THE DATA ASSIMILATION RESEARCH TESTBED A Community Facility

The DART community ensemble data assimilation facility provides students, educators, and scientists with unprecedented access to free, state-of-the-art assimilation tools. DART's comprehensive tutorial, low-order models, and examples can introduce students to ensemble data assimilation on their laptops. The same tools can produce analyses using 10-million-variable climate system models, novel remote sensing observations, and the newest supercomputers. This enables students to advance quickly from basic understanding to meaningful research projects. DART can also accelerate scientific progress by modelers and observational researchers who do not have resources to develop their own assimilation systems. Future DART releases will include enhanced parallel methods that scale for thousands of processors, novel algorithms to deal with nonlinearity and non-Gaussianity in ensembles, and carefully documented MATLAB versions of the core DART algorithms for students. DART users are also contributing new models, observation types, and algorithms. By providing a nexus for a growing community of data assimilation users and experts, DART can provide an increasingly powerful and flexible set of tools for ensemble data assimilation. © 2009 American Meteorological Society.