The document discusses how to obtain bankable meteorological data for solar power plants. It describes using ground measurements for high accuracy but also satellite data which provides spatial resolution over long periods of time. Combining ground and satellite data through validation can provide accurate hourly time series, irradiation maps, and long-term annual means.
1. How to get bankable meteo data?
DLR solar resource assessment
Robert Pitz-Paal,
Norbert Geuder, Carsten Hoyer-Klick, Christoph Schillings
2. Solar resource assessment for solar power plants
Why solar resource assessment ?
Characteristics of solar irradiation data
DLR solar resource assessment:
ground measurements
satellite data
How to get bankable meteo data?
Summary
Slide 2 > Solar resource assessment at DLR
3. Why is exact knowledge of the solar resource
so important?
High impact of the annual 50 MW Parabolic Trough solar thermal power plant
irradiation
Levelized Electricity Costs (LEC)
Levelized Electricity Costs in €/kWh
on the LEC
Annual electricity generation in GW
Mean system efficiency and
Mean system efficiency
Irradiation is a
crucial parameter for
site selection and
plant design and
Annual electricity generation
economics of plant
Solar field size (aperture) in 1000 m²
Slide 3 > Solar resource assessment at DLR
Why solar resource assessment?
4. Characteristics of solar irradiation data
What kind of irradiation data is needed?
Type:
DNI (Direct-Normal Irradiation)
GHI (Global-Horizontal Irradiation)
DHI (Diffus-Horizontal Irradiation)
Source:
ground measurements
satellite data
Properties of irradiation:
spatial variability
inter annual variability
long-term drifts
Slide 4 > Solar resource assessment at DLR
Characteristics of solar irradiation data
5. Ground measurements vs. satellite derived data
Ground measurements Satellite data
Advantages Advantages
+ high accuracy (depending on sensors) + spatial resolution
+ high time resolution + long-term data (more than 20 years)
+ effectively no failures
+ no soiling
+ no ground site necessary
+ low costs
Disadvantages
- high costs for installation and O&M Disadvantages
- soiling of the sensors - lower time resolution
- sometimes sensor failure - low accuracy in particular at high time
- no possibility to gain data of the past resolution
Slide 5 > Solar resource assessment at DLR
Characteristics of solar irradiation data
6. Inter annual variability
Strong inter annual and regional 1999
variations deviation
to mean
2000
kWh/m²a
2001
Average of the direct normal 2002
irradiance from 1999-2003
2003
Slide 6 > Solar resource assessment at DLR
Characteristics of solar irradiation data
7. Long-term variability of solar irradiance
7 to 10 years of measurement to get long-term mean within 5%
If you have only one year of
measured data, banks have to
assume, that this it is here
Slide 7 > Solar resource assessment at DLR
Characteristics of solar irradiation data
8. Time series of annual direct normal irradiance
Long-term tendency
may exist and
differs from site to site
linear regression:
+2.16 W/(m² year)
Spain
linear regression: Australia
-0.93 W/(m² year)
with stratospheric aerosol
Slide 8 > Solar resource assessment at DLR
Characteristics of solar irradiation data
9. How to get bankable Meteo Data?
Quality checked ground Derivation of irradiation from satellite
measurements to gain data to get
highly accurate data ƒ spatial distribution and
ƒ long-term time series
Validation of the satellite data with accurate
ground data
Result: accurate hourly time series, irradiation maps and long-term annual mean
Slide 9 > Solar resource assessment at DLR
DLR solar resource assessment
10. Instrumentation for unattended abroad sites:
Rotating Shadowband Pyranometer (RSP)
sh
ad
pyra
nom
ow Sensor: Si photodiode
eter ba
sen nd
sor
Advantages:
+ fairly acquisition costs
+ small maintenance costs
+ low susceptibility for soiling
+ low power supply
Disadvantage:
diffus
horizontal - special correction for good accuracy
Global HorizontaI Irradiation (GHI) necessary (established by DLR)
measurement
Slide 10 > Solar resource assessment at DLR
Ground measurements
11. Precise sensors (also for calibration of RSP):
Thermal sensors:
pyranometer and pyrheliometer,
GHI DNI precise 2-axis tracking
DHI
Advantage:
+ high accuracy
+ separate GHI, DNI and DHI sensors
(cross-check through redundant measurements)
Disadvantages:
- high acquisition and O&M costs
- high susceptibility for soiling
- high power supply
Slide 11 > Solar resource assessment at DLR
Ground measurements
12. Satellite data: SOLEMI – Solar Energy Mining
Meteosat Prime Meteosat East
SOLEMI is a service for high resolution and high quality data
Coverage: Meteosat Prime up to 22 years, Meteosat East 10 years (in 2008)
Slide 12 > Solar resource assessment at DLR
Satellite data
13. Radiative Transfer in the Atmosphere
1400
Extraterrestrial
1200
O2 and CO2
Direct Normal Irradiation
1000
Ozone
800
Rayleigh
(W/m²)
600
Water Vapor
400
Aerosol
200
Clouds
0
00:00
02:00
04:00
06:00
08:00
10:00
12:00
14:00
16:00
18:00
20:00
22:00
00:00
Hour of Day
Slide 13 > Solar resource assessment at DLR
Satellite data
14. How two derive irradiance data from satellites
The Meteosat
satellite is located in
a geostationary orbit
The satellite scans
the earth line by line
every half hour
Slide 14 > Solar resource assessment at DLR
Satellite data
15. How two derive irradiance data from satellites
The Meteosat
satellite is located in
a geostationary orbit
The satellite scans
the earth line by line
every half hour
The earth is scanned
in the visible …
Slide 15 > Solar resource assessment at DLR
Satellite data
16. How two derive irradiance data from satellites
The Meteosat
satellite is located in
a geostationary orbit
The satellite scans
the earth line by line
every half hour
The earth is scanned
in the visible and
infra red spectrum
Slide 16 > Solar resource assessment at DLR
Satellite data
17. How two derive irradiance data from satellites
The Meteosat
satellite is located in
a geostationary orbit
The satellite scans
the earth line by line
every half hour
The earth is scanned
in the visible and
infra red spectrum
A cloud index is
composed from the
two channels
Slide 17 > Solar resource assessment at DLR
Satellite data
18. How two derive irradiance data from satellites
Atmospheric
transmission is
calculated from
global data sets
elevation
Slide 18 > Solar resource assessment at DLR
Satellite data
19. How two derive irradiance data from satellites
Atmospheric
transmission is
calculated from
global data sets
elevation
ozone
Slide 19 > Solar resource assessment at DLR
Satellite data
20. How two derive irradiance data from satellites
Atmospheric
transmission is
calculated from
global data sets
elevation
ozone
water vapor
Slide 20 > Solar resource assessment at DLR
Satellite data
21. How two derive irradiance data from satellites
Atmospheric
transmission is
calculated from
global data sets
elevation
ozone
water vapor
aerosols
Slide 21 > Solar resource assessment at DLR
Satellite data
22. How two derive irradiance data from satellites
Atmospheric
transmission is
calculated from
global data sets
elevation
ozone
water vapor
aerosols
The cloud index is
added for cloud
transmission
Slide 22 > Solar resource assessment at DLR
Satellite data
23. How two derive irradiance data from satellites
Atmospheric
transmission is
calculated from
global data sets
elevation
ozone
water vapor
aerosols
The cloud index is
added for cloud
transmission
All components are
cut out to the region
of interest
Slide 23 > Solar resource assessment at DLR
Satellite data
24. How two derive irradiance data from satellites
Finally the irradiance
is calculated for
every hour
Slide 24 > Solar resource assessment at DLR
Satellite data
25. Example for hourly time series
for Plataforma Solar de Almería (Spain)
1200
ground
1000 satellite
800
W/m²
600
400
200
0
13 14 15 16 17 18
day in march, 2001
Slide 25 > Solar resource assessment at DLR
Validation of the data
26. Comparing ground and satellite data: time scales
12:45 13:00 13:15 13:30 13:45 14:00 14:15
Hourly average Meteosat image Measurement
Ground measurements are typically
pin point measurements which are
temporally integrated
Satellite measurements are
Hi-res satellite pixel in Europe
instantaneous spatial averages
Hourly values are calculated from
temporal and spatial averaging
(cloud movement)
Slide 26 > Solar resource assessment at DLR
27. Comparing ground and satellite data: “sensor size”
solar thermal
power plant satellite pixel
(200MW ≈ 2x2 (∼ 3x4 km²)
km²
ground
measurement
instrument
(∼2x2 cm²)
Slide 27 > Solar resource assessment at DLR
28. Comparison with ground measurements and accuracy
general difficulties: point versus area and
time integrated versus area integrated
1200
ground
1000 sate llite
800
W/m²
600
400
200
0
DNI time serie for 1.11.2001, Almería 0 6 12 18 24
Slide 28 > Solar resource assessment at DLR
Partially cloudy conditions, cumulus humilis hour of day
29. Comparison with ground measurements and accuracy
general difficulties: point versus area and
time integrated versus area integrated
1200
ground
1000 sate llite
800
W/m²
600
400
200
0
DNI time serie for 30.4.2000, Almería 0 6 12 18 24
Slide 29 > Solar resource assessment at DLR
overcast conditions, strato cumulus hour of day
30. Validation in Tabouk, Saudi Arabia
all-sky clear-sky
All Sky Clear sky only
Validation of the data
31. Validation results
RMSE (Root Mean Square Error)
Decreasing deviation between
ground and satellite derived
data with increasing duration
of the integration time
Bias Site bias
Measurement Network Saudi Arabia +4.3%
Variing within different sites Several Sites in Spain -6.4% to +0.7%
Morocco -2.0%
Algeria -4.8%
Slide 31 > Solar resource assessment at DLR
Validation of the data
32. Monthly ground and satellite derived irradiation data
hourly monthly mean (DNI_ground) in Wh/m², Solar Village 2000
hourly monthly mean (DNI_satellite) in Wh/m², Solar Village 2000
Slide 32 > Solar resource assessment at DLR
Validation of the data
33. Adjusting Ground and Satellite Data
Simple Method: Scaling with the Bias
E.g. with a Bias of -5%, every value is multiplied with 1.05
+ Very easy to apply
- Modification of frequency distribution at the extreme end
a factor > 1 may produce unrealistic high values
a factor < 1 may omit high values
Suitable for average values
Slide 33 > Solar resource assessment at DLR
34. Adjusting Ground and Satellite Data
Advanced Method: Error analysis and correction functions
Analysis of the Deviations:
At clear sky? (e.g. due to incorrect atmospheric data)
During cloud situations? (e.g. incorrect cloud modelling)
Development of a correction function dependent e.g. on the cloud
index. 7
c(x)
6
5
Korrekturfaktor
4
3
2
1
0
-0.2 0 0.2 0.4 0.6 0.8 1 1.2
Wolkenindex
Bias before Correction Bias after
Slide 34 > Solar resource assessment at DLR
35. Summary
Ground measurements are accurate but expensive and in suitable regions
mostly rare (especially DNI)
New ground measurements do not deliver time series for the past
Satellite data offers spatial resolution and long-term time series of
more than 20 years into the past
Combination of ground and satellite date yields: irradiation maps,
long term annual means and time series – all with good accuracy
Realistic long term meteo data helps to avoid risk surcharges of banks due to
conservative assumptions and increase the financial viability of your project
Slide 35 > Solar resource assessment at DLR
Summary
36. Thank you for your attention !
Acknowledgements
Richard Meyer, Institute for Physics of the Atmosphere,
DLR Oberpfaffenhofen
Sina Lohmann, Institute for Physics of the atmosphere,
DLR Oberpfaffenhofen
Antonie Zelenka, MeteoSwiss, Zürich
Volker Quaschning, FHTW Berlin
Slide 36 > Solar resource assessment at DLR
Acknowledgements
38. Satellite data and nearest neighbour stations
Satellite derived
data fit better to a
selected site than
ground
measurements
from a site
farther than
25 km away.
Slide 38 > Solar resource assessment at DLR
Validation of the data
39. Derivation of the Cloud-Index from satellite images
VIS-channel IR-channel
(0.5 – 0.9µm) (10.5 –12.5 µm)
Original-image
Meteosat-7
Reference-image
Derived cloud
information
(Cloud-Index)
40. Time series of direct normal irradiance
aerosol opt. thickness
linear regression:
Spain 2.16 W/(m² year)
1.56 W/(m² year)
aerosol optical thickness
Pinatubo
El Chichon
with
without
stratospheric aerosol
with
without stratospheric aerosol
linear regression:
Pinatubo Australia - 0.93 W/(m² year)
El Chichon
- 1.22 W/(m² year)
Slide 40 > Solar resource assessment at DLR