A Fundamental Climate Data Record for the AVHRR
description
Transcript of A Fundamental Climate Data Record for the AVHRR
A Fundamental Climate Data Record for the AVHRR
Jonathan MittazManik Bali & Andrew Harris
CICS/ESSICUniversity of Maryland
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
Funded project through NCDC for 3 years (part of NOAA Climate Data Record Program)
Goal To provide recalibrated AVHRR Level 1B radiances for the thermal IR
channels (3.7, 11 and 12 μm channels) which are as accurate and bias free as possible and where the uncertainty on the radiances are better understood.
Source Data NOAA AVHRR Level 1B data
Deliverables – not yet fully defined by likely to be one or more of Code to calculate new radiances from current AVHRR Level 1B files Recalibrated Level 1B data files (all AVHRRs in KLM format) NCDC specific format (netCDF for example)
Part of SW/IR Imager FCDR Team – Team Lead : Bob Evans
AVHRR IR CDR Project
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
• Use a physically meaningful calibration algorithm (current operational calibration (Walton et al. 1998) is not)
• Apply a uniform calibration methodology to the complete AVHRR data record– Current AVHRR Level1B data have a changing
calibration methodology over time. Walton et al. calibration is available for NOAA-7,9,10,11,12,14 and all AVHRR/3s but is significantly biased.
• Reanalyze AVHRR pre-launch data to obtain instrument non-linearity
Calibration Algorithm
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
Calibration Targets -ECT (180->320K) & Space Target @ 70K
No thermal shielding – very simple test chamber. Future pre-launch tests should be done better
Run at 5 instrument temperatures of 10, 15, 20, 25, 30°C
Pre-launch Data
Calibration Test Chamber
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
Example of pre-launch problems (Mittaz, Harris & Sullivan)
Application of the Walton et al. calibration on the pre-launch data from which it was derived shows large biases – sign of severe problems with the pre-launch data and methodology
Can be fixed by the use of a physically based methodology – means that all pre-launch data has to be re-analyzed
Some pre-launch calibration parameters will still be corrupted
Pre-launch Data (2)
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
• Use Physically based calibration equation (pre-launch and TOA)
• Use Top Of Atmosphere calibration sources (e.g. (A)ATSR, IASI etc.) when available to correct parameters contaminated during pre-launch testing (underlined in red)
• Use model of instrument to obtain calibration when contamination exists (solar contamination) when possible
• Remove periods of bad calibration from record• Monitor calibration as a function of time and correct when necessary
Calibration Approach (TOA)
22
)()()(')(EarthSEarthS
ICTS
ICTSICTICTEarth CCCC
CCCCRR
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
AVHRR calibrated by operational scheme – shows large temperature dependent biases
Combination of incorrect algorithm and pre-launch contamination
Correct by fitting corrupted calibration parameters to a TOA calibration radiance source (in this case IASI)
Operational and New Calibration comparisons (IASI)
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
• Assessment of AVHRR/3 stability over 6 months - stable (<0.1K)
Even operational calibration has constant biases to 0.05K
New calibration shows small trends at the < 0.08K level (note change in scale wrt previous plot by ~ factor 10)
220-230K
220-230K
290-300K
290-300K
MetOp-A AVHRR Stability
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
• Now done longer term study – MetOp-A AVHRR stable over 3+ years. Close to climate change requirements (Ohring et al. 2004)
Accuracy = 0.1K Stability (per decade) = 0.04K
MetOp-A AVHRR stable over 3+years
SST Data (>270K)
11 µm
Bias = 0.03KGradient = 0.014 K/decade
12 µm
Bias = 0.03KGradient = 0.004 K/decade
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
MetOp-A AVHRR Thermal Trend
Small drift in average orbital temperature (0.2K in 4 years) with clear seasonal variability
Constancy of temperature may in part explain stability of AVHRR calibration
0.2K
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
• Baseline instrument for re-calibration is the (A)ATSR series– Designed to be climate ready– Accurate and stable to < 0.05K (apart from the AATSR 12µm channel see later)– Data available from 1991 to present day (covers the AVHRR/2 AVHRR/3
instruments)• Data available via FTP
– One months worth of data ~130Gbytes – takes ~ 3 days to download• AVHRR data matched with (A)ATSR data (first attempt parameters)
– Match individual AVHRR GAC ‘pixels’– Take into account true AVHRR GAC footprint– Both AVHRR and (A)ATSR data should be spatially coherent (current limit σ<1K
over ~12x12km area)– Satellite ZA agree to < 1°– Data limited to close to nadir (current limit < 10°)– For daytime 3.7µm channel keep relative azimuth angle to < 30°– Maximum time difference between AVHRR and (A)ATSR data < 10 minutes – Correct (A)ATSR data for differences in spectral response functions
Use of the AATSR as a TOA Calibration Source
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
• Compare 11 and 12 µm channel AVHRR data calibrated using the parameters derived from IASI matches - 11µm good agreement, 12 µm not
Comparison of MetOp-A AVHRR with AATSR (IASI parameters)
Good agreement with a slight (-0.05K) bias – small tweak can make the data match
Strong trend to -0.5K at cold temperatures – highlights issues with AATSR 12µm channel (AATSR Cal Team informed)
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
• In Operation since March 2001 – thought to currently have a bad calibration (e.g. ‘out of family’ from NOAA MICROS pages).- test case for AVHRR near terminator/problem checking
AATSR/NOAA-16 AVHRR Comparison (11µm)
Using calibration from MetOp-A gives a trend and bias(but smaller trends than current calibration)(Data from Feb 2003)
Recalibration removes trend/bias
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
• 7 years after previous calibration now close to terminator– Shows a distinct change in the calibration – time dependent effect
NOAA-16 data taken Feb 2010
Data is biased and shows a trend relative to Feb 2003 calibration
Bias parameters (α,α’) have larger values than in 2003 – impact of change in thermal state
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
Average Instrument Temperature (NOAA-16)
Unlike MetOp-A (0.2K in 4 years), NOAA-16 shows large temperature variations – becomes extreme from ~ 2008. A change in the thermal environment may explain the change in the 2010 calibration biases relative to 2003.
Scan motor problems
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
Misses some contamination
Uses a simple constant to fill
Better detectionof events
Modeled gainincluding uncertaintyestimate
Have much better detection of times of contamination – users can be more certain they are not including bad or corrupted data
Also have a model for the gain for contaminated times including an uncertainty estimate
Again better detectionof events
Misses some contamination
Detection and correction of bad data - solar contamination (3.7μm)
NOAA-14
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
Nighttime
Daytime
Strong correlation of nighttime gain with Earth scene radiance
Note predictive capability of new calibration (also can be used for solar contamination)
Up to 0.25K error in daytime BT @ 295K
Detection and correction of bad data – Earthshine
• 3.7 µm contaminated by Earthshine (light from Earth scattering via Blackbody in calibration) – fix with model of Self emission
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
• Contamination of the Space clamp view (e.g. by the Moon) will make the true instrument gain be unknowable – so these events are removed
Detection and correction of bad data – Space Count contamination
To have an accurate FCDR you need to accurately remove/flag all bad data
Cooperative Institute for Climate and SatellitesUniversity of Maryland5825 University Research Court, (Suite 4001)College Park, MD 20740-3823Tel: (301) 405-2147 Fax: (301) 405-8648http://www.essic.umd.edu/cics
Institute Director: Dr. Phillip [email protected] Director: Andrew [email protected]
• For future missions - good pre-launch testing is critical and needs to be done properly• In orbit comparisons against TOA reference sources is also critical to remove biases
– Most of the tools are in place to recalibrate AVHRR/3 series• AVHRR/2 waiting on pre-launch analysis
• AVHRR has the capability of being used for accurate climate studies – MetOp-A is currently accurate and stable
• Requirement for time/temperature dependence for the calibration– Clear in NOAA-16 (calibration very different in 2010 compared to 2003– Constant instrument temperature -> constant calibration? (MetOp-A)
• Need to remove accurately remove/estimate bad data from record– Tools are in place
• Remaining issues– SRF shift needs to be included for (A)ATSR data– 3.7 µm channel
• Automatic implementation of Earthshine correction – AATSR 12 µm channel needs to be fixed– Look into using RTM data/AVHRR overlap periods when accurate TOA sources not available
(pre-1991)
CONCLUSION