Investigaciones y Recursos Solares Avanzados

Nowcasting of Solar Radiation (GHI and DNI) using satellite images

Imprimir PDF


olar power plants needs realible information about the solar resource for different tasks: namely planning, operation, control and maintenance. For each task IrSOLaV offers different temporal time steps and forecasting horizon which cover most of needs of solar power plant operators. The "nowcasting" horizon is a form of very short-range weather forecasting, covering a specific area.

A "nowcast" is loosely defined as a forecast for the coming 6-hour period, based on detailed observational data like radiometers, pyrheliometers, satellite images or sky camaras among other. These predictions are suitable for the operation and control of poer plants. For the electric energy planning, IrSOLaV offers hourly predictions of GHI and DNI with a temporal forecasting horizon of 72 hours using statistical and physical models.

Nowcasting data for the most recent history (up to the present time) can be regularly delivered via ftp, web services or email in real-time for performance assessment and monitoring. Solar radiation predictions from 0 up to 180 hours, is delivered online, every 6 hours.

Satellite images, thanks to its spatial and temporal resolution, are an excellent source of data to make prediction in the very short term or nowcasting. Its utility is mainly due to the considerable effect of clouds over solar irradiance that reaches the ground surface.

The temporal variation of cloud structures is mainly due to the movement of the clounds itself. For this reason motion vector field (MVF) or atmospheric motion vector fields (AMF) are an appropriate tool for the prediction of the movement of clounds.


The application of motion vectors directly over the cloud coefficient allows to obtain the predictions of solar irradiance form satellite images for a temporal horizon of very short-term or nowcasting (next 6 hours). The values predicted are updated each hour. The procedure to derive MVFs from clound coeficcient is based on identifying regions from two consecutives images and calculates the optimal displacement vector between both regions from mean square difference for each pixel. Movement cloud which are obtained are applied to the actual image to derive the position of cloud in the next temporal instant. Finally, a smoothing filter is applied to remove small structures which vary randomly.

Next, some graphics are shown for the nowcasting prediction in Badajoz AEMET (Spanish National Weather Service) location (Spain). Here the day 01/07/2012 with the measurement and predictions of hourly GHI and DNI for the whole day done at 06:00 UTC is presented:

And the predictions updated for the hour 11:00 UTC:


You are here: Home Servicios Solar Radiation Prediction Predicción Nowcasting GHI/DNI

IrSOLaV is contributing to:

IrSOLaV is member in the following Associations:


Last update: 06/01/2016


  • Address: Calle Santiago Grisolia 2 (PTM) Tres Cantos - 28760 - Madrid - Spain
  • Fix Telephone: (+34) 911263612
  • E-mail:
  • Web Address:


  • Address: 63 Daryagani New Delhi 110002 - India
  • Fix Telephone: (+91) 11 4156 2950
  • Mobile: (+91) 99100 61150
  • E-mail:
  • Web Address:
  • Address: Estoril 120 Oficina 809, Las Condes, Santiago - Chile