solar energy management technical details

03 Aug solar energy management technical details

Technical details:

• The user through the fully dynamical interface is able to navigate, zoom and click at any pixel of the 1.5 million matrix (the spatial resolution is 3 km at nadir) retrieving solar energy potential information for 3 hours ahead and 3 hours back time horizons at 15-min intervals.
• The continuous monitoring of solar energy is every 15 minutes with a delay of 10 minutes from the actual time. This delay has to do with the Meteosat Second Generation 4 (MSG4) scanning time and the online production of the cloud microphysics data used as input to the fast radiative transfer model.
• The short-term forecasting covers the 3 hours ahead time horizon at 15-min intervals. It is based on deterministic image processing technologies (dense optical flow estimation) and is optimized for the purposes of this service. For more information click here.
• The fast radiative transfer modelling approach is based on large scale pre-calculated look up tables (i.e. 2.5 million simulations using the libRadtran library of radiative transfer models) in synergy with machine learning and high perfomance computing.
• The production time of the 3 hours back and 3 hours ahead time horizons for the region of Europe and North Africa is almost 1 minute and 40 seconds. The new data projection time windows are: xx:11:40, xx:26:40, xx:41:40 and xx:56:40, where xx corresponds to the hour of the day followed by the minutes and seconds. The projected hours of the day cover the time window 04:00 – 19:00 UTC.
• The assymetrical shape of the colored map is a result of the Earth disk view by the MSG4 projected on a Cartesian coordinate system. The shape of each selected pixel follows the same geo-mapping specific features.
• The colored map projection represent the current time and is based on tiles production for all zoom levels. The tiles production is refreshed dynamically during navigation, zoom and click by the user, while the time of new tiles are: xx:14, xx:29, xx:44 and xx:59, where xx corresponds to the hour of the day followed by the minutes.
• For the aerosol effect on solar energy the aerosol optical depth 1 day forecasts from the Copernicus Atmosphere Monitoring Service was used at 1 hour temporal resolution. This information is refreshed every night at 20:59 UTC using the 12 hour CAMS run available after 20:00 UTC every day.
• For the almost 20 million radiative transfer model simulations required every 15 minutes (1 matrix of 1.5 million pixels for the current time and 12 matrixes of 1.5 million pixels for the 3 hours ahead time horizon, i.e. 12 time steps), a distributed and high performance computing was exploited at a UNIX featured environment. For this purpose a 32-core independent core architecture is used followed by a 256 Gb RAM, distributed into 16 virtual machines working in parallel and in real time.
• The daily dataflow including the input and output data retrieving, production and projection is of the order to 550 Gb handled by a 12Tb RAID10 storage architecture.

Go back to the solar energy management service.