DART Manhattan Documentation

DART/CAM spaghetti plot

00Z 01 Feb 2003. 20 of 80 member ensemble. T85 CAM GPH at 500hPa.

Sensitiviy Analysis on Hurricane Katrina

Ensembles allow exploration of circulation on hurricane position.

GPS RO observations

Rapid support for new observation types.

Chemical Weather Forecasting and Analysis




CAM inflation values for U Wind

after 1 month of adaptive inflation.

CAM inflation values for U Wind

after 1 month of DAMPED adaptive inflation.

DART/CMAQ CO assimilation


Planetary Boundary Layer

Probabilistic nowcasting with a single column model.

A good rank histogram

Lorenz '96 - the true state is as likely as any of the ensemble members.

A bad rank histogram (filter divergence)

Lorenz '96 - the true state generally falls outside the ensemble members.

Tracer advection

One of the tutorial examples explores sources/sinks.

Impact of PS observations on V winds

DART has novel algorithms to explore localization.

Ensemble Trajectories

DART has routines to explore and diagnose.

Ensemble Trajectories

Explore the impact of the increments.

time series of 'total error'

Quantify the effects of observations or parameter settings.

CAM@T85 observation space performance

DART natively supports direct comparisons with observations.

CAM@T85 assimilation performance

See if all the observations are being used.

DART research is broadly categorized along three avenues: one is the research toward data assimilation algorithmic and computational efficiency, another is toward implementing and exploiting the information in observations with one model or another, and another is to learn more about the behavior of an individual model - perhaps by looking at systematic features of the increments.

If you would like to add your research highlight using DART, please let us know!


Research Projects involving DART


Model Performance

DART/CAM spaghetti plot Ensemble Data assimilation can provide qualitative and quantitative uncertainty for quantities of interest to weather forecasters. [link to more information]

Kevin Raeder, raeder@ucar.edu


Whole Atmosphere Community Climate Model (WACCM)

WACCM graphic  

This is an excerpt from Applying the science and technology of data assimilation by Brian Bevirt 07/11/2017 as part of a CISL News series describing the many ways CISL improves modeling beyond providing supercomputing systems and facilities.

These plots show measured and modeled zonal mean temperatures between 70N and 90N during the January 2009 sudden warming of the stratosphere. The bottom plot shows the observed temperatures (in degrees Kelvin, see legend at right), the center plot shows how this state of the atmosphere was simulated by the specified-dynamics version of the WACCM model, and the top plot shows WACCM's improved result after using DART to assimilate middle atmosphere observations.

The key point in this figure is that WACCM+DART captures both the stratosphere warming and mesosphere cooling that are seen in the observations. Also seen in the specified-dynamics version of WACCM, the elevated stratopause that forms at high altitudes around day 30 descends too fast compared to the observations. The elevated stratopause is maintained at a high altitude in the WACCM+DART simulation. This has implications for the descent of species from the mesosphere into the stratosphere. Accurate representation of the mesosphere dynamics is important for the ionosphere variability during sudden stratosphere warming events. (Figure courtesy of Nick Pedatella, HAO)

[link to more information]

Nick Pedatella, nickp@ucar.edu Hanli Liu, liuh@ucar.edu Jing Liu, jingliu@ucar.edu


Open Geospace General Circulation Model (thermosphere/ionosphere/magnetosphere)

OpenGGCM graphic The primary goal of this project is to combine the OpenGGCM (Open Geospace General Circulation Model) with the NCAR Data Assimilation Research Testbed (DART), which implements an Ensemble Kalman Filter (EnKF) and will enable the sequential assimilation of ionosphere, thermosphere, and magetosphere data.

We will also optimize model parameters by including them into the state vector. This will improve model accuracy even when no data are assimilated. [link to more information]

Jimmy Raeder, J.Raeder@unh.edu


Chemical Data Assimilation

CAM-Chem/DART CO Column We are currently applying an ensemble-based chemical data assimilation system, consisting of regional to global chemical transport models (CAM-Chem, WRF-Chem) in conjunction with DART, for a joint assimilation of meteorological observations and satellite-derived CO measurements from MOPITT and aerosol optical depth (AOD) measurements from MODIS. The chemical data assimilation system has been recently used for near-real time chemical forecasting (see https://espo.nasa.gov/arctas/) to support flight planning during the NASA Arctic Research of the Composition of the Troposphere from Aircraft and Satellites (ARCTAS). [link to more information]

Ave Arellano, arellano@ucar.edu


GPS RO Observations and Tropical Cyclone Forecasting

GPS RO geometry schematic Profiles of atmospheric quantities deduced from GPS Radio Occultation data are available in otherwise data-sparse regions and provide information used to forecast the behavior of tropical cyclones. The COSMIC/FORMOSAT-3 mission has been providing about 2000 data profiles per day since September 2007. [link to more information]

Hui Liu, hliu@ucar.edu


Sensitivity Analyses

DART/WRF sensitivity plot Forecast sensitivity analysis provides an objective means of evaluating how initial condition errors affect a forecast and where to gather additional observations to reduce forecast errors. Most sensitivity studies use the adjoint of a linearized forecast model to determine the gradient of a forecast metric with respect to the initial conditions. Adjoints suffer from a number of difficulties including coding, linearity assumptions, and moist processes. Ensemble-based sensitivity analysis provides an attractive alternative to adjoint-based methods because it combines data assimilation and sensitivity analysis in a consistent manner. This image illustrates the effect of zonal winds aloft on the position of Hurricane Katrina. [link to more information]

Ryan Torn, torn@atmos.albany.edu


Assimilation of CO

CMAQ CO assimilation result This project describes an integrated approach to modeling atmospheric chemistry with trace gas data assimilation. Specifically, we ran CMAQ from within DART to assimilate both synthetic and real observations of CO for the period of June 2001. [link to more information]

Alexis Zubrow, azubrow@unc.edu


Assimilation on MARS

DART/MARS_Lawson graphic The planetary atmospheres group at Caltech has produced a global and planetary version of NCAR's WRF (Weather Research and Forecasting) Model. We are using DART to attempt data assimilation within the Mars atmosphere using the Mars version of WRF, MarsWRF, as our GCM. [link to more information]

Greg Lawson, wglawson@gps.caltech.edu


Inertio-Gravity waves

DART/CAM normal modes This project aims at quantifying the impact of various motion types in analysis and forecast fields by using normal modes. The DART/CAM is the main analysis system used in the project. The first question addressed is about how large part of the atmospheric energy is associated with the inertio-gravity motions, an important part of the global circulation primarily because of their role in the tropical system. [link to more information]

Nedjeljka Zagar, nedjeljka.zagar@fmf.uni-lj.si


Planetary Boundary Layer

DART/CAM spaghetti plot A long-term goal of this work is to find an efficient system for probabilistic planetary boundary layer (PBL) nowcasting that can be employed wherever surface observations are present. One approach showing promise is the use of a single column model (SCM) and ensemble filter data assimilation techniques. [link to more information]

Dorita Rostkier-Edelstein, rostkier@ucar.edu
Josh Hacker, hacker@ucar.edu


The Lorenz '96 model

Lorenz, E. N., and K. A. Emanuel, 1998:
Optimal sites for supplementary weather observations: Simulations with a small model.
J. Atmos. Sci.55, 399-414.

bad rank histogram The Lorenz '96 model is one of our favorite models. In our implementation, it is a 40-variable model that can be used to test inflation algorithms, the effects of localization schemes, the integrity of the DART installation itself, the state-space diagnostic routines; it is extensively used in the tutorial, and can even be run as a standalone executable to test the MPI support on a machine. [link to more information]

Jeff Anderson, jla@ucar.edu, and
Tim Hoar, thoar@ucar.edu
good rank histogram


Please suggest ways for us to improve DART.