MIRS Science and other Updates

1
MIRS Science and other Updates

• S.-A. Boukabara and K. Garrett

2
Output Format

• We want to minimize as much as possible the
  format changes. Even if they occur, it will be
  seamless (as the library we provide to read the
  data reflects the changes)
• Please suggest all parameters you need to see in
  output files, so we can add them.
• The following are being added
• Corrected TBs
• Solar zenith angle
• Relative azimuth angle
• Version number of algorithm that generated file.

3
Publications

• MIRS Presentations for the upcoming MicroRad
• The retrievals of sea-ice concentration from
  variationally-retrieved microwave surface
  emissivities. Kongoli, Boukabara, Yan, Weng and
  Ferraro
• The retrievals of effective grain size and snow
  water equivalent from variationally-retrieved
  microwave surface emissivities. Kongoli,
  Boukabara and Weng
• Theoretical study of the impact of scanning
  geometry on microwave sounding of temperature and
  humidity. Boukabara, Garrett and Weng
• NOAA-08 An Optimal Atmospheric Dataset for
  Algorithm Training and Covariance Matrix
  Generation.. Garrett, Boukabara and Weng
• The goal is to have these presentations turned
  into 4 papers submitted to the special issue of
  MicroRad08 or another journal by March 08.

4
MIRS Release to Scientific Community

• Got approval so far from
• STAR Security Coordinator,
• Funding agency (OSD, Tom Schott, NDE Jim Silva)
• Al Powel (pending licensing agreement in place)
• All we need is a licensing agreement (we have an
  almost complete draft based on ECMWF and NWS
  agreements). Will shortly circulate it for
  approval.
• We have everything in place technically to
  release the package (password protection, online
  download system, etc)

5
MIRS Interface with NDE

• We had a follow-up technical meeting with NDE at
  NSOF to describe technical details of MIRS
  Interface/Integration within NDE system.
• Several changes are likely necessary in our
  technical implementation and output formats
• Main change is likely to be
• Passing output files in PCF (instead of
  constructing them in script. They would need to
  be done by GUI).
• Extending version numbers inside all data and src
  files (as we might be passing individual patches
  in case of emergency fixes)
• Transferring several items in the PCF into the
  Tuning file, to shield them from being registered
  as possible modifiable options.

6
Validation System (sounding)

• Our validation results presented by Kevin at the
  last meeting were confirmed independently, by
  Tony Reale and Bomin Sun. So our baseline
  performances are fine.
• Plan is to extension MIRS daily validation
• Comparison to COSMIC Data (refractivity). We have
  access to these data through DDS (credit to Cathy
  Nicols). Need to work on a reader (on going
  contact with Lidia and Thomas)
• Comparison to ECMWF data. Issue with accessing
  data. Contact in progress with Walter Wolf and
  Mitch Goldberg to revise the ECMWF agreement if
  needed.
• We are entering a different phase in the
  validation stage checking profile by profile for
  (1) lapse rates, (2) supersaturation, etc. (even
  if statistically we are doing very good, we need
  to make sure every profile meets certain
  standards).

7
Impact of Relative Humidity and loosening
constraint on First Guess (Land/Temp)
This is based on comparisons with ICDB datasets
8
Impact of Relative Humidity and loosening
constraint on First Guess (Land / WV)
Setting the RH maximum value to 100 reduces the
stdeviation
This is based on comparisons with ICDB datasets
9
Impact of Relative Humidity and loosening
constraint on First Guess (Ocean/Temp)
This is based on comparisons with ICDB datasets
10
Impact of Relative Humidity and loosening
constraint on First Guess (Ocean/WV)
Setting the RH maximum value to 100 reduces the
stdeviation
This is based on comparisons with ICDB datasets
11
Whats next

• We will update the OSDPD with a flag change
  (extUseQ) and the max RH (100 instead of 120).
  This helps a number of individual profiles and
  improve slightly the WV stats. This also helped
  the overall retrieval fields by removing the
  stripes due to angle stratification of the first
  guess.
• Re-training of the first guess based on GDAS
  (many super-saturated profiles found in GDAS
  fields used for training).
• We might update the regression coeffs files to
  OSDPD if this turns to improve the results.
• We did mostly 90-95 of the work to reach good
  performances of T and Q. The remaining 5-10 are
  going to be step by step improvements.