[an error occurred while processing this directive] [an error occurred while processing this directive]
Report of the CSM Natural Variability Working Group Meeting
Fourth Annual Climate System Model Workshop
by R. Saravanan and Edwin K. Schneider, Co-chairs
June 24, 1999
The first part of the meeting was devoted to the following presentations:
David Battisti (University of Washington) described results from CCM3 integrations using mixed layer ocean and prescribed land surface properties to study tropical Atlantic variability. An interesting feature was the sensitivity of the variability to land surface processes in the Amazon basin.
Ping Chang (Texas A&M University) presented an analysis of three ensembles of CCM3 integrations forced by observed SST variability focusing on the tropical Atlantic. The results suggested that there was the potential for positive air-sea feedback in the deep tropics in the western part of the Atlantic basin.
R. Saravanan (NCAR) discussed another experiment using CCM3 coupled to a mixed layer ocean, also focusing on the tropical Atlantic, and showing that the spatial variations in the mixed layer depth played an important role.
Gudrun Magnusdottir (University of California, Irvine) presented results from a series of CCM3 integrations forced with large sea surface temperature (SST) and ice anomalies in the North Atlantic. The response to ice forcing was significantly stronger, and there was very large variability in the response patterns for different calendar months.
Phil Duffy (Lawrence Livermore National Laboratory) presented results from the Coupled Model Intercomparison Project (CMIP), focusing on the variability of surface air temperatures over ocean and over land. Of note was the considerable discrepancy between model-simulated variability and the observed variability over land, pointing to possible deficiencies in land surface parameterizations.
Michael Ghil (UCLA) presented a SSA analysis of "observed" SST variability looking at the persistence of SST anomalies. This analysis picked out a decadal mode of variability in the Pacific basin, quite similar to that identified in more traditional analyses of SST time series that exclude the El Nino-Southern Oscillation (ENSO) variability.
The second half of the meeting was devoted to discussion of the following topics:
1. CHANGE OF NAME: After a brief discussion, it was agreed that the name of the working group should be changed from "Natural Variability Working Group" back to "Decadal to Centennial Variability Working Group" to better reflect the distinction between this group and the Seasonal-to-Interannual variability Working Group (SIWG).
2. DIAGNOSTIC TOOLS FOR CSM: There were at least two informal proposals from earlier meetings, from Sumant Nigam (University of Maryland) and David Neelin (UCLA), to use simplified models to diagnose CSM behavior and help improve it. Nigam had proposed using a linearized primitive equation model and Neelin had proposed using a simplified tropical model. The consensus opinion was that any simplified modeling effort that would like to be "officially" associated with CSM must satisfy the following criteria:
3. MODEL DEFICIENCIES (other than those identified by SIWG): It was pointed out that coupled tropical variability in the CSM is poor even in the Atlantic and the Indian Oceans, not just in the Pacific. The presentation by Magnusdottir identified clear deficiencies in CCM3's simulation of the North Atlantic stormtrack.
4. ACCESS TO MODEL DATA: It was pointed out that data from long model integrations is often spread over a very large number of files on the Mass Storage System. For example, a single file often just contains a month's worth of data, which necessitates a large number of file reads to process a 300-year run. It was recommended that a decade or so of data be combined into a single file, which would speed up data access completely.
5. LONG MODEL INTEGRATIONS: There was a clear need for obtaining computational resources to carry out long model integrations that would be useful to the community. A prioritized list of suggested integrations is attached below. The need for doing some higher resolution integrations (T85 or T106) to test the sensitivity to resolution was noted. It was also pointed out that an ensemble size of 5, which may be adequate for identifying the gross features of the ENSO signal, was too small for identifying weaker non-ENSO signals. The suggestion was made that the typical ensemble size be increased to at least 10.
6. EFFECT OF LAND SURFACE PROCESSES IN ATMOSPHERIC VARIABILITY: There was considerable interest in carrying out CCM3 integrations with prescribed, rather than interactive, land surface properties, such as soil moisture. It was felt that future versions of CCM should incorporate this option.
PRIORITIZED LIST OF LONG MODEL INTEGRATIONS:
Due to problems with the coupled climatology of CSM, we feel that progress on the scientific issues associated with decadal-to-centennial climate variability can best be made at this time by long integrations of the atmosphere forced by specified SST or coupled to a mixed layer ocean. This belief was confirmed by the interesting results given in the presentations listed above.
I. Tropical Pacific SST-Mixed Layer integration: 200 years (5,000 C90 hrs)
CCM3 forced by observed SST in tropical Pacific and coupled to a slab ocean model elsewhere, with spatially and seasonally varying slab depth. The purpose of this integration would be to capture the effect of oceanic mixed layer feedbacks on atmospheric variability, while still representing the strong air-sea coupling in the tropical Pacific. In particular, the temporal persistence of midlatitude and tropical Atlantic atmospheric variability would be better represented in such a configuration.
II. "Perfect model" AMIP-type integration: 100 year (2,500 C90 hrs)
Use SST from a coupled CSM integration to force an atmosphere-only CCM3 integration. Such an integration would help answer the question of whether AMIP-type atmosphere general circulation model (AGCM) integrations using specified SST can reproduce the variability in a coupled model (in a "perfect" model context).
III. High resolution (T85) AMIP2 ensemble: 5x20 years (15,000 C90 hrs approx.)
Higher resolution ensemble CCM3 integrations forced globally by observed SST. This suite of integrations would allow us to study how the variability simulated by CCM3 is affected by resolution. Of particular interest are the extratropical response to ENSO and local air-sea interaction in the tropical Atlantic.
IV. 100-year CCM4 integrations using GISST4 dataset: 5x100 years (12,000 C90 hrs)
These integrations can be used for intercomparisons with other modeling groups, which are planning a similar set of integrations (Shukla).
V. Extended climatological SST integration: 800 years (20,000 C90 hrs)
Extend the climatological annual cycle integration of CCM3 from 200 years to 1000 years. Such a long integration would provide a sufficiently large dataset for a careful analysis of the statistical properties of internal atmospheric variability.
List of Participants:Michael Alexander, University of Colorado Caspar Ammann, University of Massachusetts David Battisti, University of Washington Jason Bell, Lawrence Livermore National Laboratory (LLNL) Uma Bhatt, International Arctic Research Center Cecilia Bitz, University of Washington Maurice Blackmon, NCAR Rainer Bleck, Los Alamos National Laboratory Byron Boville, NCAR James Boyle, LLNL Esther Brady, NCAR Grant Branstator, NCAR Christopher Bretherton, University of Washington Kenneth Caldeira, LLNL Maria-Antonietta Capotondi, NOAA-CIRES Yi Chao, Jet Propulsion Lab Wei Cheng, University of Miami Wooyoung Choi, Los Alamos National Laboratory Yongjiu Dai, University of Arizona Aiguo Dai, NCAR Gokhan Danabasoglu, NCAR Charlotte DeMott, Colorado State University Scott Doney, NCAR C. Mark Eakin, NOAA Peter Eltgroth, LLNL Benjamin Felzer, UCAR Robert Gallimore, University of Wisconsin-Madison Avijit Gangopadhyay, University of Massachusetts-Dartmouth Michael Ghil, UCLA Bala Govindasamy, LLNL Stephen Griffies, NOAA/Geophysical Fluid Dynamics Laboratory (GFDL) James Hack, NCAR Andrea Hahmann, University of Arizona Charles Hakkarinen, Electric Power Research Institute Cecile Hannay, IARC, Frontier University of Alaska Hiromaru Hirakuchi, Central Research Institute of Electric Power Industry Marika Holland, NCAR William Holland, NCAR Qi Hu, University of Nebraska-Lincoln Zhen Huang, Los Alamos National Laboratory (LANL) Chih-Yue Kao, LANL Akira Kasahara, NCAR Marat Khairoutdinov, Colorado State University Kenneth Kunkel, Illinois State Water Survey Chung-Chieng Lai, LANL Shian-Jiann Lin, NASA/Goddard Space Flight Center JoAnn Lysne, NCAR Gudrun Magnusdottir, University of California Eric Maloney, University of Washington Julie McClean, Naval Postgraduate School Gerald Meehl, NCAR Richard Moritz, University of Washington Sumant Nigam, University of Maryland Joel Norris, NCAR Bette Otto-Bliesner, NCAR David Pierce, Scripps Institution of Oceanography Andrey Proshutinsky, University of Alaska-Fairbanks Philip Rasch, NCAR Steven Running, University of Montana Edward Sarachik, University of Washington Ramalingam Saravanan, NCAR Adam Schlosser, Center for Ocean-Land-Atmosphere Studies (COLA) Edwin Schneider, COLA Albert Semtner, Naval Postgraduate School Michael Spall, Woods Hole Oceanographic Institution Shan Sun, NASA/Goddard Institute for Space Studies Kevin Trenberth, NCAR Joe Tribbia, NCAR Anastasios Tsonis, University of Wisconsin-Milwaukee David Williamson, NCAR Zong-LiangYang, University of Arizona Stephen Zebiak, IRI Jinlun Zhang, University of Washington