GSICS Bias Monitoring Routine comparisons of satellite radiances against reference
Post on 07-Feb-2016
32 Views
Preview:
DESCRIPTION
Transcript
GSICS Bias Monitoring• Routine comparisons of satellite radiances against reference
GSICS Correction• Function to correct issued radiances• For consistent calibration with reference
GSICS Reports & Guidelines• Recommendations to modify practices• Design and Operation of future satellite instruments
For Operational Environmental Satellites Infra-red recalibration (GEO and LEO) Demonstration status
(current operational satellites) Near real-time and re-analysis Reflected Solar Band recalibration (GEO/LEO) In development within GSICS Microwave – Sounders & Imagers (LEO) In development with GPM XCAL Historic Instruments In development at EUMETSAT…
Migrating from Metop-A/IASI to Metop-B/IASI as GSICS
Inter-Calibration Reference for Geostationary IR ImagersTim Hewison1
1: EUMETSAT, Eumetsat-Allee 1, D-64295 Darmstadt, GermanyPlease send questions and comments to tim.hewison@eumetsat.intEUM/RSP/VWG/13/697414 NOAA Satellite Conference, College Park, MD, USA, 8-12 April 2013
Global Space-based Inter-Calibration System
• What is GSICS?– Initiative of CGMS and WMO– An effort to produce consistent, well-
calibrated data from the international constellation of environmental satellites
• What are the strategies of GSICS?– Best practices/requirements for
prelaunch characterisation– Improve on-orbit calibration by
developing an integrated inter-calibration system• Initially by LEO-GEO Inter-satellite/
inter-sensor calibration
• This will allow us to:– Improve consistency between
instruments– Produce less bias in Level 1 & 2
products– Retrospectively re-calibrate archive
data– Better specify future instruments
How We Make GEO-LEO IR GSICS Corrections
Conclusions• Double differencing method based on the three-way comparison of Metop-A/IASI and Metop-B/IASI in the channel
space of SEVIRI• Small differences between these functions provides an indirect comparison between the IASI instruments on Metop-A
and –B,- which is not possible by direct means due to their orbital configurations
• This approach allows different references to be used to generate Fundamental Climate Data Records (FCDRs) by inter-calibrating a series of instruments, while ensuring their traceability to a common reference
Why do we need to transfer References?
GSICS Principles
1. Comparison of Collocated Radiances
Simultaneous near-Nadir Overpassof GEO imager and LEO sounder
• Collocation Criteria:• ΔLat<35° ΔLon<35°• Δt < 5 min• Δsecθ < 0.01
(Atmospheric path diff.)
• Concentrated in tropics~1000 collocations/orbit~1-4 orbit/night
Schematic illustration of the geostationary orbit (GEO) and polar low Earth orbit (LEO) satellites and
distribution of their collocated observations
2. Data Transformations (Spectral and Spatial)
Spectral Convolution:• Convolve LEO Radiance Spectra with GEO Spectral Response Functions• to synthesise radiance in GEO channels
Spatial Averaging:• Average GEO pixels in each LEO FoV• Estimate uncertainty
–due to spatial variability–as Standard Deviation of GEO pixels
• Use in weighted regression
LEO FoV~12km
~ 5x5 GEO pixels
Example radiance spectra measured by IASI (black) and modeled by LBLRTM (grey), convolved with the Spectral
Response Functions of SEVIRI channels 3-11 from right to left (colored shaded
areas).
Illustration of spatial transformation. Small circles
represent the GEO FoVs and the two large circles represent the LEO
FoV for the extreme cases of FY2-IASI, where n xm=3x3 and SEVIRI-IASI, where n xm=5x5
3. Comparison by Regression
• Compare collocated observations• GEO radiance
• Spatially averaged
• Regressed against• LEO radiance spectra,
• convolved with GEO SRF
• Weighting= Noise + Variance of GEO
radiances• to estimate uncertainty on each collocation
Weighted linear regression of LGEO|REF and <LGEO> for Meteosat-9 13.4μm channel based on single
overpass of IASI
Bias Monitoring from GSICS Corrections
Inter-calibration Bias Changes in Meteosat-7/MVIRI
Water Vapour Channel:Constant Bias ~+2.5K
Infrared Channel:Twice yearly oscillationLong-term trendBias grows -2K to -3K / 5
yr
=> Would have strong impact on product derivation, e.g. OLR or UTH
Time series plot showing relative bias of IR channels of Meteosat-7/MVIRI (MSG2) wrt
Metop-A/IASI, expressed as brightness temperature difference
for standard radiance scenes (1976 US Standard Atmosphere with clear
sky).
Inter-calibration Bias Changes in Meteosat-9/SEVIRI
Most channels show small (<0.4 K) and stable biases during this period.
13.4 μm : negative bias,• due to absorption by ice• grew larger between decontaminations• in December 2008 and February 2013• when bias was reduced by about 0.7 K.
Time series plot showing relative bias of IR channels of Meteosat-9/SEVIRI (MSG2) wrt
Metop-A/IASI, expressed as brightness temperature difference
for standard radiance scenes.
Inter-calibration Bias Changes in Meteosat-10/SEVIRI
2012-07-05 Launched2012-12-12 OperationalSmall (<0.4 K) biases in most channels
13.4 μm : negative bias,• due to absorption by ice• grew larger between decontaminations• in 2012-08 and 2013-12• when bias was reduced by about 0.7 K.
3.9 μm : variable bias,• due to interference by thin ice film building up on optics• follows sinusoidal variations
Time series plot showing relative bias of IR channels of Meteosat-10/SEVIRI (MSG2) wrt
Metop-A/IASI, expressed as brightness temperature difference
for standard radiance scenes
TRACEABILITY /
UNBROKEN CHAINS OF
COMPARISONS
What are GSICS Products?
• Systematic generation of inter-calibration products• for Level 1 data from satellite sensors• to compare, monitor and correct the calibration of monitored instruments
to community references• by generating calibration corrections • with specified uncertainties• through well-documented, peer-reviewed procedures• based on various techniques to ensure consistent and robust results
• Delivery to users• Free and open access• Adopting community standards
• To promote• Greater understanding of instruments’ absolute calibration,
by analysing the root causes of biases• More accurate and more globally consistent retrieved L2 products• Inter-operability for more accurate environmental, climate & weather
forecasting products
• Satellites (and their instruments) have finite life spans• Allows other satellite’s observations to be metrologically
traceable to a single reference
Advantages in Combining Multiple References• Robustness• In case of failure of one reference• Allows transition between references – e.g. Metop-A-
>B
• Greater coverage of diurnal cycle• Both scene and instrument calibration variability• Important for 3-axis stabilized spacecraft
Þ Define only one as the calibration referenceÞ All others are regarded as calibration transfer
standards
GSICS Correction
Correction using
Reference 1
Correction using
Reference 2
Delta Correction Reference
2-1
Double-Differencing with Meteosat-9/SEVIRI as Transfer
Radiometer
• Direct Comparison Metop-B & -A impossibleOrbits are 50min/180° out of phase
• Meteosat-9/SEVIRI can be collocated with overpasses of each Metop
Use as transfer standard
• Define Delta Corrections in SEVIRI channel spaceStrictly Meteosat-9/SEVIRI But SRF Differences expected to have negligible impact (TBC)
• Delta Corrections have same form as GSICS Correctionse.g. linear function of radiance – but may be zero!But include finite uncertainty
Schematic diagram showing how double differencing against a third sensor as an intermediate transfer reference can be used to inter-
calibrate two instruments without requiring direct comparison of their observations. Dashed red lines show collocated observations from pairs
of instruments. Black arrows show calibration transfers.
Metop-A/ IASI
Metop-B/ IASI
Meteosat-9/ SEVIRI
(Met9/ SEVIRI-MetopB/ IASI)-(Met9/ SEVIRI-MetopA/ IASI)
Time Series of Standard Biases(Met9/SEVIRI-MetopB/IASI) & (Met9/SEVIRI-
MetopA/IASI)Shaded areas represent k=1 uncertainty
Distribution of GSICS Corrections to Users with “Delta Correction” to Convert between references
Meteosat-9/SEVIRI Channel
Mean Difference of Standard Biases [K]
Std Dev of Difference of Standard Biases [K]
IR3.9 +0.006 0.022
IR6.2 +0.028 0.009
IR7.3 +0.025 0.029
IR8.7 +0.014 0.021
IR9.7 -0.036 0.019
IR10.8 -0.025 0.022
IR12.0 -0.017 0.020
IR13.4 -0.000 0.010
Statistics of Double Difference (Met9/SEVIRI-MetopB/IASI)-(Met9/SEVIRI-
MetopA/IASI)From Demonstration GSICS Re-Analysis Corrections
From first 2 months data from Metop-B/IASI: 2012-12-16/2013-02-14
Biases for standard scene radiances (clear sky)
[Total Uncertainty (k=1) ~0.01K]
• Double differences are small (<|40mK|) for all channels • Good News!• But still need delta correction!• With uncertainties!
Meteosat-9/SEVIRI Channel
Mean Difference of Standard Biases [K]
Std Dev of Difference of Standard Biases [K]
IR3.9 -0.72 0.28 (<<u)
IR6.2 +0.11 0.05
IR7.3 +0.15 0.18
IR8.7 -0.11 0.30
IR9.7 -0.17 0.08
IR10.8 -0.30 0.29
IR12.0 -0.25 0.24
IR13.4 +0.67 0.28
Biases for cold cloud radiances (220K)
[Total Uncertainty (k=1) :~0.7K for IR3.9 ~0.2K for IR10.8]
• Double differences are small (<|1K|) for all channels • More Good News!
• Inter-calibration of Meteosat-10/SEVIRI show changes, particularly the 13.4μm and 3.9μm channels, due to ice contamination on the optics, modifying the spectral responses.
• These changes were rapid during the first few months, but have become more stable and the calibration of all channels is now within 1 K of the both Metop-A/IASI and Metop-B/IASI.
top related