Analyzing the impacts of non-gaussian errors in gaussian data assimilation systems
CSU Project #: 5-34692Sponsor: National Science Foundation (NSF)
Primary Investigator: Dr. Steven Fletcher
Co-Investigator(s): Dr. Andrew S. Jones
Start Date: September 1st, 2012
End Date: August 31st, 2015
Objectives:
- During the first year the lognormal version of the Bayesian MLE will be programmed along with the transform approach.
- Two important mathematical features will be derived and coded in the first year. These are a lognormal Bayesian quality control measure, and performance metrics that do not rely on any Gaussian assumptions.
- Also in the first year validation data will be obtained. This will be in the form of GPS integrated total precipitable water vapor for the moisture variable.
- During the first year the brightness temperatures will be obtained in the first year that will be used as the observational component of the retrieval system. The area that will be looked in the first year and into the second will be off the coast of Japan during 2004.
- Also in the second year the retrieval system will be tested over the Atlantic near to the Caribbean islands, specifically in the hurricane season of 2006.
- It is planned to test the effects of the different retrieval systems in the WRF-VAR system for this test case to study the impacts the retrieval have on the forecast of this tropical storm.
- These profiles will then be assimilated into the three difference versions of the WRF-VAR system to study the impacts of forecasting these events with incorrect retrieved data as well as assimilating the data incorrectly.
Progress Reports:
** No Progress Report **CIRA Themes relevant to this project:
- Data Assimilation