Evaluation and applications of a new
satellite-based surface solar radiation data
set for climate analysis
Jörg Trentmann1, Richard Müller1, Christine TrägerChatterjee1, Rebekka Posselt2, Reto Stöckli2
Satellite Application Facility on Climate Monitoring (CM SAF)
1Deutscher
April 2011
Wetterdienst, 2MeteoSwiss
1
Motivation
• Surface Solar Irradiance highly relevant:
 Climate Monitoring and Climate Analysis
 Solar Energy
• Available data sets (ISCCP, GEWEX, ERA-Interim)
agree well on the mean
• Differ substantially in the temporal
evolution
• Coarse spatial resolution
April 2011
2
The CM SAF approach
• Retrieve surface solar irradiance (SIS) from the
geostationary Meteosat satellites (1983 – today)
• Apply a well-established method:
Heliosat
(Cano et al.,1986, Hammer et al., 2003)
• Provide the data at high temporal and spatial resolution
• Free and easy accessible
• Validate the data with BSRN surface station data
• Evaluate the data with alternative data sets
April 2011
3
The Heliosat method
Fundamental Assumptions
1. The clear-sky surface radiation can be accurately
calculated
(information on the water vapor and aerosol is required)
2. For each satellite pixel and time slot the minimum
reflectance of each months represents clear-sky
conditions
(i.e., effect of Rayleigh scattering + surface albedo on the
reflectance)
3. The cloud optical depth is related to the cloud-reflected
solar radiation
(= brightness of the visible satellite channel)
4. The degradation of sensor sensitivity can be monitored
with bright cloud targets (maximum reflectance)
April 2011
4
The Heliosat method
• Surface irradiance (global radiation) is
retrieved for each satellite pixel / time
slot (30 min) between 1983 and 2005.
• Average and interpolate to hourly / daily
/ monthly means on a 0.03o-lon-lat-grid.
• Data is freely available in CF-netcdf-format:
www.cmsaf.eu
• Additionally, the direct surface irradiance and the cloud
index are provided.
• Scripts to visualize and analyse the data using opensource software (cdo, R) are provided.
April 2011
5
Validation
• Monthly means surface
irradiance at 14 BSRN stations
April 2011
6
Validation
April 2011
7
Validation
Temporal Stability
a) Hovmöller Diagram
b) Temporal evolution of the bias to
BSRN surface stations
c) Apply standard techniques to
detect change points (ongoing
project, University Frankfurt)
April 2011
8
Evaluation
• CM SAF data about 2-3 Wm-2 higher than
alternative data sets
• Temporal evolution of CM SAF consistent with
alternative data sets, exception 1991
(Pinatubo??)
April 2011
9
Applications
Close-up Views
Germany
Amazon
Mediterranean
West Africa
Tanzania
April 2011
10
Applications
Surface Irradiance, Mean July, CM SAF
Topography
GEWEX
• No validation with DWD
surface network available.
• Excellent correspondence
with topographic features.
• GEWEX SRB data misses
details due to coarse
resolution.
April 2011
11
Applications
Close-up Views
Germany
Amazon
Mediterranean
West Africa
Tanzania
April 2011
12
Applications
Mean Surface Radiation, Tanzania
• No surface network
available
• Correspondence of surface
irradiance with
topographic features.
Cooperation with Jörg Bendix, University Marburg
April 2011
13
Applications
Surface Radiation at Mt. Kilimanjaro
• Substantially reduced
surface radiation around
Kilimanjaro due to enhanced
cloud cover
April 2011
14
Applications
Time Series of
Surface Radiation
37.5° E, 3.1° S
April 2011
37.6° E, 3.1° S
15
Applications
Linear Trend in Surface Radiation
(W/m2/dec)
• There are significant trends in
the surface radiation at Mt.
Kilimanjaro between 1983 and
2005
• Strong increase of surface
radiation (more than 15
W/m2/dec) over the summit (>
3000 m)
• Decrease along the northern
and southern slopes
April 2011
16
Summary
• Heliosat Method is applied to Meteosat Satellites to derive
solar surface irradiance from 1983 to 2005
• Validation with BSRN surface measurements shows high
quality of the CM SAF satellite-derived data set.
• Temporal (hourly) and spatial (0.03o) resolution of the data
set is unique.
• Spatial resolution allows detailed local studies:
Strong contrast in temporal trends of surface irradiance at
the summit (increase) and the foothills (decrease) of Mt.
Kilimanjaro, Tanzania.
• The Data Set is freely available in CF-netcdf-format at
www.cmsaf.eu, software tools are also provided
• Processing of the global AVHRR satellite data (1982 – 2009)
ongoing (see Poster XY 220, F. Kaspar et al.)
April 2011
17
April 2011
18
Extra Slides
April 2011
19
The Heliosat method
Reflectivity, 12 UTC, 2 Sept 2008
Min. Reflectivity, Rmin, 12 UTC, Sept 2008
April 2011
20
The Heliosat method
Max. reflectance, Rmax:
95 % percentile of counts
during one month in the
reference region
Reflectivity, 12 UTC, 2 Sept 2008
Temporal evolution of Rmax
April 2011
21
The Heliosat method
The definition of the Cloud Index n:
Cloud Index, 11 UTC, 1 July 2005
April 2011
The Heliosat method
• The cloud index, n, is related to the clear-sky index, k.
• The clear-sky index, k, is the ratio between the all-sky surface
irradiance, G, and the clear-sky surface irradiance, Gclear
clear sky
index
cloud index
April 2011
23
The Heliosat method
• The cloud index, n, is related to the clear-sky index, k:
k=1n
• The clear-sky index, k, is the ratio between the all-sky surface
irradiance, G, and the clear-sky surface irradiance, Gclear:
G = k * Gclear
• Gclear can be calculated by radiation transfer calculations using
the fast and accurate clear-sky model gnu-MAGIC (Mesoscale
Atmospheric Global Irradiance Code, Mueller et al., 2009,
http://sourceforge.net/projects/gnu-magic/)
April 2011
24
Validation
• Monthly means SIS from 14
BSRN stations
• Measures: Bias, Variance,
Correlation Coefficient of
Anomalies, Fraction of months with
bias above 15 Wm-2
April 2011
25