Discussions on NOAA/CTB Science Priority and Transition Plan

Recently, NOAA/CTB/Climate Science Team conducted a review on the NOAA/CTB Science Priority and Transition Plan.  Following is a discussion by Dr. Martin Hoerling of NOAA/Climate Diagnostics Center. 

1.  Establish a "process" to accelerate the transition of R&D to operations.

What sets the Climate Test Bed (CTB) apart from existing OGP-funded research through current programs and the core mission of NCEP operations is the establishment of a "process" for moving the research into operations.  There have been some efforts of which the most recent one is the "week-2" re-forecasting approach developed in research mode by CDC, and now implemented at EMC.   What are the lessons to be drawn from these ad hoc experiences?  How can the CTB build upon those, and streamline a process for transition of research to operations? This seems to be a key matter in meeting the mission goal. To accelerate this, we should go hand-in-hand with setting up such a process. For example, where does the operational side pick up responsibility, and at what juncture does a "hand-off" occur?  There will be a need for resources to both research and operations to ensure a smooth transition.  The defining of responsibilities on research and operations sides will be important in measuring performance and accountability for a smooth, more rapid transfer of suitable technology.  In addition, the process should also include the need to define what constitutes "suitable (or new and improved) technology".

2.  Paint a clearer picture of the expected skill of climate predictions, based on existing theory.

The notion of predictions of the "first" and "second" kind has been recognized for decades, and the skill prospects for such systems widely studied. Climate predictions using dynamical models have been occurring for about a decade now (see the archives of The Experimental Long-lead Forecast Bulletin).  An even longer record of simulation studies with such dynamical models exists.  Progress in dynamical climate modeling and predictions was in fact "fast" with the technological hurdles of running such models on sufficiently fast machines being a main early bottleneck.

What has not progressed, however, is the skill.  Skill of dynamical systems continues to be generally comparable to sophisticated statistical forecasts.  A substantiative question is "why has progress been slow" in improving predictions on 2 week to12 month time scales?  The theory of climate predictablity is quite clear that a leading reason lies not in the lack of access to a coupled atmosphere-ocean-land surface forecast systems, a recent development at NCEP.  Empirical confirmation to that effect can be found in the DEMETER experience.

What is the nature of the recent coupled system climate prediction progress as concerns S/I predictions?  Without questioning the scientific merit of implementing fully coupled dynamical models for operational climate predictions, it is an open question as to how much climate forecast skill can be improved beyond existing methods.

It needs to articulate the "need" that is causing NOAA to act and develop a Climate Test Bed.  Painting a clearer picture of the expected skill of climate predictions based on existing theory could make the case that even these theoretical levels are probably not currently being realized because of existing practices, which include linear empirical methods, 2-tier dynamical methods, and single model coupled system; it would be useful to better define the term "improvements", for example, objective probabilities, calibration, specific new products and scales of application including soil moisture and drought prediction.

(Contact Martin Hoerling)