Proposed naming convention for the Transverse AMIP (TAMIP) experiments
1st Draft 23rd April 2010
Author: Charlotte Pascoe (BADC) with advice from Mark Elkington (Met Office)
We propose that the Transverse AMIP (TAMIP) experiments be considered as 4 ensemble experiments within the CMIP5 framework. It is our understanding that the community of users who will use the TAMIP experiment data will want to search for data from individual TAMIP periods as these are associated with the Intense Observation Periods (IOPs) from the Year Of Tropical Convection (YOTC).
Extract from TAMIP experiment design:
"4 sets of 16 hindcasts are to be run, the first in each set starting at 00Z on the 15th of the following months and then subsequently at 30 hour intervals: October 2008, January 2009, April 2009 and July 2009."
TAMIP Naming Convention
We propose the following naming convention for TAMIP in CMIP5.
- 8.1 tamip200810
- 8.2 tamip200901
- 8.3 tamip200904
- 8.4 tamip200907
TAMIP DRS Structure
Proposed drs structure for TAMIP:
/CMIP5/MOHC/HadGEM3/tamip_200810/3hr/atmos/tas/r1s1/v1/tas_Ahour_HadGEM3_tamip200810_r1s1_2008101500-2008102000.nc
New ensemble type s# "staggered start"
We propose a new staggerd start ensemble type to describe the TAMIP experiments. Staggered start ensemble members require the following additional metadata which will be collected in the CMIP5 questionnaire:
- start date
- duration
Staggered start ensembles have one shared set of boundary conditions which specify boundary conditions from the start of the first ensemble member to the end of the last ensemble member. The user will map the ensemble to a set of input files rather than mapping each ensemble member to a specific input file. Each ensemble member adheres to that portion of the boundary conditions which are consistent with it's start date and duration.
Staggered Start for Decadal Experiments
The staggered start ensemble type can also be applied to the decadal experiments. Using staggerd start has the potential to reduce the number of 1.# decadal experiments that the user sees in the questionnaire from 45 to 6.
The 1.1 decadal hindcast/forecast experiments are described as 10 separate experiments in the CMIP5 DRS:
- decadal1960
- decadal1965
- decadal1970
- decadal1975
- decadal1980
- decadal1985
- decadal1990
- decadal1995
- decadal2000
- decadal2005
Modelling centres have also been asked to do additional runs to build ensembles for each of the decadal simulations (experiments 1.1-E 1960,1965,...,2005)
If the decadal experiment were instead described as a staggered start ensemble then all of the 1.1 decadal experiments listed above could be contracted into a single experiment. The differing start dates would be identfied by the s# staggered start ensemble notation and the ensemble of experiments for each start date would be identified by the r# ensemble realisation notation. In other words we describe the decadal experiments using 2 ensemble axes r# and s# so that the list of decadal experiments above becomes:
- decadal_r1s1
- decadal_r1s2
- decadal_r1s3
- decadal_r1s4
- decadal_r1s5
- decadal_r1s6
- decadal_r1s7
- decadal_r1s8
- decadal_r1s9
- decadal_r1s10
The staggered start method requires users to enter only one set of boundary condition information for all the 1.1 decadal experiments. Using staggered start ensmebles will reduce the long list of 45 decadal hindcast experiments to 6!
- 1.1 decadal (1.1, 1.1-E, 1.1-I)
- 1.2 decadal (1.2, 1.2-E)
- 1.3 noVolcano
- 1.4 volcano2010
- 1.5 decadal
- 1.6 chemistry
