VARIATIONAL DATA ASSIMILATION OF SOIL MOISTURE INFORMATION

Simple item page
dc.contributor.advisorKalnay, Eugenia Een_US
dc.contributor.advisorMitchell, Kenneth Een_US
dc.contributor.authorGrunmann, Pablo Javieren_US
dc.contributor.departmentMeteorologyen_US
dc.contributor.publisherDigital Repository at the University of Marylanden_US
dc.contributor.publisherUniversity of Maryland (College Park, Md.)en_US
dc.date.accessioned2005-08-03T14:22:17Z
dc.date.available2005-08-03T14:22:17Z
dc.date.issued2005-04-20en_US
dc.description.abstractThis research examines the feasibility of using observations of land surface temperatures (in principle available from satellite observations) to initialize soil moisture (which is not available on a continental scale). This problem is important because it is known that wrong soil moisture initial conditions can negatively affect the skill of numerical weather prediction models. Since this problem requires the availability of a good soil model, considerable effort was devoted to the improvement of several aspects of the NCEP Noah land surface model and its numerical properties (reliability, efficiency, updates and differentiability). When tested against the experimental station data at Champaign, IL collected by Dr. Tilden Meyers of NOAA/ARL, where the surface fluxes, precipitation, and surface temperature were available, the Noah model forced with observed downward radiative surface fluxes and near-surface meteorology, including precipitation, was able to reproduce the observations quite well. A method for data assimilation was developed and tested, in a manner similar to 4-dimensional variational assimilation (4D-Var) in the sense of applying the temporal behavior of the observed variable but with a single spatial dimension (land surface models are typically “column models”, as they do not usually compute horizontal derivatives). The results show that it is indeed possible to assimilate land surface temperature and use it to correct soil moisture initial conditions, which may manifest significant errors if, for example, the precipitation forcing the model is significantly biased. This is true, however, only if the surface forcings besides precipitation are essentially correct. When surface forcing come from the North American Land Data Assimilation System (NLDAS) as they would be available for operational use over the US, the results are not satisfactory. This is because the assimilation changes the soil moisture to correct for problems in the simulated land surface temperature that are at least partially due to other sources of errors, such as the surface radiative fluxes. We suggest that in order to succeed in the soil moisture initialization, more (and more accurate) observations are needed in order to constrain the dependence of the observation part of the cost function solely on soil moisture.en_US
dc.format.extent6567814 bytes
dc.format.mimetypeapplication/pdf
dc.identifier.urihttp://hdl.handle.net/1903/2476
dc.language.isoen_US
dc.subject.pqcontrolledPhysics, Atmospheric Scienceen_US
dc.subject.pquncontrolledData Assimilationen_US
dc.subject.pquncontrolledNumerical Weather Predictionen_US
dc.subject.pquncontrolledMeteorologyen_US
dc.subject.pquncontrolledSoil Moistureen_US
dc.subject.pquncontrolledLand Surface Modelen_US
dc.titleVARIATIONAL DATA ASSIMILATION OF SOIL MOISTURE INFORMATIONen_US
dc.typeDissertationen_US

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
umi-umd-2348.pdf
Size:
6.26 MB
Format:
Adobe Portable Document Format
Download

Collections

UMD Theses and Dissertations
Atmospheric & Oceanic Science Theses and Dissertations