Abstract
The effect of resolution on the depiction of central sea level pressure for an intense oceanic extratropical cyclone is examined through a one-dimensional Fourier analysis. Profiles of sea level pressure were manually interpolated along the latitude passing through the storm center from two subjective analyses and the 00-, 24-, and 48-h NMC Nested-Grid Model (NGM) forecasts, all valid at 0000 UTC 5 January 1989. At this time, the Experiment on Rapidly Intensifying Cyclones over the Atlantic (ERICA) intensive observing period 4 (IOP 4) cyclone attained its maximum intensity, with a central pressure of 936 mb at 41°N, 58°W in an analysis prepared by Frederick Sanders.
After the Fourier coefficients were determined for each pressure profile they were used to recompute a series of pressure profiles truncated at various maximum wavenumbers λmax. The central sea level pressures obtained from these truncated profiles asymptotically approach the central pressure of the original profile as λmax increases. An effective resolution is defined as the λmax at which the truncated central pressure comes within 1 mb of the original central pressure. This investigation reveals an effective resolution for the NGM of approximately wave-number 50 (604 km at 41°N) compared to wavenumber 100 (302 km at 41°N) for the hand analyses. A similar examination of the magnitude and location of the maximum v component of the geostrophic wind computed from the pressure profiles supports these estimates of effective resolution. However, maximum gradient wind speed was found to be relatively insensitive to resolution, while maximum geostrophic relative vorticity was overly sensitive to resolution. Consequently, neither of the latter two were used to derive effective resolutions.
The recomputed pressure profiles truncated at wavenumber 50 are virtually identical to the original profiles for the NGM cases. However, the truncated pressure profiles for the hand analyses yield central pressures that approach the NGM 00-h value. Thus, resolution differences alone account for approximately half of the 20-mb central-pressure error in the NGM 24- and 48-h forecasts, while apparent deficiencies in the model, initial conditions, or boundary conditions must account for the other half. In other words, the best performance that could be expected from the current NGM configuration is a central pressure of approximately 945 mb rather than 936 mb, assuming the same background pressure.
