Programm Photovoltaik Ausgabe 2008 ... - Bundesamt für Energie BFE
Programm Photovoltaik Ausgabe 2008 ... - Bundesamt für Energie BFE
Programm Photovoltaik Ausgabe 2008 ... - Bundesamt für Energie BFE
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Introduction<br />
In the framework of IEA Solar Heating and Cooling (SHC) Task 36 “Solar Resource Knowledge Management”<br />
Meteotest investigates mainly the possibilities and quality of global radiation forecast.<br />
The Task 36 is divided into 3 main subtasks:<br />
a) Standard qualification for solar resource products (includes benchmarking of different radiation<br />
estimation models based on satellite measurements).<br />
b) Common structure for archiving and accessing solar resource products (includes prototype of<br />
online tool for accessing data).<br />
c) Improved techniques for solar resource characterization and forecast; improve satellite retrieval<br />
methods for solar radiation products; conduct climatological analysis of solar resources.<br />
The aim in the radiation forecast subtask is to define the quality of the existing models and to enhance<br />
the quality. A further aim is also to check the quality of forecasted PV production data. Main partners<br />
in this subtask are University Oldenburg and New York State University at Albany.<br />
Additionally Meteotest did some work within part c), where a new turbidity climatology was made.<br />
Work done and first results<br />
Radiation forecast<br />
Meteotest made a short analysis of the operational global radiation forecast of the MM5 model, which<br />
runs with input data of GFS (global model of US National Weather Service) and in spatial resolution of<br />
30 km.<br />
The global radiation was used in form of direct model output. We tested the hourly values at 4 different<br />
stations in Switzerland (Basel, Geneva, Lugano and Schaffhausen). The validation was made with<br />
measured values from the network of MeteoSwiss.<br />
The validation forecast ranged from 9 – 63 hours (9, 12 15 h for day 1 - 3)<br />
The test period was June 16th 2006 – October 16th 2006 (1.33 years)<br />
The validation showed the following results:<br />
1. Mean biases are small<br />
2. relative root mean squared errors are (prmse):<br />
All seasons, day 1, all hours: 53%; summer, day 1, 12 UTC: 38%<br />
3. The quality (prmse) is: dependent on time of day (factor 1.3); strongly dependent on season<br />
(factor 2)<br />
The ongoing validations of the subtask members showed partly different results. The groups working<br />
with MM5 models got similar results in Spain.<br />
New York State University at Albany (Richard Perez) used digital weather forecast of cloud cover of<br />
US National Weather Service. Those results showed a Prediction Root Mean Squared Error of<br />
20-40% depending on the region (Desert Rock, Boulder, Goodwin Creek).<br />
University of Oldenburg used ECMWF (European weather model). Those results showed a Prediction<br />
Root Mean Squared Error of 30-37% depending on the region (Germany and Spain). This could be<br />
partly due to a better radiation model within the ECMWF model.<br />
As quality is very much depending on site and time of the year, the team members will conduct next<br />
year validations for the same regions and periods for all models.<br />
Seite 272 von 288<br />
Solar Resource Management, J. Remund, Meteotest<br />
2/3