NCAR Data Assimilation Initiative - AKA Data Assimilation Research Section

The NCAR Data Assimilation Initiative was founded to create and to then lead a research community for data assimilation where individuals benefit from sharing ideas, methodologies, and software tools as well as access to a data assimilation testbed. NCAR has a large number of researchers for whom data assimilation is an essential part of their ongoing or planned research. New developments in theoretical data assimilation and in software engineering are making collaborations between data assimilation experts, modelers, observational specialists and statisticians easier and more productive than was possible in the past. The maturation of the Initiative resulted in the Data Assimilation Research Section (DAReS): a component of the Institute for Mathematics Applied to Geosciences. The primary goal of DAReS is to continue to advance the theory and practice of ensemble data assimilation. Also, DAReS accelerates the progress of many other NCAR projects by providing a centralized data assimilation expertise which can be coordinated with existing observational and modeling expertise.

Ensemble Filters for Geophysical Data Assimilation

Download DART

The Data Assimilation Research Testbed Facility : DART

The Data Assimilation Research Testbed (DART), is a software environment for making it easy to match a variety of data assimiliation methods to different numerical models and different kinds of observations. DART has been through the crucible of many compilers and platforms. It is ready for friendly use and has been used in several field programs requiring real-time forecasting. The DART source code is distributed through our anonymous subversion server (meaning you don't need an account on our machines) after you fill out a trivial form asking for your name, email address, and institution. We need to gather summary statistics to maintain an email list should there be an important update and to help us quantify our impact on the community. This is an ultra-low-traffic email list. We have never sent an email to the entire list! Please be assured we will keep your email address confidential.

It has been more than a year since the Jamaica version was released (on 12-Apr-2007). There have been some bugfixes and improvements to the Jamaica version that are available in the current development version, such that the DAReS support team would prefer you to work with the current development version of the code (subversion calls it the HEAD) The DART source code and documentation can be downloaded at www.image.ucar.edu/DAReS/DART/DART_download

The Jamaica and subsequent versions use an updated scalable filtering algorithm and (optionally) MPI for greatly improved performance. Much effort has gone into extensible observation processing and support for real observations. The inflation algorithms (including a posterior inflation option) have been improved and are now part of the filter namelist (for those of you in the know). The most current version of the documentation is distributed with the code, but if you want an idea of what it takes to build DART and run experiments before downloading, look at the Jamaica release document. There is also a Jamaica differences document that is intended to be an exhaustive list of differences between the Jamaica release and the previous version (the Hawaii release).

Schematic of Ensemble Data Assimilation - from the DAReS Perspective

This is the DART view of ensemble data assimilation for models that run as separate executables. Starting at the top and working clockwise: Everything is driven by a Fortran namelist and the presence or absence of observations. A Fortran executable named 'filter' reads a namelist, an initial state for the ensemble, and a file containing observations and goes to work. Given the observations and an initial state, 'filter' assimilates the observations and then determines how far to advance the model (using information from the namelist and the observation file). 'filter' forks a shell script to the system and it is this shell script that is responsible for three things: 1) for converting the DART state vectors and 'advance_to_time' to the format required by the underlying model, 2) advancing the model, and 3) converting the model output into a form suitable for 'filter'. [The script is responsible for the lower portion of the diagram.] The model advances each ensemble member (either in turn or all-at-once) and the model output is converted to the input format expected by 'filter'. The shell script finishes and signals 'filter' to continue. We are now back at the beginning and the cycle continues as long as there are observations to assimilate or until the control information in the Fortran namelist is met. When that happens, a set of restart files is written (suitable to continue an experiment with more observations) and diagnostic files are written. These diagnostic files allow for the exploration of the assimilation before and after each assimilation step and for exploration of the assimilation in 'observation space'; each real observation is paired with the estimates of the observation from all of the ensemble members (if desired). Minimally, the ensemble mean estimate of the observation and the ensemble spread of the estimates is recorded.

We regularly give tutorials on Ensemble Data Assimilation

Summer Graduate Workshop on Data Assimilation for the Carbon Cycle: 8-13 July 2007

This summer school exposed students in the geosciences, ecology, and mathematics to multidisciplinary science through a focus on estimating the sources and sinks of carbon for the Earth system. One goal is to train the next generation of researchers to work within a multidisciplinary science team that combines geoscientists, ecologists, applied mathematicians, and statisticians. The home page for the workshop is available [here].

Joint SAMSI/IMAGe Workshop: June 2005

The ability to combine observations with a numerical model is critical to understanding and predicting geophysical systems like the earth's atmosphere. This summer school was presented by statisticians and geophysicists who are leaders in the field of data assimilation. By bridging the gap between basic and applied research on ensemble data assimilation, the workshop provided participants with an understanding of the most recent advances and the most critical unsolved problems in this rapidly growing field. Lectures and discussion were supplemented by a series of computational explorations using the Data Assimilation Research Testbed facility at the National Center for Atmospheric Research. By the end of the workshop, participants were equipped with the tools to attack the problems posed in the lectures and to undertake research in data assimilation for a large variety of applications. The home page for the workshop is available [here].

Our central email address is dart@ucar.edu, which will hit 'everyone' and find its way to the best person.
The categories that follow are not set in stone, everyone has some expertise in all areas.

Shipping information: