CISL Annual Report 2005 > CISL research accomplishments

Data Assimilation Research accomplishments

Data Assimilation (DA) is the set of statistical and computational methods for combining observational data with numerical models. Although its most visible application is in operational weather forecasting, DA can also be used to validate and understand the features of geophysical models and interpolate among observations that are irregularly observed over space and time. The centerpiece of the Data Assimilation Research Section (DAReS) is a software environment, the Data Assimilation Research Testbed (DART), that makes it easy to investigate the use of DA with a specific numerical model. The use of ensemble Kalman filter methods have made it possible for DART to support a variety of large and sophisticated numerical models including NCAR's weather forecast model (WRF) and the atmospheric component of NCAR's climate model (CAM).

At the other extreme, DART also supports many simple dynamical models that are excellent for teaching DA, and the DART tutorial is a lively collection of examples integrated with DART's capabilites.

Coupled with a major new DART release (Iceland version), a research highlight is the quantification of the forecast skill of CAM. The synthesis of CAM with observational data using DART produces estimates of the atmosphere that are comparable in accuracy to reference analyses. Just as important, incorporating CAM into a DA framework makes it easier to understand how the dynamics of the model differ from observations, and a pilot study this year has been able to understand how variation in the gravity wave drag parameter affects the performance of CAM.

Some other research results in DAReS are the amount of information afforded by GPS observations of water vapor, the use of radar-based observations for predicting severe weather, and the application of ensemble methods to a model of the mesosphere (ROSE).