The main goal of this research is to prepare data assimilation methodology capable of extracting maximum information from the future GOES-R data. We employ a data assimilation system which includes the following components: (i) Maximum Likelihood Ensemble Filter (MLEF, an ensemble data assimilation method developed at CIRA), (ii) Weather Research and Forecasting (WRF) model and (iii) CIRA-developed forward radiative transfer operator for all weather conditions. The main applications of the system are for assimilation of synthetic GOES-R ABI radiances and real radiances from the currently available satellites (such as MSG radiances) in severe weather cases. As an example we show the analysis and background errors obtained in assimilation experiments for cyclone Kyrill, presented in the form of probability histograms (see the figures below).
Probability histogram of the errors of the 10.35 µm radiances, calculated with respect to synthetic (“observed”) radiances, for the first guess (FG), the analysis (ANL) and the experiment without assimilation (NO_OBS). Ensemble size of 48 members is used. Results from first (left) and seventh (right) data assimilation cycle are presented. Note the improvements (i.e., errors are clustering more around zero after seven data assimilation cycles) in FG and ANL as compared to NO_OBS. Units for radiances are W m-2 sr-1 cm.
This research activity includes the following participants: Dusanka Zupanski, Milija Zupanski, Louie Grasso, Mark DeMaria, Renate Brummer, Isidora Jankov, Daniel Lindsey and Manajit Sengupta.
More information about ensemble data assimilation at CIRA can be found at the webpage http://www.cira.colostate.edu/projects/ensemble. Related research includes ensemble data assimilation activities under NOAA Hurricane Forecasting Improvement Project (HFIP) and NASA Global Precipitation Measurement (GPM) Program (webpage for the GPM is http://gpm.gsfc.nasa.gov).