SR SSD 2001-17
Implementation of Operational Seasonal Climate Forecast Model
NCEP Climate Prediction Center
(Ed. Note: What follows is a slightly edited version of a recent Notice of Intent to Change issued by the NCEP Climate Prediction Center.The proposed change(s) are subject to approval by NCEP directors and CAFTI.)
Brief Description of Change.
NCEP plans to implement the first component of a new operational seasonal climate forecast suite. This will include a ~200km resolution (T62L28) climate atmospheric general circulation model and associated software, along with data sets for producing ensemble multi-seasonal climate forecasts, all in order to support CPC operational seasonal climate outlook products.
The multi-seasonal climate forecasts will have a one-month forecast cycle. Each month, a 20-member ensemble forecast run of about seven months in length will be produced, beginning on or about the 26th day of the previous month and completed on or about the fifth day of the current month. In addition, during each one-month forecast cycle, a 10-member ensemble of retrospective forecasts (hindcasts) for the past 21 years (1979-2001) will be produced. The hindcasts are initiated in the same calendar month of each year as the current forecasts. The set of hindcasts is needed to produce a climatology of the model for the given month when the forecasts are initiated, in order to derive anomalies of the model forecasts to be used for CPC product generation.
The hindcasts are to be produced during the approximately 20-day period of the previous month, between the completion of the previous month forecasts (on or about 6th of the previous month) and the initiation of the present month forecasts (on or about the 26th of the previous month).
Reason for Change.
NCEP has been running routine experimental ensemble climate forecasts using a climate atmospheric model since January 1995. The model forecasts are provided to CPC as numerical guidance to support their operational seasonal climate outlook products. To derive official seasonal climate outlook products, the numerical model guidance is weighted by its historical performance and combined with several additional statistical tools developed by CPC.
The model has been running in routine experimental fashion, that is, it has been run by EMC/CPC scientists on the research and development part of the NCEP central computer. The reason for the proposed change is to upgrade the model from the "routine experimental" status to "operational" status in order to ensure the reliability and timely production of the seasonal climate forecast guidance, and thus to better support the operational mission of CPC. The model will be implemented on the production side of the NCEP central computer which is augmented with additional 16 nodes for operational climate forecast suite.
Schedule for Change.
- Final Testing: Real-time parallel testing of the ensemble multi-seasonal climate forecasts at T62L28 resolution for three one-month forecasting cycles: July, August, September 2001.
- CAFTI Approval: To be requested in August 2001.
Official Implementation: On or about September 6, 2001 for the October 2001 forecast cycle.
Description of Testing.
The full resolution (T62L28) model was run in parallel with the lower resolution (T42L28) model for three one-month forecast cycles from February to April 2001. The full resolution model will be running in real time on the production side of the NCEP IBM for three one-month forecast cycles from July to September 2001, in parallel to the routine experimental forecast runs using the same model and at the same resolution.
Anticipated Impact on Forecasts.
This is the first component of a new climate forecast suite. The forecasts have been used by CPC on experimental basis. The present change is to implement the model into the NCO production suite to ensure reliable and on-time production of the forecasts to support the CPC operational seasonal climate outlook products.
Evaluation has been ongoing at CPC in the form of production of routine experimental model
guidance to support the CPC operational seasonal climate outlook products. The current higher
resolution model (~200km) ran in parallel with the lower resolution model (~300km) from
February through April 2001. Evaluation by CPC and EMC concluded that the higher resolution model
exhibited improvements over the lower resolution model, particularly in reduced mean bias in
temperature and height fields. Additional details can be found at
The T62 model has been running in routine experimental fashion since June 2001 (see
http://www.emc.ncep.noaa.gov/cmb/atm_forecast/index.html), and evaluation of value added by the ensemble dynamic model forecasts to the seasonal climate forecasts is ongoing at both EMC and CPC, and by our partners in the research community.
Additional components of the climate forecast suite will include an ocean data assimilation system and a coupled ocean-atmosphere model. These will be implemented at a later time.
Points of Contact for Further Information on the Change.
Ming Ji (301-763-8000 Ext 7564 Ming.Ji@noaa.gov)
Henry Juang (301-763-8000 Ext. 7517 firstname.lastname@example.org)
Jae Schemm (301-763-8000 Ext. 7565 Jae.Schemm@noaa.gov)