Diapositive 1

1
Megha Tropiques MADRAS algorithm status BRAIN
Franck Chopin (LMD/ICARE) Nicolas Viltard (CETP)
2
Principle of BRAIN

• BRAIN is a bayesian-based algorithm meant to
  retrieve rain and precipitation profile from TMI
  data
• Its retrieval database is made of co-located PR
  and TMI data
• It works over land and ocean with slightly
  different principles and database (only 85 Ghz
  over land)

Colocation example Orbit 10915 Diamonds PR
pixels Bold Diamons nadir PR Pixel Plus TMI
pixels Bold stars Middle of TMI swath
Blue line nadir PR Pixel
gt Position of PR and TMI relative center swath
changes during the TRMM revolutions
3
Principle of BRAIN database building
4
Flow-diagram of BRAIN retrieval part
5
BRAIN database characteristics and sanity checks
Brain vs. PR for validation dbase
Database histogram of rain intensity occurence
Error and S. Dev of error for validation dbase
6
Retrieval example
Reference PR rain at 37 Ghz resolution
7
Flow-diagram of BRAIN TB simulation
8
TB simulation from dbase scenes and comparison
with TMI
Tbs observed
Tbs simulated from PR swath cloud model
9
TB simulation and influence of ice
parametrisation in RTM
Tbs observed
The 85GHz is particularly sensitive to ice
parameterization and specially the
density-diameter law used in RTM
Tbs simulated from PR swath cloud model
Two realisations of TB 85 Ghz H, with only the
mass-diameter that was changed...
10
Flow-diagram of BRAIN for other satellites
TMI TB
RTM Error assessment
11
Example adaptation for SSM/I beta version
No transfert radiative performed, just Tb and
rain resolution changed
12
HISTOGRAM COMPARISON BETWEEN SSM/I AND TRMM
GREEN SSMI HISTOGRAM RED RESCALED TRMM
HISTOGRAM
H19
V21
H85
H37
13
Flow-diagram of BRAIN integrated with all
platforms
TMI TB
RTM Error assessment
14
Conclusions

• Still a lot of work to be done...
• Adaptation to MADRAS (code part) should start
  early 2006 (6 months)
• MADRAS beta database should start also early 2006
  (3 months)
• Complete base with radiative transfer should be
  done by end of 2007 with improved ice-phase
  (probable start after AMMA)
• Use of 157 Ghz will be studied in parallel

Open questions
What about coupling of MADRAS and SAPHIR ? Should
we use ancillary data ? What about coupling with
MSG ? gt nicolas.viltard_at_cetp.ipsl.fr