resolution Visible (HRV) channel of MSG-SEVIRI. This channel has a spatial resolution of 1 km at sub satellite point and this allows to detect localized fog on small valleys. At the same time, the high temporal resolution of HRV data permits to monitor the fog evolution. Combining textural and threshold tests, a distinction between fog and low clouds has been conducted for that pixels identified as low or middle clouds by the C_MACSP cloud detection algorithm (Ricciardelli et al., 2008). Optimal thresholds have been calculated using a specific training dataset built with data collected in different areas of the satellite coverage, seasons of the year, and times of the day. The fog detection algorithm has been validated against METeorological Aerodrome Reports (METAR) observations. Beyond fog detection, another information useful to improve solar PV power production estimation is the probability of fog occurrence. A model to forecast fog occurrence has been implemented using several weather variables from a numerical weather prediction (NWP) model (surface and dew point temperatures, relative humidity, wind speed). The probability of fog occurrence is expressed by an index ranging from 0 to 1. The validation of the forecasted probability of fog occurrence is currently ongoing .