1

CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Integrating CM SAF data with PVGIS for

Estimation of Solar Energy System
Performance

Thomas Huld, Richard Müller, Attilio Gambardella

 2
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Solar data needs for solar energy applications

Solar radiation data are needed for solar energy system
planning, deployment and operation, but the type of data
needed varies
• Spatial resolution: generally as high as possible
• Historical data needed for planning/deployment, recent data
(or forecasts) needed for operation/monitoring
• Long-term averages are sufficient for some applications,
while high time resolution is useful for others
 3
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Solar Energy Systems

Two main groups:

• Thermal systems
– Domestic hot water and space heating
– Concentrating solar thermal power plants

• Photovoltaic (PV) systems

– Fixed mounted flat-plate systems, stand-alone
– Fixed mounted flat-plate systems, grid-connected
– Sun-tracking flat-plate systems
– Sun-tracking concentrating systems
 4
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Data needs for solar energy estimation

• Flat-plate grid-connected PV: global and diffuse/beam long-

term monthly averages. Tracking systems are more sensitive
to the beam component
• Off-grid PV and solar hot water systems: long time series of
hourly/daily global and diffuse/beam values
• Concentrating PV: direct normal irradiance (DNI), at least
long-term monthly averages
• Solar thermal power plants: long time series of DNI
 5
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

What is PVGIS?

• PVGIS is a database of solar radiation and temperature data

combined with a web interface that lets users calculate the
energy output of photovoltaic (PV) systems.
• PVGIS is also a scientific tool that allows us to do research
on the performance of PV systems over large geographical
areas and estimate the potential for solar energy deployment
in Europe and Africa
 6
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

What is PVGIS? (2)

• The PVGIS solar radiation databases for Europe and Africa

were obtained in the following way:
– Europe: Long-term monthly averages global and diffuse irradiation
from ground stations are interpolated onto a spatial grid. Data
collected by ESRA over the period 1981-1990
– Africa: Daily values of global irradiation sums from the HelioClim-1
database, estimated from MFG satellite data, with a spatial resolution
of 15’. Time period 1985-2004
• The online calculation also includes the effects of shadows
from nearby mountains. The shadow calculation is based on
the SRTM-3 DEM with a resolution of 3 arc-seconds (~90m
at the equator).
 7
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Web interface features

• Covers Europe and Africa

• Various types of PV installations: fixed or tracking, models
for different PV technologies
• High-resolution terrain data allows calculation of the effects
of shadowing
• Google Maps interface with search and zoom facilities.
 8
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Click here

Then click here

1000-2000 hits
per day
 9
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Reasons for success

•Simple interface: 2 clicks and you have results

•Easily understood results: energy production in kWh
•Immediate access:
–Relevant information is not buried 8 menu levels deep
–No registration needed
–It’s free!
 10
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Research applications

• Mathematical models of PV performance, together with our

colleagues of the ESTI laboratory
• Performance of PV technologies in different climates
• Performance of different PV mounting strategies
• Potential for PV in rural electrification in Africa
 11
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

PVGIS: Known problems

• Europe: Heterogeneous spatial distribution of ground station

data. Some ground station data not reliable. Data are old.
• Africa: Lack of diffuse component. Low primary spatial
resolution, and lack of validation.
• General: beam irradiance is not good enough to calculate
DNI with any confidence. Time resolution too low for some
applications.
 12
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

A wish list of Improvements

• Quality improvements in the solar radiation data

• Higher temporal resolution, for research on intermittency and
for off-grid PV installations
• Direct normal irradiance for concentrating PV applications
• My pipe dream: PVGIS Asia

Collaboration with CM SAF via DWD on high-resolution

satellite-based data. Incorporation into PVGIS in 2010.
 13
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Radiation data from CM SAF

MSG:
• 4 years of hourly global and beam data, 06-2006 to 05-2010
• Geographical coverage: Most of Europe and North Africa:
20N to 58N, 15W to 35E
• Spatial resolution 1’30’’ (about 2.5km).
MFG:
• 8 years of hourly global and beam data, 1998-2005
• Covers most of Europe and North Africa: 0N to 58N, 15W to
35E
• Spatial resolution 1’48’’
 14
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Construction of the new PVGIS dataset

• Calculation of long-term monthly averages:
– Calculate multiyear average of each time slot for beam and global
– Then sum all average time slots in each month
– Overall average is arithmetic average of MFG and MSG data
• Calculate monthly beam and diffuse clear-sky indices as
needed for the PVGIS algorithm
• Downscaling so far only includes terrain shadows in the
online calculation
 15
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Comparison of PVGIS-3 with CM-SAF and other

datasets
• Europe:
– PVGIS vs. CM-SAF data from MFG and MSG
– PVGIS vs. an average of four datasets:
ƒ Helioclim-3 (MSG-based)
ƒ SatelLight (MFG-based)
ƒ Meteonorm (mainly based on ground stations)
ƒ NASA SSE (various sources)
• Africa:
– PVGIS (HelioClim-1) vs. CM-SAF data from MFG
 16
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Differences from existing PVGIS

PVGIS-3 vs. CM-SAF PVGIS-3 vs. 4 datasets

Global horizontal, relative difference in annual average

 17
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Global irradiation, comparison CM-SAF and PVGIS-3

Location MSG MFG bias CM-SAF –
bias(%) (%) PVGIS (%)
Lindenberg (DE) -3.4 -3.0 +6.9
Cabauw (NL) +0.4 +1.5 +11.6
Carpentras (FR) +2.1 +5.1 +9.0
Payerne (CH) -3.0 +3.7 +13.2
Camborne (UK) - +6.2 +8.4
Ispra (IT) +8.0 - +15.0
Milano (IT) -0.5 - +13.0
 18
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Differences from PVGIS-3, Africa

PVGIS-3 CM-SAF MFG

Global horizontal, long-term annual average
 19
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Differences from PVGIS-3, Africa

Global horizontal, relative difference in annual average

 20
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

First steps in validation of DNI

MSG MFG
BSRN station Months MBE (%) Months MBE (%)
Sde Boqer (IL) 26 +5.5 21 +1.5
Cabauw (NL) 36 +4.8 11 +14.9
Carpentras (FR) 24 +2.5 22 +8.6
Payerne (CH) 22 -4.4 24 +9.6
Tamanrasset (DZ) 36 -9.1 24 -5.9
Camborne (UK) 23 +18.2

Relative MBE between CM SAF and BSRN stations

 21
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Future developments

• New version of PVGIS with choice of datasets (1-2 weeks)

• Extension of spatial coverage of CM SAF data in PVGIS (if
Richard can fit it into his schedule)
• Incorporation of high time-resolution data from CM SAF to
calculate off-grid systems in Europe (~end of year)
• Making DNI available in the PVGIS web application (2011)
• PVGIS Asia (2011?)
 22
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

Conclusions

• Successful collaboration between CM SAF and JRC

• Clear improvements in some regions with high solar energy
penetration, such as Northern Italy
• DNI data look very encouraging
• Higher-resolution data allow us to do useful new stuff
• PVGIS saved from slow decline into irrelevancy
 23
CM-SAF 3rd User Workshop, Rostock, 06-08/09/2010

THANKS to:

The CM-SAF teams for producing a high-quality

solar radiation data set, and making it available!

YOU, for your patience