1. Precipitation Cluster Processing chain
ITALIAN AIR FORCE MET SERVICE C.N.M.C.A. - Satellite Branch
Precipitation Cluster Processing chain
Cap. Davide MELFI
H-SAF Training and Users Service, Rome, December 2009

2. TOPICS HW architecture Overview on processing chains
Changing in PR-OBS03 processing chain

3. Previous HW architecture for HSAF’s products
Now we have all hw to complete the implementation of all lines of production of HSAF.
So during the next months the wp 2100 will be dedicated to finalize the operative chain, because an important review from EUMETSAT will be done in October.
People of wp 2100 must work to following document:
Algorithm Theoretical Definition Document (ATDD)
Component Design Document (CDD)
Component Verification File (CVERF)
Product Prototyping Report (PPR)
System Software system Document (SSVD)
System Validation File (SVALF)
Data Output Format (DOF)
In particular the Component Verification File , Product Prototyping Report , System Software system Document , System Validation File ,
Data Output Format describe the functionality of OC at Host Institute. So this documents will be the guide for the inspections at Host Institute by EUMETSAT.

4. New HW architecture for HSAF’s products

5. BUFR output:
Encoded by ECMWF libemos
with B
ASCII output:
sea/land/coast indication
PNG output:
Made with GMT freeware SW
BUFR output:
Encoded by ECMWF libemos
with B
ASCII output:
sea/land/coast indication
PNG output:
Made with GMT freeware SW
GRIB2 output:
Encoded by ECMWF grib_api
PNG output:
Made with GMT freeware SW
GRIB2 output:
Encoded by ECMWF grib_api
PNG output:
Made with GMT freeware SW

6. Header
ftp://ftp.meteoam.it/utilities/Bufr_decode/

7. ftp://ftp.meteoam.it/utilities/
Processing Time
Products
ONLINE
OFFLINE
OUTPUT format
ftp://ftp.meteoam.it/utilities/
PR-OBS01
3 min (input – 2,5 h delay)
BUFR
PR-OBS02
30 sec (input – 5 min delay)
1 month /1 week + Recovery from archive
PR-OBS03
1 min
GRIB2
PR-OBS05
3 min

8. Changing in PR-OBS03 processing chain
H03: version 1.0
First release
H03: version 1.1
Input in BUFR format
Decreasing of computational time:
Optimization of the SW
Masking of IR data (cloudy pixels)
H03: version 1.2
Exclusion of Semi-transparent cloud
OBS 1: The schedule provided the issue of i.2 version before the march. The delay does not allow the reprocessing within the current year
OBS 2:Inclusion of METOP, NOAA 19 and correction of channel 4. At the moment is not clear the impact of the upgrade about the new channel.
OBS 3: The parallax correction is ready. The discrimination of thin clouds is ready. If PT approves the correction will be insert. I propose to collect a data set to cal/val
OBS 4: The algorithm should be ready for integration June. Attempt to insert OBS 4 in o.c. before the end of summer.
PR-ASS1: No enlargement of area at the moment. Will be considered in the CDOP
H03: version 1.3
Parallax error correction

9. Masking NWCSAF - Cloud type
OBS 1: The schedule provided the issue of i.2 version before the march. The delay does not allow the reprocessing within the current year
OBS 2:Inclusion of METOP, NOAA 19 and correction of channel 4. At the moment is not clear the impact of the upgrade about the new channel.
OBS 3: The parallax correction is ready. The discrimination of thin clouds is ready. If PT approves the correction will be insert. I propose to collect a data set to cal/val
OBS 4: The algorithm should be ready for integration June. Attempt to insert OBS 4 in o.c. before the end of summer.
PR-ASS1: No enlargement of area at the moment. Will be considered in the CDOP

10. Parallax error
EFFECT OF OBSERVATION ZENITH ANGLE AND CLOUD TOP HEIGHT ON PARALLAX DISPLACEMENT
SUB SATELLITE POINT SSP
PARALLEX DISPLACEMENT

11. Parallax error correction
EUMETSAT subroutine
(Marianne König )
NWCSAF - CTH
Scheme of parallax correction OBS 1.3
The input of algorithm is not the simple 10.8, it is the 10.8 filtered by:
A cloud mask;
A movement of pixel by parallax correction;
The cloud mask is arranged by the CLOUD TYPE product of NWC-SAF
The parallax correction is calculated from the CTH of NWC-SAF
Case study :42

12. Parallax error correction
May happen that two different pixels, with different height, are projected into a same pixel. We agree that higher
The figure shows us the moving of PIXEL in the new reprojection.
2 pixels in the same new one
The colder one remains

13. Delayed running HSAF 03:15 RUNNING 03:30 05:45 SEVIRI MSU SEVIRI MSU
03:00
03:15
03:30
03:45
04:00
04:15
04:30
04:45
05:00
05:15
05:30
05:45
Timeline
HSAF 03:15
RUNNING
03:30
05:45
SEVIRI
MSU
SEVIRI
MSU
SSMI
Acquisition time
Receiving time

14. Contribute of SSMI/S data to OBS5
Cum. Prec
Last 3 H.
3 H before
12-9 before
24-12 before
Last 3 H.
9-6 before
6-3 before
Last 3 H.
12-9 before
9-6 before
Ho we fill in the OBS 5
You can see the cumulate at 3, 6, 12 and 24 hours.
The blu box represents the last 3 hours time.
So The blu box are worked with AMSU data instead the red box are worked with SSM data
6-3 before
Last 3 H.

15. Thanks for attention!

16. CNMCA set up a reanalysis environment (OBS1,2,3), with a suite ready for June for comparison of same case studies with different ingredients to compare their impacts (parallax correction, etc.)
TESEO
PERSEO
MEDUSA

17. Acquisition time Receiving time Timeline
03:00
03:15
03:30
03:45
04:00
04:15
04:30
04:45
05:00
05:15
05:30
05:45
Timeline
Receiving time
Acquisition time
OBS 1 l’algoritmo è completo ed è previsto aggiornamento data base al termine del VS. Non alimenta OBS3, ritardo DMSP. Ipotesi di inserimento:
Produzione parallela e differita di 3 ore;
OBS 2 v1 sarà operativo fino alla installazione on-line della c.o. del OBS 2 v2. Attualmente alimenta OBS 3 e successivamente OBS 5
OBS 2 v2 off line fino alla fine dell’anno
OBS 3 è prevista ulteriore modifica: istogramma, ottimizzazione del sw impatto su cpu time
OBS 5 al massimo inizio maggio