Solar power forecasting
Solar Photo Voltaic Systems convert the solar energy into electric power. The power output obviously depends on the incoming radiation and on the characteristics of the panel itself. Knowledge of the trajectory of the sun, scattering processes, condition of the atmosphere and characteristics of the installation make a reliable forecast possible.
Solar radiation travels through space and reaches the earth's atmosphere. As the radiation travels through the atmosphere part of it will be scattered by atoms and little particles, like dust and water droplets. The remaing part reaches the solar panel and is called direct radiation. The amount of direct radiation that reaches the panel depends on the pathlength through the atmosphere (airmass), the kind and number of atmospheric particles that may vary with the time of the year and the angle between incoming radiation and the surface of the panel.
Part of all scattered radiation will reach the panel indirectly. This part is refered to as diffuse radiation. Radiation that is reflected by clouds is an example of diffuse radiation. The sum of direct and diffuse radiation can be larger under cloudy conditions than in case of clear sky conditions.
The last contribution to the incoming radiation is the reflected radiation, depending on the colour or albedo of the surroundings and the total configuration of mountains and urbanisation. Reflectivity usually is zero for horizontal panels and increases with the tilt angle.
Irradiance as the sum of direct, diffuse and reflected radiation can be forecasted for each location on earth and each orientation of the panel, by using the angle between incoming radiation and the panel and information on atmospheric conditions from numerical weather models. The graph shows the incoming radiation for a fixed panel oriented to the southwest and a tilt angle of 45 degrees (blue) and a panel mounted on a solar tracking system (red). The tracking system can increase the efficiency by 21% at a latitude of 53 degrees North. The forecast is valid for Amsterdam.
tuning of the forecasting model
Using the radiation forecast and the characteristics of the solar panel, like temperature dependency and refraction of the cover, it is possible to forecast the power output. For large facilities we recommend to tune the model to local circumstances, specific characteristics of the panel and systematic errors of the weather model.
short term forecasting
Forecasting the cloudcover is one of the major concerns in numerical weather modelling. Often the cloudcover has not been predicted very accurate. To improve the short term forecast until several hours in advance, it is possible to extrapolate satellite images and derive a new cloudcover forecast from the extrapolation. Aeolis has used this method succesfully in forecasting precipitation (see http://www.meteox.nl/h.aspx?r=&jaar=-3&soort=exp).