Conference Presentation
Reference
Clear sky and all-weather global and beam irradiance models: long term validation
INEICHEN, Pierre
Abstract
The meteorological satellite images as data source to evaluate the ground irradiance components become the state of the art in the field of solar energy systems. The strongest argument is the high spatial coverage, and the fifteen minutes temporal granularity when using images from MSG. They also have the advantage to provide «real time» data used for example to assess the proper operation of a solar plant. On the other hand, long-term ground data are very scarce concerning the beam irradiance. The use of secondary inputs such as polar satellite data and ground information increases significantly the precision of the algorithms, mainly for the beam component. Following a paper from Zelenka concerning the nuggets effect, the interpolation distance to the nearest ground measurement site is limited to 10 to 30 km, depending on the irradiance parameter. This strengths the satellite derived data argument. The use of data derived from models or interpolated between nearby measurements sites are strongly related to the quality of the ground measurements used in the deriving process. This means that the preparatory steps are [...]
INEICHEN, Pierre. Clear sky and all-weather global and beam irradiance models: long term validation. In: 6th PV Performance Modeling and Monitoring Workshop , Freiburg (Germany), October 24-25, 2016
Available at:
http://archive-ouverte.unige.ch/unige:92899
Disclaimer: layout of this document may differ from the published version.
1 / 1
6th PV Performance Modeling and Monitoring Workshop, Freiburg 2016
Clear sky and all-weather global and beam irradiance models: long term validation
Dr Pierre Ineichen
University of Geneva , Energy system group 66 bd Carl-Vogt, CH-1211 Geneva 4
Keywords: satellite derived irradiance; clear sky model; hourly, daily and monthly data; interannual variability; validation
Purpose and approach
The meteorological satellite images as data source to evaluate the ground irradiance components become the state of the art in the field of solar energy systems. The strongest argument is the high spatial coverage, and the fifteen minutes temporal granularity when using images from MSG. They also have the advantage to provide «real time»
data used for example to assess the proper operation of a solar plant. On the other hand, long-term ground data are very scarce concerning the beam irradiance. The use of secondary inputs such as polar satellite data and ground information increases significantly the precision of the algorithms, mainly for the beam component. Following a paper from Zelenka concerning the nuggets effect, the interpolation distance to the nearest ground measurement site is limited to 10 to 30 km, depending on the irradiance parameter. This strengths the satellite derived data argument.
The use of data derived from models or interpolated between nearby measurements sites are strongly related to the quality of the ground measurements used in the deriving process. This means that the preparatory steps are essential to ensure the quality of the data to be used as input to system simulations.
Our approach is to first apply a stringent quality control, including time stamp of the data, absolute and relative calibration coefficient of the sensors, long term stability, components coherence etc., on the both the ground and modeled data.
The usual statistical indicators such as mean bias, root mean square deviation, standard deviation of the bias, correlation coefficient, etc. are used to benchmark the product. We also applied a second order statistic (Kolmogorov-Smirnov) to characterize the frequency distributions.
The comparison is done on an hourly, daily, monthly and yearly basis, on both the global and the beam component.
The interannual variability is also studied.
Ground data
Data from more then twenty ground stations are used for the validation, with up to 16 years of continuous measurements; for the validation itself, due to the satellite data availability, only data since 2004 are used for the validation. The data acquired before 2004 are used to evaluate the interannual variability. The climate range mainly covers desert to oceanic, latitude from 20°N to 60°N, and altitudes from sea level up to 1600 meters.
High precision instruments (WMO 2008) such as Kipp and Zonen CM10 and Eppley PSP pyranometers, and Eppley NIP pyrheliometers, are used to acquire the data.
Models
During the solar irradiance deriving process, the two main steps are represented by the clear sky model that evaluates the highest irradiance possibly available at the considered site (geometric and meteorological conditions), and the all-weather model based on the meteorological satellite images.
The clear sky models are based on the knowledge of the atmospheric aerosol optical depth and water vapor content.
These two parameters can be combined into the Linke turbidity coefficient that was widely used in the last century.
For the all-weather models, the underlying fundamental assumption of retrieving the surface solar irradiance from satellite observations is that the reflected radiance, as measured by the satellite instrument, is related to the broadband atmospheric transmission, and therefore to the solar radiation reaching the ground.
The validated models are the following:
- clear sky models: Bird, Solis, McClear, CPCR2, REST2, ESRA and Kasten, the two last models are based on the Linke turbidity coefficient, the other on the aerosol and water vapor atmospheric content.
- all-weather satellite models: SolarGIS, Helioclim3 v3 and v4, Solemi, IrSOLaV, S2M, Heliomont, EnMetSol, and Climat SAF. These models offer the possibility to derive real-time data. Another set of data is also investigated:
one average year of data derived from models, or based on severeal year of measured or modeled data. These are the following: PVGIS, WRDC data bank, RetScreen, NASA SSE, Meteonorm, and ESRA.
Results
The main results are the following:
Clear sky models:
· three models exhibit roughly the same level performance and stand above the other models. These models are: McClear, REST2 and Solis. The standard deviations for these models are of the order of ±3% for the global component and ±4% to ±5% for the beam component. The standard deviations of the bias are also respectively
±3% and ±4% to ±6%. Considering that the measurements uncertainties are respectively around ±4% and ±3% for the global and the beam components, the validation results show that roughly all the models stay within this value for the global component whatever the aerosol input data set is. When considering the beam component, the standard deviations of the models depend on the aerosol and water vapor input used in the process.
All-weather models:
· global irradiance is retrieved with a negligible bias and an average standard deviation around 17% for the best algorithm. For the beam irradiance, the bias is around several percents, and the standard deviation around 34%. The standard deviations of the bias vary from 2% to 5% for the global irradiance, and from 6% to 14% for the beam component. As expected, the main dependence comes from the clear sky model and the knowledge of the aerosol optical depth. Better results are obtained with daily turbidity instead of climatic monthly values. A lower dependence with the atmospheric water vapor column and the solar elevation angle is pointed out. For the majority of the sites, SolarGis, Heliomont and EnMetSol give the best statistics for all of the components.
Illustration
Quality control: components coherence Quality control: calibration assessement
Model validation: frequency distribution
Model validation: clear sky model effect Clear sky model: behaviour assessement
Model validation: parameter dependence
