Issue 4: Reproducibility of temperature results
The comparison of temperature retrievals from four different centers showed very similar changes over time. This is a good sign that the key measure of temperature was useful for climate change purposes. But the differences suggest that retrieved temperature should not be considered a benchmark observation. Analysis quantifying the differences (both average difference (bias) and absolute value of the differences) by latitude bands and within latitude bands over land and ocean separately would help clarify the appropriate role of GPS-RO temperature retrievals in climate analyses.
Response from Kevin Trenberth:
Agreed
5 Comments
Tom Yunck
True enough, but this question should be revisited from time to time. These are early days.
Tony Mannucci
We cannot draw firm conclusions so soon in the comparison study. This statement of this issue raises important questions that need to be addressed.
Ben Ho
Response from Ben Ho
To make a consistent comparison for issue 3 (comparisons of monthly 5-degree lat mean
fractional refractivity among four centers), here pixel level GFZ refractivity profiles
from Jan. 2006 to Dec. 2006 are paired with those of UCAR for now.
Dry temperature difference may be 1.5 to 2 times larger than the fractional refractivity
difference here.
Mean fractional refractivity differences between GFZ and UCAR at different latitudinal bands
are quantified.
Time series of fractional refractivity difference between GFZ and UCAR at different vertical layers
for different latitudinal zones are also shown here.
Figs. of Detail Response
Detail Response
Results shown here indicate that the mean fractional refractivity difference between GFZ and UCAR
is within 0.03%, where their mean MAD (median absolute deviation) from 8 km to 30 km is about 0.16 %.
Results here also demonstrate that sampling errors can be eliminated (issue 3) in a pixel to pixel
comparison.
The dry temperature comparison among four centers will be conducted when data are available.
Seth Gutman
Consolidated Comments by Gutman, Yoe and Reale
Seth Gutman's Response
I agree.
Jim Yoe's Response
Yes, but for the stratosphere generally the retrieved temperatures may be the most reliable means to determine trends of temperature and of tropopause height.
Jens Wickert
Response from Jens Wickert
In general I would expect larger deviations in temperature (compared to the refractivity), when comparing data from different groups but the same satellite. Different implementations of the initialization should be responsible for this. But I'm also sure that the differences can be explained by detailed investigations.
Each processing step to "higher level" data products (temperature and water vapor) from the excess phases, bending angles and refractivity introduces small, but uncertainties.