Preliminary summary of CPT breakout meeting, Santa Fe, 7 July 2004 Chris Bretherton A productive meeting of many of the CPT members discussed progress since last fall, and future CPT plans/strategies. The meeting was divided into four talks (summarized below) followed by discussion (which led to the following action items): ---------------------------------------------------------------------- CPT action items: 1. Schedule Fall meeting - 21-22 Oct in Seattle or next week thereafter was best for attending CPT participants. Will either of these dates work for most others too? Chris B. will try to nail this down by end of July. 2. RCE intercomparison - we agreed to modify the RCE intercomparison into a Walker circulation forced by differential cooling of a slab ocean, specifying 1x and 2X CO2 runs. Both 2D and 'bowling-alley' domain (long and narrow) 3D CRMs can participate, plus SCM's configured to be dynamically connected in a singly or doubly periodic way. Chris B showed results from an example 2D CRM run with this configuration, and will discuss this proposal with GCSS WG4. He will try to get candidate intercomparison specs and output datasets on the CPT site by the end of July. 3. CPT/AMWG collaboration on subgrid parameterization of cloud and radiation processes. General agreement that an approach based on assumed horizontal subgrid pdf's of water/condensate is most promising, per email discussion of a few weeks ago. Discussion continued from Spring AMWG meeting. NCAR CPT liaison Cecile Hannay will implement Pincus/Klein 'column generator' into CAM to stochastically sample pdf's and vertical overlap assumptions to provide inputs to radiation and cloud physics routines. NCAR ASP postdoc Ben Johnson will take the lead in designing a pdf scheme for total water, in collaboration with CPT members, esp. Steve Klein, Robert Pincus, and Phil Rasch. We'll need to talk to Bill Collins about using this machinery to improve the cloud specification/overlap for radiation in CAM. This will be an ongoing activity for the next couple of years, but a goal is at least to get the column generator implemented in CAM by the October meeting. Ben will be invited to the October CPT meeting to lead a discussion on design of a pdf approach. 4. Rigorous use of GCSS WG1/4 cases. Several past and ongoing GCSS cases are available for single-column testing, esp. BOMEX, ARM and DYCOMS shallow cumulus/stratocu cases, and ARM., Amazonia, and COARE deep convection cases. We should make an effort to run all these cases with all three participating SCMs, since observational and CRM comparisons exist for all these cases and they have proved illuminating for other SCM groups worldwide. Caralyn Lappen will be asked to help with SCAM simulations of the deep convection cases, and Ming and Julio will do the SCM simulations with the GFDL and GMAO models. Chris will coordinate getting links to input datasets onto the CPT web page for these cases. It would be nice if we could get this done by our October meeting. 5. Use of GCM single-column output data. The GFDL model has a surprising amount of condensate at 700-800 mb in subsidence regions, while the NCAR model has lots of condensate within 50 mb of the surface. Is this due to intermittent events or persistent clouds at these levels? Ming and Steve Klein will look at the high time-res single-column GFDL output and hopefully Cecile can make comparable single-column output from the CAM to try to better understand this. Ming can report on his findings in October. In addition, we need high time res single-column advective/boundary forcing data from a control simulation at some subset of all the designated single-column points. Maybe Ming and Cecile can work on getting these extracted from their models too. Julio can do the same with GMAO if he has the software capability. This might provide a framework for an SCM intercomparison of boundary layer cloud processes and their climate sensitivity in the three models. 6. CPT PI web sites. I'd like to request all CPT core PIs to put up a web site that I can link to our internal CPT web page to help communicate your CPT-sponsored and other relevant activities to the group, by the time of our October meeting. ---------------------------------------------------------------------- Talk summaries Chris Bretherton's talked about his work with Matt Wyant on Bony diagnosis (binning via monthly mean 500 mb omega) of the CPT models. The models show similar patterns of LW/SW cloud forcing, but they arrive at this via very different vertical condensate/cloud fraction profiles in both convective (om500 < 0) and BL cloud (om500 > 0) regimes. ISCCP simulator results from GFDL show considerable ifferences from actual om500-binned ISCCP observations; NCAR too (though this was not debugged at the meeting time), so we should be able to learn from this. Chris mentioned Minghua's slides in which Minghua argued that omega-binning at hourly resolution would improve the connection of omega to cloud. However, the number of datasets in which an hourly omega can reliably by determined is rather small. Chris also discussed the proposed radiative-convective equilibrium (RCE) intercomparison. He presented 64x64 km domain CRM simulations of the case specified on the CPT web site, which look nice, but pointed out that results differ profoundly on a larger (576x576 km) domain due to convective self-aggregation, which ultimately focusses all the convection into a single small patch surrounded by very dry air elsewhere. Isaac Held talked about RCE simulations using various idealizations of the GFDL GCM, including aquaplanet nonrotating uniform SST simulations at GCM horizontal resolution on a sphere. He pointed out that a single column model can behave rather differently than a dynamically coupled horizontally homogeneous array of SCMs due to dynamical/diabatic feedback instabilities like self-aggregation, and suggested that an dynamically coupled SCM array is perhaps a better analogue to the GCM behavior. He also showed impacts of turning off deep convection parameterization, etc. and stressed the need to fundamentally better understand tropical circulation/deep convection/cloud/radiation feedbacks for both parameterization and understanding cloud feedbacks. Marat Khairoutdinov talked about a large CRM simulations of the 52-day TRMM KWAJEX case, and an impressively high resolution (dx = dy = 100 m, with comparable vertical resolution, Lx = Ly = O(50 km)) simulation of the GCSS WG4 Amazonia diurnal cycle case. Visualizations of both cases make clouds look pretty realistic. Marat has implemented Robert Pincus's ISCCP simulator for use with his CRM model output. Model output datasets are large, but CPT members can email Marat to look at them. Brian Mapes presented a preliminary version of his talk the next day for the CCSM Atmospheric Model Working Group (AMWG), describing the CPT work to date and upcoming plans. We discussed and refined his excellent summary presentation. Attendees (I may have missed some) Tom Ackerman Leo Donner Jim Hack Cecile Hannay Isaac Held Marat Khairoutdinov Jeff Kiehl Steve Klein Brian Mapes David Mansbach Robert Pincus Dave Randall Dave Raymond Phil Rasch