IRI real-time seasonal climate forecasts and models

As of April 2017, the IRI has changed its forecasting methodology, described here: https://iri.columbia.edu/our-expertise/climate/forecasts/seasonal-climate-forecasts/methodology/

The IRI has been issuing seasonal climate forecasts of precipitation and near-surface temperature for the globe since 1997, and on a monthly basis since 2001. IRI’s seasonal forecasts, prior to April 2017, were based on a two-tiered dynamically based multimodel prediction system. A brief description follows, with more details given in Barnston et al. (2010).

A set of 2-tiered seasonal forecasts are produced every month at IRI using ECHAM4.5 and CCM3.6 Atmospheric General Circulation Models (AGCMs) forced by both persisted (PSST) and scenario SSTs (SSST), the SSST forecasts consisting of multi-model averaged SST “scenarios” designed to include a measure of the uncertainties in the SST forecast (see Barnston et al. (2010) for details). The forecast SST scenarios, which are applied only to the tropics (25N-25S), come from the SST forecasts of the CFSv2 model, the NOAA/CPC Constructed Analogue statistical model, and the LDEOv5 simple coupled model from Lamont Campus of Columbia University. Outside of the tropics, damped persistence forecasts are used. There are 3 subscenarios derived from the multi-model mean of the of the SST forecast scenario, designed to incorporate the uncertainty of the SST forecast. One subscenario is the multi-model mean forecast itself, and the other two subscenarios are the multi-model mean forecast plus and minus an uncertainty factor derived from the hindcast performance record, respectively. The persisted SST scenario persists the observed SST of the most recent 1 month. These two SST scenarios are used separately to force both ECHAM4.5 and CCM3.6 for which 24 members are generated based on initial conditions, taken from long AGCM runs driven with observed SSTs, for the date of the initialization.

Persisted and forecasted SSTs ensemble members are run out to 4 and 9 months, respectively, allowing to produce persisted SST forecasts for the upcoming season and forecasted SST forecasts for the four overlapping 3 months seasons to come. The ensemble members from both ECHAM4.5 and CCM3.6 are then combined together with other AGCM forecast ensembles to build the IRI Multi-Model Ensemble (MME) forecast system. The other ingredients of the MME are forecast and persisted SST-driven ensembles from GFDL and COLA2.26 AGCMs that are run by the respective collaborating institutions (as of September 2014), together with the 24-member CFSv2 coupled model ensemble from NCEP.

Additional sets of ECHAM4.5 retrospective/real-time forecasts are also available, forced with SST forecasts from a single model, namely (1) NCEP’s Constructed Analogue (CA) method (from January 1957), and (2) the NCEP CFS forecast (CFSv1 from 1982-2012, and CFSv2 from 2012-present).

Many of the forecast and hindcast datasets from the constituent models are available in the IRI Data Library, as described below. When using IRI forecasts or related data please cite both the published paper describing the dataset and the datasets themselves.

References

Barnston, A. G., S. Li, S. J. Mason, D. G. DeWitt, L. Goddard, and X. Gong, 2010: Verification of the First 11 Years of IRI’s Seasonal Climate Forecasts. J. Appl. Meteor. Climatol., 49, 493–520.

Van den Dool, H. M., 1994: Searching for analogues, how long must we wait? Tellus, 46A, 314–324.

AGCMs run at IRI

ECHAM4.5

http://iridl.ldeo.columbia.edu/SOURCES/.IRI/.FD/.ECHAM4p5/.Forecast/

The ECHAM4.5 atmospheric general circulation model (AGCM) was originally developed at the Max Plank Institut fur Meteorolgie (MPI) in Germany. From 1997 to 2002 an earlier version of the model ECHAM3.6, was applied to the IRI seasonal forecast system, however since 2001, its 4.5 version is used at T42 resolution with 19 vertical layers (Roeckner et al. 1996).

All the hindcast and real-time forecast data (24 ensemble for each PSST and SSST as well as for additional CASST) are now accessible in the IRI Data Library. When using the ECHAM4.5 data, please cite Roeckner et al. (1996), Li et al. (2008), and Barnston et al. (2010), and please acknowledge the IRI Data Library.

ECHAM4.5 Key References

Barnston, A. G., S. Li, S. J. Mason, D. G. DeWitt, L. Goddard, and X. Gong, 2010: Verification of the First 11 Years of IRI’s Seasonal Climate Forecasts. J. Appl. Meteor. Climatol., 49, 493–520.

Li, S., L. Goddard, and D. G. DeWitt, 2008: Predictive skill of AGCM seasonal climate forecasts subject to different SST prediction methodologies. J. Climate, 21, 2169-2186.

Roeckner, E., and Coauthors, 1996: The atmospheric general circulation model ECHAM4: Model description and simulation of present-day climate. Max-Planck-Institut fur Meteorologie Rep. 218, Hamburg, Germany, 90 pp.

CCM3.6

http://iridl.ldeo.columbia.edu/SOURCES/.IRI/.FD/.CCM3v6/.Forecast/

As a freely available community climate model (CCM), the CCM3.6 AGCM was developed at the National Center for Atmospheric Research (NCAR) in the United States. It hasthe same horizontal resolution (T42) and ensemble size (i.e., 24 for PSST and SSST, respectively) as the ECHAM4.5 model. Also, the initial atmospheric conditions are supplied by restart files from an integration in which CCM3.6 has been forced with observed SST for many years through the forecast start date. From 1997 to 2003 an earlier version of the model, CCM3.2, with 10 ensemble members only for PSST or SSST runs, was applied to the IRI forecast system. When using the CCM3.6 data, please cite Hack et al. (1998), Hurrell et al. (1998), and Kiehl et al. (1998), and please acknowledge the IRI Data Library.

CCM Key References

Hack, J. J., J. T. Kiehl, and J. W. Hurrell, 1998: The hydrologic and thermodynamic characteristics of the NCAR CCM3. J. Climate, 11, 1179–1206.

Hurrell, J. W., J. J. Hack, B. A. Boville, D. L. Williamson, and J. T. Kiehl, 1998: The dynamical simulation of the NCAR Community Climate Model version 3 (CCM3). J. Climate, 11, 1207–1236.

 Kiehl, J. T., J. J. Hack, G. B. Bonan, B. A. Boville, D. L. Williamson, and P. Rasch, 1998: The National Center for Atmospheric Research Community Climate Model. J. Climate, 11, 1131–1149.

AGCMs run by Collaborating Institutions

COLA

http://iridl.ldeo.columbia.edu/SOURCES/.IRI/.FD/.COLA/.T63/.Forecast/

The COLA v2.2.6 AGCM has horizontal resolution of T63, and 18 sigma levels in the vertical. Its land surface model is the simplified Simple Biosphere model (Xue et al., 1991). The deep convection scheme is the relaxed Arakawa-Schubert, and the shallow convection is the scheme of Tiedke. See Schneider (2002) for a more complete description.

COLA Key References

Schneider, E. K., 2002: The causes of differences between equatorial Pacific SST simulations of two coupled ocean-atmosphere general circulation models.  J. Climate, 15, 449-469.

Xue, Y., P. J. Sellers, J. L. Kinter III, and J. Shukla, 1991: A simplified biosphere model for global climate studies. J. Climate, 4, 345-364.

GFDL

http://iridl.ldeo.columbia.edu/SOURCES/.IRI/.FD/.GFDL/.AM2p12b/.Forecast/

The GFDL Tier-2 seasonal forecast model (GFDL, 2004; Delworth and Coauthors, 2006) for the IRI multi-model ensemble seasonal prediction system is the most recent AM2.1 version. This AGCM uses a finite-volume, latitude-longitude numerical core with a horizontal resolution of 2.5-lon X 2.0-lat grids and 24 hybrid vertical levels (top ~30km). Some of the key physical parameterizations include: Relaxed Arakawa-Schubert convection, full radiative transfer with aerosols, prognostic cloud scheme, planetary boundary layer scheme, orographic gravity wave drag and an interactive land model.

GFDL.AM2p14 Key References:

Delworth, T. L. and Coauthors, 2006: GFDL’s CM2 Global Coupled Climate Models. Part I: Formulation and Simulation Characteristics. J. Climate, 19, 643-674.

GFDL Global Atmospheric Model Development Team, 2004: The new GFDL global atmosphere and land model AM2/LM2: Evaluation with prescribed SST simulations. J. Climate, 17, 4641-4673.

GFDL.AM2p14 Key References:

Delworth, T. L. and Coauthors, 2006: GFDL’s CM2 Global Coupled Climate Models. Part I: Formulation and Simulation Characteristics. J. Climate, 19, 643-674.

GFDL Global Atmospheric Model Development Team, 2004: The new GFDL global atmosphere and land model AM2/LM2: Evaluation with prescribed SST simulations. J. Climate, 17, 4641-4673.