Evaluation of Local Ensemble Transform Kalman Filter System for the Global FSU Atmospheric Model
Tuesday, 16 December 2014
This paper shows the results of a implementation of the data assimilation system to obtain the initial condition to the atmospheric general circulation model (AGCM) for the Florida State University/USA. The better quality of forecasts is given the more accurate the estimate of the initial conditions. The process of combining observations and short-range forecast to obtain an analysis is called data assimilation. The data assimilation system called “Local ensemble transform Kalman filter (LETKF) is implemented. A prediction estimates ensemble in state space represents the model errors in that scheme. The LETKF is tested with the AGCM Florida State University Global Spectral Model (FSUGSM). The model is a multilevel (27 vertical levels) spectral primitive equation model with a vertical σ-coordinate. All variables are expanded horizontally in a truncated series of spherical harmonic functions (at resolution T63) and a transform technique is applied to calculate the physical processes in real space. The LETKF data assimilation experiments are based in two types of the synthetic observations data (surface pressure, absolute temperature, zonal component wind, meridional component wind and humidity) to evaluate the LETKF system for FSUGSM. The data assimilation experiments are based on observational systems simulation experiments where the “nature“ is assumed to be known, and adding random noise to the nature run. The first experiment, the “nature” fields are the FSUGSM forecasts without data assimilation, afterwards, we use the “National Centers for Environment Prediction” reanalysis to obtain the “nature” fields. The observations are localized at every other grid point of the model. The forecast ensemble size is 20 members. The numerical experiments have a one-month assimilation cycle, for the period 01/01/2001 to 31/01/2001 at (00, 06, 12 and 18 GMT) for each day. We compare the behavior of the model by comparing with its forecast, observations and nature fields. A source of information for the quality of any data assimilation is the observation-minus-forecast (OmF) and the observation-minus-analysis (OmA) statistics. The histogram of OmF and OmA for a range of spatial and temporal scales is calculated. The results showing the analysis from the assimilation of the two observations experiments will be presented.