Joint CGD/EOL Seminar

Bringing climate models and observations together using a weather forecast approach: Scenes from the tropical Pacific

Brian Medeiros NCAR/CGD

 This presentation will focus on results using the Cloud-Associated Parameterization Testbed (CAPT) to evaluate and understand aspects of the Community Atmosphere Model. CAPT is a simplified "weather forecast" mode in which the model is initialized with a realistic atmospheric state and integrated for a few days to produce a forecast. The method is idealized in the sense that the realistic initial state is (typically) an operational analysis, so it does not incorporate a native data assimilation step. The model's large-scale circulation remains somewhat close to the observed state for some time (days), while the fast processes (i.e., parameterized physics) develop errors quickly and start evolving the state away from the observed state and toward the model's preferred climate. Interrogating the processes involved in these fast errors helps to identify sources of error in climate simulations and informs the model development effort. An advantage of the CAPT framework is that it allows for direct comparison with observations, including field campaign data. We will review the relevant aspects of the methodology, and then investigate results focusing on clouds and convection over the Pacific Ocean. The first results will show how observations can be compared with CAPT forecasts to evaluate the model and identify biases. Those biases can then be understood by detailed analysis of the model output. A later example takes a different tack. We will see that the model captures important features of the observed diurnal cycle over the open ocean. Some aspects of the observed diurnal cycle are not well understood, and here we use CAM in the CAPT framework to run experiments to try to understand the mechanisms of diurnal variation of winds and convection over the central Pacific. These examples are intended to show how observations can help us understand and improve the model, and then how the model might be used to shed light on observed phenomena.

Tuesday, 20 May 2014

3:30 p.m.

NCAR-Foothills Lab

3450 Mitchell Lane

FL2-1022, Large Auditorium

 Seminars are live webcast: http://www.fin.ucar.edu/it/mms/fl-live.htm

* Refreshments are served before seminar.  *

About you
First name: 
Meghan
Last name: 
Stell
Phone extension: 
2043
Announcement contact
First name: 
Gaylynn
Last name: 
Potemkin
Phone extension: 
1618
Announcement duration
Start: 
Monday, May 12, 2014
End: 
Tuesday, May 20, 2014