Using GPS-RO to evaluate Climate Data Records from MSU/AMSU Carl Mears, Remote Sensing Systems
AMSU Characteristics Cross-Track sounders that measure near/on the Oxygen absorption complex at 60 GHz. Different channels measure average temperature for different deep layers in the atmosphere. 6 channels (9-14) measure the stratosphere. AMSU 9 is similar to MSU channel 4 So the AMSU 9 record extends back to late 1978 Other AMSU channels (10-14) Begin in mid 1998. Measure higher in the stratosphere than MSU4/AMSU9
AMSU Stratospheric Channels Channel 13 Channel 12 Channel 11 Channel 10 TLS MSU4 and AMSU9
MSU/AMSU Climate Data Records Instruments are cross-calibrated using extensive comparisons of results from co-orbiting satellites. This type of analysis Can Detect: Drifts in results from single satellites. Drifts caused by changing local measurement times/diurnal tides. Can t Detect: Drifts that occur in all satellites at the same time instrument aging? Radiosondes cannot provide unambiguous validation, particularly in the stratosphere.
Can GPS-RO Data be used to investigate (validate?) trends in AMSU stratospheric channels?
GPS AMSU comparisons In the most recent decade (2001-2010) we can compare AMSU results to results from GPS-RO Focus on CHAMP results: Advantages: Long time series available (2001-2008) Relatively small data volume Relatively stable number of observations per month Possible Issues/Disadvantages: Timing of CHAMP observations not always optimal Not that many collocations Study AMSUs on NOAA-15, NOAA-16, and AQUA Long time series. AQUA is in controlled orbit local measurement time is almost constant. Note: Naïve User of GPS-RO Data!
Approach Use UCAR atmprf data from CHAMP (2009_265 version) Use our RTM to calculate synthetic AMSU brightness temperatures from CHAMP profiles for AMSU channels 9-12 Use individual collocations (not monthly averages calculated before collocation) Spatial collocation: is the center of CHAMP observation in 2.5 degree AMSU grid cell? Temporal collocation: is observation within 6 hours? Collocation criteria relatively loose. Why?
NOAA-15/CHAMP Collocations February, 2005 100 50 Latitude 0-50 -100-6 -4-2 0 2 4 6 Time Difference (Hours)
NOAA-15/CHAMP Collocations April, 2005 100 50 Latitude 0-50 -100-6 -4-2 0 2 4 6 Time Difference (Hours)
Distribution of Differences AMSU 9 minus CHAMP
Distribution of Differences AMSU 10 minus CHAMP
Distribution of Differences AMSU 11 minus CHAMP
Distribution of Differences AMSU 12 minus CHAMP
Summary Increasing standard deviation with increasing channel number is expected. Both AMSU and CHAMP are noisier with increasing altitude.
Interannual Trends Can we use CHAMP/AMSU collocations to look for anomalous trends in AMSU? January 2005 1 month ~1400 collocations Optimistic analysis σ mean =~0.03K AMSU CHAMP (K) Actual AMSU-CHAMP differences (and their variability) are much larger
Global Mean of Collocated Differences AMSU Channel 9 Trends are 2003-2008 for all 3 satellites Channel 9 on NOAA-16 has a long-term drift relative to NOAA-15 and AQUA.
Global Mean of Collocated Differences AMSU Channel 10
Global Mean of Collocated Differences AMSU Channel 11
Global Mean of Collocated Differences AMSU Channel 12 Note Scale Change!
Trend Differences (AMSU CHAMP) These are not small! Comparable to or larger than AMSU cooling trends.
Some Observations Shape of Short Term Variability similar for all 3 AMSUs This is despite differences in local time, instrument temperature, etc. (Remember, local time for AQUA is fixed) Short term variability is similar in shape for different channels, but becomes larger for AMSU channels higher in the stratosphere. 2003-2008 trends are similar, and are all of the same sign. Difference trends tend to increase higher in the stratosphere, similar to the variability
Discussion of Causes: Can we blame AMSU? Short-term variability so similar between satellites. Thus difficult to explain by AMSU calibration drifts or sampling issues. Difference trends very similar between satellites. Makes the diurnal cycle/diurnal tides an unlikely culprit. Make errors due to changes in instrument temperature an unlikely cause. What about CHAMP? Ideas welcome.