The vortex relocation scheme operates on the background forecasts prior to the assimilation of observations, and performs a procedure that extracts a vortex from the model large-scale environment flow and relocates it at the observed location. While the scheme has been demonstrated to reduce track errors in older versions of the GFS, recent sensitivity experiments show that for newer generations of the GFS, this mechanical relocation may produce increased errors in track forecasts at longer lead times despite reducing the initial position uncertainty. In this study, we will present the results from such sensitivity experiments utilizing the operational hybrid 3D EnVar-based GDAS system to initialize the operational resolution GFS, comparing results from a control with an experiment that excludes the vortex relocation.
As part of the current operational hybrid 3D EnVar scheme, an 80-member ensemble is evolved and updated using a serial Ensemble Square Root Filter (EnSRF) to prescribe estimates of the flow-dependent, evolving background error covariance for the deterministic assimilation. Currently, the ensemble gets no special treatment for tropical cyclones other than the assimilation of bogus winds (when generated) and advisory minimum sea-level pressure. We will provide motivation for a scheme that is effectively a combination of position assimilation and mechanical relocation, and demonstrate that such a scheme can be utilized to improve error covariances for tropical cyclone vortices, which will hopefully yield improved deterministic initialization for the full resolution deterministic GFS.
- Indicates paper has been withdrawn from meeting
- Submission entered in competition