CAB Notes
2/28/07
Peter Gent
Micelel Reinecker
Don Anderson
Rick Anthes
Paola
David Rind
Phil Jones (sub for John Drake)
Dave Bader
Anjuli Bamzaii
Rick Rosen NOAA
Jerry Melillo?
Tom Delworth
Tom Crowley
Jim Hurrell
Isaac Held
Jay Fein
Cecilia Biitz
Eric Itzwiere
Ricky Rood
David Legler
Peter Milne
Neil Swanberg ?
Li Zhang NSF
Scott Denning
Annarita Mariotti NOAA/CPO/CPPA
Jim Kinter
(23)
Peter Gent—presentation “reducing the biases in CCSM
Science highlights: 26 papers, special issue of journal of climate (internal and external authores show collaborative nature of ccsm)
Arctic sea ice highlight (Ceclia Bitz has been working on) In the next decade there’s a substantial A1B scenario reduction.(2010-2019), abrupt decrease (very little sea ice left in sept 2040-2049. Sept is almost ice free by 2050 in ccsm projections. In AGU paper.
As a result, it’s thought that polar bears will be a threatened species. Observed ice cover 1992 image show polar bears use edge of ice sheet.
Priorities:
· coupled FV dynamical core (done)
· Improve major biases in CCSM 4 (double ITCZ and ENSO accomplished)
Updates:
· POP code on ocean. 2 improvements 1)eddy parameterization 2) coefficient varies in the vertical (perhaps some detail on mixing, Peter had a lot to say)
· New advection scheme in ocean model
· PCWG has transitioned to a new version of CICE. (slide with details)
· Land Model: Community Hydrology Project (slide has list of what they’ve done) Amazon slide
· One of the biggest problems for CCSM has been the ENSO. We fixed a lot of this. 3 major changes and they’ve made a big improvement. Neale and Richter had a very good correlation (connective transport and plume improvements). It’s the comb of reducing subtropical and x that procude this result. The deep convective scheme that we’re going to use is being decided by SSC. (probably will be neale richter, though wu is very good in a quantative). Good example of different people coming together with different ideas. Very constructive and scientific. Isaac asks if there’s any other development that would have this much impact? Peter: hard to say but it was dramatic.double itcz errors reduced as much as 40 -50 %. Isaac convective thing has a huge impact. Is the effort devoted to this commensurate with impact? Gent: it’s taken us too long to make changes to the deep convective scheme. Every group wants to make best component, but the model is onlyas good as it’s weakest link. Tom ? have you looked at sea surface? Is this coupled? Gent yes it’s coupled. Tom: occas in obs you do see a double itcz. Why does it show up sometimes in real world? Gent: in reality it does happen for short time but switches back quicker than model. Anjuli MJos? Gent MJOs improved but not right.
· Cloud fraction. We’ve always had problem of too much low cloud in arctic in winter. Reordering cloud fraction calculation. 50% too high. After mod, values reduced over North America.
· Have now put carbon climate code into CCSM 3 with runs at T31x3. Now have stable thing that reps carbon cycle. With this in ccsm3, the carbon cycle makes changes but we still have a stable climate model (the big message). Our carbon cycle seems to have a smaller perturbation than other models.
· Conclusions:
o In lest year we’ve signif red biases
o Enso freq better
o Improved ocean and cice
o About to assemble interim ccsm3.5 to add carbon cycle to up to date version of ccsm
o Paola question: warm fresh water bias. Pent in upwelling regions the water is too warm not enough stratus cloud. We haven’t been able to resolve these biases. The atm circ in tropical has improved with neale-richter. Ricky Rood: do these runs what’s happened with smag diffusion in ocean Gent all without smag in ocean. Rr: diffusion on atm models? Gent have done some runs with diff in atm that made considerable impr in some things but disappeared in coupled version. Rr spectral of fv? Gent: fv. Jay fein: what about chem changes? Gent: they’re really spinning up with chem modules. Not quite ready to do coupled with chem in them. Were originally wanted soemthings that cost a lot. What they really want to predict is ozone in troposphere, but not quite ready. Anjuli: ARGO? Gent: paper in special issue. Michele Reinecker: ? Gent: each of the wg has a list of targets for improvements. It’s a matter of focusing them on the ones that make the most difference. Rick Anthes: why CCSM is less sensitive to carbon cycle, good or bad? Gent: it’s good, the bigger discrep is the amplitude of the positive feedback. Denning the carbon cycle for ccsm 4 does not include dynamic veg. couple of thigns to focuson 1) water wasn’t getting into ground.2)carobn nit, most other models are not n limited and have a tendency to take up too much n. Issac: are you planning on predicting stratepheric ozone? Gent no, stay with troposphere.
Break.
Tom Delworth: Atlanta Decadal Workshop in Miami:
Will have to go over slides. Make sure Tom gets final edit of slides before posting. Workshop recommendations: from slide. Fien: initialed decadal something ready for prime time. Same question about regional modeling? Ocean had same scenario some years ago. Even though it wasn’ ready, a lot was learned form trying. Delworth: feels these thigns should be vetted because of resources needed. Worries abut model drift being interpreted as prediction. Hurrell: can we revisit issue as it relates to CCSM?
Peter Gent: CCSM 4 Implementation Plan
Rienecker asked for CCSM 4 implementation plan at last year’s CAB andd idea was discussed at last year’s annual workshop. Imp plan based on AR5 2013 IPCC deadline (we’re assuming that although this deadline has not been announced the timeframe will be the same as years past). Have all exploratory stuff done by 2008.
In CCSM4, the SSC wanted to improve biases, have some form of carbon cycle, and indirect effect of aerosols. We’re doing this in stages rather than leaving it all to the end. Stage 1 completed by March 2007: Interim version 3.5. Should be a significant advance on BGC controls. Stage 2 to be completed in calendar year 2007. Stage 3 by end of 2008: validate and understand. Target resolution FV 1.9x2.5 (more equiv to t42). Higher resolution for short term sims of FV 1x1.25 (more equiv to t85), x1ocean. Rind: if target is certain resolution, shouldn’t you be developing that in stages 1 and 2. Gent: it’s a matter of having time. Bader: higher res is essential for accurate climate simulation. Rood: our exp at 2.5x1 there’s significant difference. A quantum jump. Held: ENSO is quite sensitive. Gent: Most of the ipcc runs last time were at t85. Bader: dyn veg,
Bill Collins had an idea that we should change to verify CCSM 4 by how well it simulates the 20th century. This requires more runs, but Peter thinks this is the way to go. What if it doesn’t validate, do you change the forcings or the parameters?
Short term climate simulations:
Enter proposed from from Peter’s slides.
Enter advantages from slides.
Enter challenges from slides.
Gent poses questions to group: didn’t catch. Held: running higer res for shorter times and running form real initial conditions
Enter scientific opps from slides.
Crowley: any advantage from impl viewpoint to do high res model runs earlier? Gent: pretty good idea. Have to settle bgc over the next few months. Rind: what is purpose of hig res run? Denning: these are testbeds for the new physics,. Paola: biases in upwelling areas are they not so crucial and a bunch of stuff. Rienecker: presented questions from Eric Sundquist-liked the prospectus for short term sim, very concerned about land use change being included, it has been deferred for too long and doesn’t appear to be an issue addressed by the lmwg, . Rienencker: the idea of one person to do ocean assimilation is not enough, messagefrom cab is that short term sim is very supported. Gent: asked jay about priority of short term sim. Fein: the next gen model to exercise in next ipcc, we realize the results will be 7 years out and interests will have changed, believes the interst in this will be much more usable and valuable depending on the quality of the procuct. A good way to energize the project. Rood: business infrastructure investment are on the timescale of 40-50 years and many will be affected by sea level change. Crowley: get out beyond 20 years because its within a lifetime and some stuff will be happening. Can you produce results that you label as experimental. It may be very difficult in terms of manpower, and for the summer CAB meeting present a preliminary contingency and see whether you can realistically progress and people may realize that we’r ejsut spread too thin and may nto be able to do it. Wwe know it can do doen in theory, but need put out an exercise to see if we can pull it off. Held: The real decision is how much effort do you put in generating high res with stable enough coupled model. Are your resourced enoughto dev a stable coupled thingy. Denning: thinks it pretty dangerous to say that these are going to be use3ful forecasts. Approach it as science. Jay: said early these are useful if they’re correct. Rind: how do you know if the projectin for thenext 20 years is valid? Rood: a great step forward, software plan is missing, what are yougoing to do with esmf. Jay: issue of disservice basedon products usefulness. From eric sundquists original question--What should we bediong for next ipcc? What if we add something that totally fouls up the model and we can’t move forward. To the extent possible we should define an absol min of what we want ot come up with for ipcc and they prove to be reasonable, then we can determine thenext steps on short term sim. Gent: final comment, if the ipcc deadine were pushed back a year or two all the modeling centers could do a much better job. Rind thinks deadline will be forced by observations like Greenland melting, whether we know the info is good or not is a red herring, this is one of those best sources of info, having the best models making predictions is the best source of information.
Tom Crowley “Ice age earth”
Used CCSM 3. Fundamentally different pattern over islands above Australia in ice age earth. Jet maximum located over the ocean, not land. (cyclogenesis is reduced?) Some parts of Alaska were ice free because a huge amount of heat transfer was steered there by winds. Less upwelling along upper west coast of US. During ice age two areas where currents are rising up as opposed to one now.
Movie: Wants to make this availabve as asn educational tool. Bader how do you know the anomalies are correct? Crowley: there’s a lot of data on land that can validate model through observations.
Isaac Held GFDDL update
Collaborations between GFDL and CCSM:
Ongoing: 2 climate process teams
Atmospheric-May meeting at GFDL
Both models contributing ½ degree time slice simulation to NARCCAP
Other possible collaborations: FV dynamical core: cubed sphere
Dave Rind “GISS Update
GISS model development, an alternate reality compared to CCSM. Not talking about AR5 needs as driving factor for model development. A lot of effort is being made to get a realistic soil something. Coccolithophores. :^) Trying to change vertical to be more aida like. Have tried to have model transports in place to prevent problems with having designed a model a certain way, then having people use it in an unintended way. If you don’t look at these things ahead of time people will have ot use it as it is. Gent: how far in front of aR5 will you plan for it? Rind: it’s useful to see how your model compares to others. It does not dominate our discussions. Fein: is there any advantage to a coomon set of forcings across models? Rind: we’ve run with a range of solar forcing. For volcanic forcings, they’re dwarfed by the past.
Michele Rienecker GSFC Update
WE have our something model running and is going thorugh validation. Happy with performance. Couple to stratispheric chemistry. Just started running aerosol sims. Interactins with NCAR. Max and Phil Rasch have beentrying to maintain joint FV core. Would like to ssee a national organized repositiory. Work,ing with NCAR on …missed it.
Steve Lord NCEP
Science Results first 12 slides
Multimodel ensembles
CFS reanalysis
SEC is systematic error correction. CFS (which appears to be their model) is currently frozen.
Jim Kinter “Season Prediction with te CCSM” at COLA
Main goals:
- IPCC class model
- multi model prediction
- using initial conditions from mom3 in pop
- global atm and land surface
Gent have you done any experiments starting with MOM. Kinter everything shown was. You get a big initial shock. Kinter this choice of initial states is worst case scenario for cfs. See westward conflagration in ccsm.
Executive Session
Michele writes summary.