We model & validate

Your go-to application for high-resolution wind & solar data to assess the long-term yields for your future renewable assets, improving the bankability of your pipeline, and evaluate the performance of your operational plants.

Modeling methodology

  • Satellite imagery
  • Parallax correction
  • TMY Pxx
Satellite imagery

The cloud, radiation and precipitation properties are retrieved from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) on board of four different geostationary weather satellites:

  • Meteosat-11 (PRIME), covering Europe and Africa with a 3x3 km² resolution. Data is available as from 2004.
  • Meteosat-8 (IODC), covering the Europe, Africa, South-West Asia and the Indian Ocean with a 3x3 km² resolution. Data is available as from 2019.
  • GOES-East' Northern Hemisphere Extension, covering North- and Central-America with a 2x2 km² resolution. Data is available as from 2019
  • Himawari-8 Real-Time, covering Oceania and South-East Asia with a 2x2 km² resolution. Data is available as from 2021.

Identifying clouds, radiation & precipitation using CPP

The Cloud Physical Properties (CPP) algorithm is developed at KNMI to derive meteorological products from the Meteosat Second Generation (MSG) satellite. Based on the MSG-CPP, we identify the cloud, radiation & precipitation properties in 4 distinct steps:

First, cloud-free pixels are separated from cloud-contaminated & cloud-filled pixels with the GEO v2018 algorithms developed by the NWC SAF.

Second, the cloud optical and microphysical properties (i.e. top temperature, phase, optical thickness & particle size) are derived using algorithms developed in the CM SAF (Benas et al. 2017; Roebeling et al. 2006). The retrieval of most these properties relies on observations of solar backscattered radiation, and is thus limited to daytime (solar zenith angle smaller than 84°).

Third, the total surface solar irradiance and its direct & diffuse components are derived using the methods described in Greuell et al. 2013.

At last, the precipitation intensity is estimated based on the retrieved cloud properties during daytime (Roebeling & Holleman 2009) and based on statistical relationships with the observed infrared brightness temperatures (Brasjen et al. 2015).

Parallax correction

A parallax correction corrects the derived irradiance properties for the relative position of both the satellite and the sun.


P50 or Pxx typical meteorological years (TMY) are created based on Cebecauer & Suri 2015, but different weighing actors are applied to tailor the creation of TMY's to the application of solar power simulations.

Validation & accuracy

  • TUV Rheinland 2018
External validation by TÜV Rheinland (2018)

TÜV Rheinland was asked to perform a detailed 3rd-party validation of 3E’s satellite derived solar irradiation data. The aim was to quantify the error of the satellite date in comparison with high quality public meteo stations and to evaluate the results.

TÜV Rheinland has validated 3E’s satellite derived solar irradiation data over 35 meteo stations in Germany. The results of the validation include bias, standard deviation (SD) of bias and root mean square error (RMSE) between all available measurement stations and satellite data. After processing, filtering and quality control of the data sets 215 complete years (i.e. 1-14 years per stations between 2005 and 2017) remained for the validation. The time and spatial aggregated results are:

  • a mean bias of 0.7% with a standard deviation of 2.5%
  • a monthly and daily RMSE of respectively 4.6% and 10.6%

TÜV Rheinland concluded that "this high accuracy from the results over all years confirms the excellent quality of 3E’s solar irradiation data (GHI) in the validated moderate-climate region."

Download report of TÜV Rheinland