Mason Archival Repository Service

Improvement of Soil Moisture Prediction through AMSR-E Data Assimilation

Show simple item record

dc.contributor.author Sahoo, Alok Kumar
dc.creator Sahoo, Alok Kumar
dc.date 2008-04-28
dc.date.accessioned 2008-07-01T19:39:32Z
dc.date.available NO_RESTRICTION en
dc.date.available 2008-07-01T19:39:32Z
dc.date.issued 2008-07-01T19:39:32Z
dc.identifier.uri https://hdl.handle.net/1920/3142
dc.description.abstract This dissertation is aimed at evaluating the soil moisture estimation from satellites as well as land surface models and improving it using a data assimilation technique. The entire study was conducted over the Little River Experimental Watershed, Georgia for the year 2003; one of the four selected watersheds to validate the current AMSR-E satellite soil moisture data. Soil moisture data from a comprehensive in-situ observation network at this watershed were first used to study the spatial and temporal soil moisture characteristic of the watershed. There was a high degree of spatial and temporal correlation among different measurement stations which was required to validate other datasets with lower spatial and temporal frequency. Hence, those in-situ observations were treated as ground truth to validate other soil moisture datasets in this dissertation. A satellite based soil moisture product was generated from AMSR-E satellite brightness temperature data using the LSMEM radiative transfer model. This research product was found to be statistically better than the current AMSR-E soil moisture product when both the datasets were compared against the in-situ observations. Similarly, three land surface models pertaining to different model physics and parameterization were simulated to generate soil moisture over the watershed. There was quite a bit of disagreement among model soil moisture results which was also reflected in other water and energy cycle variables since they were mostly controlled by soil moisture. Noah model soil moisture was found to be better than those of other two models even though it had a constant positive bias. When the LSMEM soil moisture observations were assimilated into the Noah land surface model using the EnKF algorithm, the Noah model predictions got improved significantly. This was confirmed by calculating the improvement metric over the Noah openloop simulations. The EnKF algorithm was found to be sensitive to the model initialization and spin-up conditions. In the end, the assimilated soil moisture results were used to demonstrate two real world applications. It was found that the relationship between the winter/spring soil moisture and vegetation during growing season was different for different vegetations types. This assimilated sol moisture map was also able to show the spatial and temporal extent of the 2003 May flooding event over Tennessee, Alabama and Georgia accurately. The conclusion chapter discusses the limitations we faced during this research work and many research extensions that can be performed to this research work. This assimilated soil moisture shows lot of promise for real world applications. This product can operationally be produced at finer spatial and temporal scales which is required for any kind of real world applications.
dc.language.iso en_US en
dc.subject soil moisture en_US
dc.subject data assimilation en_US
dc.subject land surface models en_US
dc.subject microwave satellite remote sensing en_US
dc.subject little river experimental watershed en_US
dc.title Improvement of Soil Moisture Prediction through AMSR-E Data Assimilation en
dc.type Dissertation en
thesis.degree.name Doctor of Philosophy in Computational Sciences and Informatics en
thesis.degree.level Doctoral en
thesis.degree.discipline Computational Sciences and Informatics en
thesis.degree.grantor George Mason University en


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search MARS


Browse

My Account

Statistics