GR_Day2_Data Storage System

advertisement
Data Storage System, Model
Output & Analysis Tools
PRECIS Workshop, Exeter, 27th – 30th May 2014
© Crown copyright Met Office
Contents
• Data locations
• STASH codes
• File naming convention
• PP analysis utilities
© Crown copyright Met Office
$ARCHIVEDIR
• PRECIS outputs its data to a directory within the
$ARCHIVEDIR directory
• $ARCHIVEDIR/RUNID
© Crown copyright Met Office
The STASH code
• Each variable type (temperature, precipitation, etc.) from
the model is written to a separate directory defined by
that variable’s 5-digit ‘STASH’ code
• $ARCHIVEDIR/RUNID/03236
• 03236 = Mean temperature at 1.5m
• $ARCHIVEDIR/RUNID/05216
• 05216 = Mean total precipitation
• For a full list, see Appendix C of the PRECIS technical
manual
© Crown copyright Met Office
Output file naming convention
• Files follow this general pattern:
• [RUNID]a.[file type][Date].[STASH code].[format]
• Examples:
•abdaba.pji9140.03236.pp
•abdaba.pji9140.03236.pp
• All filename types are listed in Appendix D of the technical
manual
© Crown copyright Met Office
Data analysis utilities
included with PRECIS
© Crown copyright Met Office
Output Data format
• The PRECIS models output data in the PP binary data
format. This is the Met Office’s own format.
• Conversion tools are provided to convert PP data to
• netCDF
• Grib (for GrADs)
• ArcInfo ascii (e.g. for GIS/spreadsheets)
• See Section 6.2 of the technical manual for further
discussion.
© Crown copyright Met Office
Analysis Utilities
• To use the output data in any way, you will need to
make use of some kind of data analysis software utility
• The utilities GrADs, Ferret and CDAT are included on
the PRECIS DVD, but no training in the use of these is
given nor will any technical support in their use be
provided!!!
• There is a suite of data analysis tools for PP data
which are installed as part of PRECIS and are fully
supported by the PRECIS team
© Crown copyright Met Office
Questions
© Crown copyright Met Office
The UM Date Stamp (reference)
• Reduces number of characters used in setting date
• First digit indicates the decade:
[0-9a-z]
• Actually number of decades since 1800. E.g.:
• 0  1800-1810
• a  1900-1910
• g  1960-1970
• r  2070-2080
• Second digit indicate year in decade
[0-9]
• Third digit represents month:
[1-9a-c]
• Fourth digit is the day of the month:
[1-9a-v]
• Fifth digit is the hour of the day:
[0-9a-o]
© Crown copyright Met Office
Some examples (reference)
• r4c10
• 0:00z 1st December 2074
• h93ab
• 11:00z 10th March 1979
• j3jun
• June 1993
• l35eb
• 11:00z 14th May 2013
© Crown copyright Met Office
“Climate Mean” Output
(reference)
• For models with a 360 day calendar (such as those
driving the HadAM3P GCM) up to four climate meaning
periods may be set
• These allow for long timescale means to be calculated
within the model (rather than by the user)
• Choice of means is determined by a comma separated
list representing shortest to longest
• Meaning periods are nested, and each one is specified
by how many multiples of the previous (shorter) period
it is.
© Crown copyright Met Office
“Climate Mean” Output (reference)
• Example 1: 30,3,4,10
• 30 :
30 day (monthly) means
• 3 : 30 * 3 = 90 day (seasonal) means
• 4 : 90 * 4 = 360 day (annual) means
• 10 : 360 * 10 = 10 year (decadal) means
© Crown copyright Met Office
“Climate Mean” Output (reference)
• Example 2: 30,4,3,5
• 30:
30 day (monthly) means
• 4 : 30 * 4 = 120 day (4-month seasonal) means
• 3 : 120 * 3 = 360 day (annual) means
• 5 : 360 * 5 =
© Crown copyright Met Office
5 year (decadal) means
“Climate Mean” Output
(reference)
• Models with the Gregorian calendar (ERA) have only
monthly, 3-month seasonal and annual means available.
• The Base Date for climate meaning sets the reference date
for the beginning of means.
• The ‘infinite time series’ of all climate means will coincide
with this date.
• Example: If the base date is set to 01 Apr 1960, four month
seasonal means will be apr/may/jun/jul, aug/sep/oct/nov,
and dec/jan/feb/mar. Annual means will be represented by
01 Apr – 30 Mar.
• Take the initial spin-up year into consideration when
choosing the base date.
© Crown copyright Met Office
CDO and NCL – what
are they?
CDO and NCL are tools for data analysis and
visualisation.
Different tools are useful for different activities e.g. CDO
tools for analysis and NCL for visualisation.
In the worksheets these tools are used together to
process, analyse and visualise PRECIS output.
© Crown copyright Met Office
CDO – Climate Data Operators (1)
Developed by the Max Planck Institute for use with GRIB 1/2,
netCDF 3/4, SERVICE, EXTRA and IEG data.
Designed specifically for climate and NWP data analysis, there
are more than 600 operators available.
Can be run on Linux, Windows, MasOS and others.
© Crown copyright Met Office
CDO examples
• Calculate the JJA mean:
cdo selmon,6/8 infile.nc jja.nc
cdo timmean jja.nc jja.mean.nc
or, pipe the operators together using ‘-’
cdo timmean –selmon,6/8 infile.nc jja.mean.nc
• Re-grid data onto a target grid:
cdo griddes target_grid.nc > mygrid
cdo remapbil.mygrid
© Crown copyright Met Office
file.nc regridded_file.nc
CDO – Climate Data Operators (3)
Free to download and documentation and support forums can be
found at https://code.zmaw.de/projects/cdo
© Crown copyright Met Office
NCL – NCAR Command Language (1)
• Developed by NCAR (Nation Center for Atmospheric Research) for
use with GRIB 1/2, netCDF 3/4, shape files and ascii data.
• Free interpreted language designed for scientific data processing
and visualization.
• Can be run on Linux, Windows, MasOS and others.
• Can be typed directly into the NCL command line (very time
consuming), or scripted.
• Produces plots of publishable quality.
© Crown copyright Met Office
NCL – NCAR Command Language (2)
Extensive documentation and hundreds of example scripts and plots
at http://www.ncl.ucar.edu/index.shtml
© Crown copyright Met Office
NCL – NCAR Command Language (3)
Looks more complex than
it is.
© Crown copyright Met Office
Using tools together
The worksheets combine these tools to post process and analysis
PRECIS data.
1. Use CDO to analyse the netCDF data.
2. Use NCL to visualise (and process) the data.
© Crown copyright Met Office
Download