6.6
The characteristics of the initial state corrections obtained with the CMC key analysis errors algorithm: A dynamical perspective
Jean-Francois Caron, McGill University, Montreal, QC, Canada; and M. K. Yau, S. Laroche, and P. Zwack
Different techniques have been developed to estimate the initial state errors responsible for poor short-range to medium-range forecasts. One technique implemented at the Canadian Meteorological Center (CMC) is the key analysis errors algorithm, based on the adjoint technique and which adopts the energy norm to measure the forecast error. In this work, the estimated key analysis errors were analyzed with regards to dynamical balance and observational data in order to evaluate their degree of realism. Results show that both rotational and divergent part of the estimated key analysis errors are strongly out of balance initially and that the mass component of the initials corrections contains an important part of dissipating modes. Comparison with observational data showed that the corrected analysis is systematically further away from the observation than the control analysis. We tested a few approaches to isolate the balanced component of the initial corrections from the CMC key analysis errors algorithm, in an attempt to capture the relevant part of estimated key analysis errors. The best results are obtained with the nonlinear balance PV inversion technique. It is shown that the PV component of the initial corrections contains the essential information for reducing short range forecast errors. The remaining imbalance part of the initial corrections does not grow in time and does not contribute to the forecast improvement as well. The removal of the imbalance part of the initial corrections makes the corrected analysis slightly closer to the observations, but remains systematically further away as compared to the control analysis. Different ways to improve the estimation of the key analysis errors will also be discussed.
Session 6, Assimilation of Observations (Ocean, Atmosphere, and Land Surface) into Models: Assimilation Methods; Minimization Techniques; Forward Models and Their Adjoints; Incorporation of Constraints; Error Statistics
Wednesday, 1 February 2006, 8:30 AM-12:00 PM, A405
Previous paper Next paper