CSL Allocations 2009

The following projects had CSL allocations from June 2009 to March 2011.

Community Climate System Model (CCSM) – IPCC

Project Lead: Peter Gent, NCAR
Allocated: 8,100,000 GAUs
Sponsors: DOE, NSF

The fifth assessment report (AR5) of the Intergovernmental Panel on Climate Change (IPCC) is scheduled to be published in early 2013. This report can only assess papers that are either published in reputable journals, or accepted for publication. So, working backwards from this timeline, the CCSM project believes it has to finish the complete suite of AR5 runs by the end of December, 2010. Thus, the period of this CSL allocation, June 2009 through November 2010, almost exactly coincides with the period when the AR5 runs have to be accomplished. The CCSM project has committed to complete two different sets of runs for the AR5, neither of which it has done before. The first will include an interactive carbon cycle, and will consist of longer runs out to 2100, with one or two runs extended to 2300, see Figure 1. These runs will be done with a somewhat modest resolution version of the model: 2° in the atmosphere and land components and 1° in the ocean and sea ice components. The CCSM project aims to complete this set of runs on its CSL allocation at NCAR. The second set of runs will be shorter, decadal forecasts using a higher resolution version of the model: 0.5° in the atmosphere and land components and 1° in the ocean and sea ice components, see Figure 2. Several to many of these runs will involve interactive chemistry, and some will use the Whole Atmosphere version of the CCSM. These runs will require a much larger amount of computer time, and the CCSM project aims to use allocations on Department of Energy computers in order to complete this set of AR5 runs.

Community Climate System Model (CCSM) – Science

Project Lead: Peter Gent, NCAR
Allocated: 8,100,000 GAUs
Sponsors: DOE, NSF

The CCSM project is in the middle of assembling the next version of the coupled model, CCSM4. The new versions of the atmosphere, ocean, land, and sea ice have been delivered, and the first coupled runs are underway. In order for the CCSM4 to be defined, there needs to be both an 1870 control run that has an acceptable top of the atmosphere balance near zero, and a 20th Century run that reproduces well the observed climate over the past century. This will require tuning of all the four components, and especially the new feature of CCSM4, which includes for the first time the indirect effects of aerosols. Once this is completed, then the additional components of the CCSM4 will also have to be defined. These are the carbon cycle component, chemistry component, and the Whole Atmosphere component, all of which up to now have been based on the intermediate version of the model, CCSM3.5.

Once the CCSM4 is finalized, then it will immediately start to be used to make the runs that are required for submission to the IPCC AR5. The CCSM project is planning to do a complete set of lower resolution carbon cycle runs on CSL resources, and a companion proposal to this one has also been submitted. However, there are a myriad of important science questions that require separate runs from the IPCC set. The science questions and required runs from the various CCSM Working Groups are contained in this proposal. Note that the Whole Atmosphere WG is now a part of the CCSM, and so its proposal is included here, and is not a separate request, as it was in the last round of CSL requests.

Thus, the focus of the CCSM project will switch somewhat from model development to model use from June 2009 for the next 18 months. However, a number of model developments will proceed, especially in the atmosphere component. One of these is the requirement to use a different horizontal grid that is not based on latitude/longitude, which is required so that the component parallelizes over many thousands of processors. Thus, this proposal also requests computational resources for model development over the 18 months from June 2009. CSL computer resources are the lifeblood of the CCSM project, are actively managed based on the priorities set by the Science Steering Committee, and are the glue that keeps the CCSM functioning as a community project.

Seasonal to Interannual Predictability in a Changing Climate

Project Lead: Benjamin Cash, Center for Ocean-Land-Atmosphere Studies (COLA)
Allocated 2,700,000 GAUs
Sponsors: NSF, NOAA 

The predictability of climate variations on seasonal to interannual (S-I) time scales may vary in response to decadal variability in the physical climate system. Decadal variability arises from both the intrinsic dynamics of the climate system and changes in the external forcing. This naturally gives rise to the question of whether or not the variability and predictability of the dominant features of S-I variability, among them El Niño and the Southern Oscillation (ENSO), the Madden-Julian Oscillation (MJO) and the Indian monsoon, change on decadal time-scales, and if those changes are themselves predictable. 

Building upon COLA’s long-standing expertise in the investigation of S-I variability and predictability, we propose to explore the decadal predictability and variability of S-I phenomena through a series of numerical experiments using national climate models; namely the National Centers for Environmental Prediction (NCEP) Climate Forecast System (CFS), the National Center for Atmospheric Research (NCAR) Community Climate System Model (CCSM 3.0 and 3.5) and the National Oceanic and Atmospheric Administration (NOAA) Geophysical Fluid Dynamics Laboratory (GFDL) Coupled Model (CM 2.1). Particular emphasis is placed on the NCEP CFS, as this model is used operationally to produce seasonal forecasts. As a result, the CFS has been extensively tested in forecast mode against observational data, including twice-daily real-time forecasts routinely made by NCEP, and there is greater potential with CFS for research insights to lead quickly to operational improvements.

The proposed experiments seek to address two closely related sets of questions:
(1.) How is S-I variability affected by decadal variability in the climate system? How do selected model biases impact the simulated decadal and S-I variability? (2.) How does S-I predictability vary by decade? How do differences in initialization schemes and model components impact these variations?

Ensemble Data Assimilation for Climate Model Development

Project Lead: Jeffrey Anderson, NCAR
Allocated 1,770,000 GAUs
Sponsors: NSF, NASA 

Data assimilation (DA) was originally developed to generate initial conditions for numerical weather prediction (NWP) models. Modern DA algorithms can do much more than that. By combining observed quantities with short-term forecasts from models, DA increases information about the model, the observing system, and the physical system. DA is used to produce reanalysis products that provide the best estimate of the state of the climate system. DA can be used to diagnose model deficiencies and to quantitatively compare the capabilities of different models. Sensitivity analysis that facilitates determining the relationship between model variables at different points in space and time can be performed with appropriate DA tools. DART is a community ensemble data assimilation facility developed by the Data Assimilation Research Section (DAReS) at NCAR. DART is in use with a number of large geophysical models including CAM and CAM/CHEM. Co-I Robert Pincus has incorporated the GFDL AM2 model into DART in collaboration with DAReS. The Weather Research and Forecast Model (WRF), traditionally used for mesoscale prediction, is also being used with DART by a number of researchers.

Development and Application of Seasonal Climate Predictions

Project Lead: Stephen E. Zebiak, Columbia University
Allocated: 1,215,000 GAUs
Sponsors: NOAA

The goal of this research is to improve the prediction of seasonal climate variations, such as rainfall and near­surface air temperature, for application in climate risk management problems. Seasonal climate predictability stems from the components of the climate system with slower time scales, particularly the upper ocean and its interaction with the tropical atmosphere. Due to the atmosphere’s sensitivity to initial condition “chaos”, seasonal forecasts of precipitation and near­ surface air temperature must be probabilistic if they are to be useful in applied areas such as water resource management or the forecasting of outbreaks of diseases such as malaria or dengue. The proposed experiments therefore will use relatively large Monte Carlo (or ensemble) simulations to develop probabilistic seasonal forecast systems.

Multi-RCM Ensemble Downscaling of NCEP CFS Seasonal Forecasts (MRED)

Project Lead: Raymond W. Arritt, Iowa State University
Allocated: 900,000 GAUs
Sponsors: NOAA

Over the past two decades regional climate models (RCMs) have become widely used to downscale global climate simulations. In contrast the ability of RCMs to downscale seasonal forecasts has received little attention. In this project we address the question, Do RCMs provide additional useful information for seasonal forecasts? through a systematic test of the RCM downscaling methodology. Specifically, RCMs will be used to downscale 25 years of National Centers for Environmental Prediction (NCEP) Climate Forecast System (CFS; Saha et al. 2006) wintertime forecasts that will be re-made as part of this project. A forecast ensemble will be constructed using multiple RCMs, each of which will produce 15-member ensembles for each winter season (1 December – 30 April) over a period of 25 years (1982 to 2006) using initial and lateral boundary conditions from the CFS forecasts. An initial focus on the winter season allows evaluation of topographic forcing, snowmelt, and the potential usefulness of higher resolution, especially for near-surface fields influenced by orography. Each RCM will cover the conterminous US (CONUS) at approximately 32 km discretization, comparable to the North American Regional Reanalysis (NARR) against which the RCM results will be evaluated. We will evaluate individual RCM and CFS forecasts as well as ensemble forecasts and metrics of ensemble spread. More extensive analysis will be performed to link improvements in downscaled forecast skill to regional forcings and physical mechanisms.

Incorporation of the Hybrid Coordinate Ocean Model (HYCOM) into the Community Climate Systems Model (CCSM): Evaluation and Climate Applications

Project Lead: Jianjun Yin, Florida State University
Allocated: 900,000 GAUs
Sponsors: DOE

The HYbrid Coordinate Ocean Model (HYCOM) is an ocean model that is widely used in the ocean modeling community. However, its performance in climate models has not yet been systematically evaluated. In HYCOM, the vertical coordinate is isopycnal in the open, stratified ocean, and smoothly reverts to z-coordinate in the mixed layers and unstratified ocean, and to a terrain-following coordinate in shallow coastal regions. HYCOM combines the strength of an isopycnal model in simulating the ocean interior, a depth coordinate model to simulate the upper ocean processes and a terrain-following model to simulate the coastal regions. HYCOM is perhaps the most widely used model that is primarily based on isopycnic coordinates and it is the backbone of the next- generation U.S. NAVY ocean prediction model. But up to now, there has not been any systematic comparison study that would quantify the impact of using an ocean model that is primarily based on isopycnic coordinates versus one that is based on depth coordinates.

Our scientific proposal entitled “Incorporation of the HYbrid Coordinate Ocean Model (HYCOM) into the Community Climate System Model (CCSM): Evaluation and Climate Applications” has been funded (starting from August 2007) by the Scientific Discovery through Advanced Computing (SciDAC) program of the US Department of Energy (DOE). The primary goals of our DOE proposal are to: (a) assess the impact of the isopycnic versus depth coordinates in ocean and climate system modeling and (b) use the coupled CCSM3/HYCOM as a research tool for studies on past, present and future climates.