I am currently working on implementing the Local Ensemble Transform Kalman Filter (LETKF) on a ROMS model of the Chesapeake Bay. The LETKF is an advanced method for data assimilation and was developed by the Weather and Chaos group at the University of Maryland, College Park. I am applying it to the ChesROMS model, which is being developed by scientists at NOAA, University of Maryland, CRC (Chesapeake Research Consortium) and MD DNR (Maryland Department of Natural Resources). This research began during my dissertation at the University of Maryland and was advised directly by Raghu Murtugudde and Christopher W. Brown.  

Identical twin experiments have shown that the LETKF quickly reduces the analysis error, even with a very sparse data set. In addition, the ensemble spread has been utilized to explore the seasonal variability of the Chesapeake Bay and to advise new observation locations. The H-operator (observation operator) for the Bay was developed so that observation could be assimilated in real locations.

I am continuing this work as part of my postdoc at Johns Hopkins along with Tom Haine and Darryn Waugh. We are currently exploring the impact of forcing errors on the assimilation and ways to correct these errors using the assimilation. We have also begun work with assimilating real satelite and in situ data. We are also working with other scientists to derive salinity information from satellite temperature and ocean color measurements and assimilate these as well.

In addition to the coastal ocean domain, I am working on ensemble data assimilation of the Martian atmosphere. This is part of a NASA grant and involves people from the University of Maryland, GFDL, and AER Inc. I have interfaced the LETKF with John Wilson's NASA/NOAA Mars GCM to improve prediction of the Martian weather and our understanding of the Martian climate. Initial studies have been run using simulated observations in a perfect model framework and have confirmed the viability of the LETKF for the assimilation of satellite observations from the Thermal Emission Spectrograph (TES) on the Mars Global Surveyor (MGS). Ongoing work is in assimilating real observations from the TES mission. Details of the initial results can be found in my dissertation and a paper on the first stage of the work was published in Icarus.

As part of this reseach I am currently working on coding the assimilation of radiance information and also in testing the Optimal Spectral Sampling (OSS) radiative tranfer model that we plan on using. The OSS model was developed at AER in Lexington, MA and I am collaborating with a few researchers there.

My research has also included developing energy equations for the breeding method. Breeding allows us to identify instabilities in the ocean and the shapes of prediction errors, among other things. The breeding energy equations allow the diagnosis of the type of energy conversion that takes place. Using the Modular Ocean Model (MOM), originally created by the Geophysical Fluid Dynamics Laboratory (GFDL), and modifying it to run breeding experiments, we have applied the energy equations to the study of Pacific Tropical Instability Waves.  My advisors for this project are James Carton and Eugenia Kalnay.  Results have been presented at a number of conferences (see below) and a publication on the research was published in GRL in 2009.

In addition, this method is now being applied to the Martian atmosphere to identify and diagnose instabilities. I have been assisting in an effort led by Steven J. Greybush at the University of Maryland, College Park, which will be a portion of his Ph.D. dissertation. A first paper on this topic is currently being written.

In addition, I am part of a collaborative effort between CPTEC/INPE in Brazil and the University of Maryland to develop an operational global oceanic data assimilation using the LETKF and the MOM4 ocean model. The first phase of the project was completed in May 2008 and involved interfacing the LETKF with the MOM4 model. The MOM4 setup used has 1°x1° horizontal resolution with 50 vertical levels and is coupled to an ice model. Assimilating temperature, salinity, and current observations in identical twin experiments have shown a quick and asymptotic reduction on both analysis errors and the errors of the subsequent forecasts. Results from the first phase of this project were presented at the Brazilian Meteorological Congress and a short paper appeared in the conference proceedings. 

The second part of the research involved applying the LETKF to a regional model of the Southern Atlantic Ocean.