Observations of wind profiles within the tropical cyclone boundary layer have until recently been quite rare. The recent massive increase in observations due to the operational implementation of the global positioning system dropwindsonde has emphasised that a low-level wind speed maximum is a common feature of the tropical cyclone boundary layer. Here is proposed a mechanism for producing such a maximum, whereby strong inward advection of angular momentum generates the supergradient flow. The processes that maintain the necessary inflow against the outward acceleration resulting from gradient wind imbalance are identified as being (i) vertical diffusion, (ii) vertical advection, and (iii) horizontal advection. Two complementary tools are used to diagnose the properties and dynamics of the jet. The first, presented here, is a linear analytical model of the boundary layer flow in a translating tropical cyclone. It is an extension of the classical Ekman boundary layer model, as well as of earlier w... Abstract Observations of wind profiles within the tropical cyclone boundary layer have until recently been quite rare. The recent massive increase in observations due to the operational implementation of the global positioning system dropwindsonde has emphasised that a low-level wind speed maximum is a common feature of the tropical cyclone boundary layer. Here is proposed a mechanism for producing such a maximum, whereby strong inward advection of angular momentum generates the supergradient flow. The processes that maintain the necessary inflow against the outward acceleration resulting from gradient wind imbalance are identified as being (i) vertical diffusion, (ii) vertical advection, and (iii) horizontal advection. Two complementary tools are used to diagnose the properties and dynamics of the jet. The first, presented here, is a linear analytical model of the boundary layer flow in a translating tropical cyclone. It is an extension of the classical Ekman boundary layer model, as well as of earlier w...

Observations of wind profiles within the tropical cyclone boundary layer until recently have been quite rare. However, the recent spate of observations from the GPS dropsonde have confirmed that a low-level wind speed maximum is a common feature of the tropical cyclone boundary layer. In Part I, a mechanism for producing such a maximum was proposed, whereby strong inward advection of angular momentum generates the supergradient flow. The processes that maintain the necessary inflow against the outward acceleration due to gradient wind imbalance were identified as being (i) vertical diffusion, (ii) vertical advection, and (iii) horizontal advection, and a linear analytical model of the boundary layer flow in a translating tropical cyclone was presented and used to diagnose the properties of the jet and the near-surface flow. A significant shortcoming was that the jet was too weak, which was argued to be due to the neglect of vertical advection. Here, a high-resolution, dry, hydrostatic, numerical ... Abstract Observations of wind profiles within the tropical cyclone boundary layer until recently have been quite rare. However, the recent spate of observations from the GPS dropsonde have confirmed that a low-level wind speed maximum is a common feature of the tropical cyclone boundary layer. In Part I, a mechanism for producing such a maximum was proposed, whereby strong inward advection of angular momentum generates the supergradient flow. The processes that maintain the necessary inflow against the outward acceleration due to gradient wind imbalance were identified as being (i) vertical diffusion, (ii) vertical advection, and (iii) horizontal advection, and a linear analytical model of the boundary layer flow in a translating tropical cyclone was presented and used to diagnose the properties of the jet and the near-surface flow. A significant shortcoming was that the jet was too weak, which was argued to be due to the neglect of vertical advection. Here, a high-resolution, dry, hydrostatic, numerical ...

The boundary layer in a tropical cyclone is in some respects unlike that elsewhere in the atmosphere. It is therefore necessary to evaluate boundary layer parameterizations for their suitability for use in tropical cyclone simulation. Previous work has shown substantial sensitivity to the choice of scheme and identified specific shortcomings in some schemes, but without recommending which schemes are most suitable. Here, several schemes, representative of those available in popular modeling systems, are reviewed and applied in a simplified modeling framework. Based on a comparison with observations and on theoretical grounds, one popular class of schemes is shown to be badly flawed in that it incorrectly predicts the near-surface wind profile, and therefore should not be used. Another is shown to be sensitive to diagnosis of the boundary layer depth, a difficult problem in the core of the tropical cyclone, and caution is advised. The Louis boundary layer scheme and a higher-order closure scheme are, so far as can be discerned, without major problems, and are recommended. The recommendations and discussion herein should help users make a more informed choice of boundary layer parameterization, and to better understand the results that they obtain.

The GPS dropsonde allows observations at unprecedentedly high horizontal and vertical resolution, and of very high accuracy, within the tropical cyclone boundary layer. These data are used to document the boundary layer wind field of the core of Hurricane Georges (1998) when it was close to its maximum intensity. The spatial variability of the boundary layer wind structure is found to agree very well with the theoretical predictions in the works of Kepert and Wang. In particular, the ratio of the near-surface wind speed to that above the boundary layer is found to increase inward toward the radius of maximum winds and to be larger to the left of the track than to the right, while the low-level wind maximum is both more marked and at lower altitude on the left of the storm track than on the right. However, the expected supergradient flow in the upper boundary layer is not found, with the winds being diagnosed as close to gradient balance. The tropical cyclone boundary layer model of Kepert and Wang is used to simulate the boundary layer flow in Hurricane Georges. The simulated wind profiles are in good agreement with the observations, and the asymmetries are well captured. In addition, it is found that the modeled flow in the upper boundary layer at the eyewall is barely supergradient, in contrast to previously studied cases. It is argued that this lack of supergradient flow is a consequence of the particular radial structure in Georges, which had a comparatively slow decrease of wind speed with radius outside the eyewall. This radial profile leads to a relatively weak gradient of inertial stability near the eyewall and a strong gradient at larger radii, and hence the tropical cyclone boundary layer dynamics described by Kepert and Wang can produce only marginally supergradient flow near the radius of maximum winds. The lack of supergradient flow, diagnosed from the observational analysis, is thus attributed to the large-scale structure of this particular storm. A companion paper presents a similar analysis for Hurricane Mitch (1998), with contrasting results.

Part I of this paper presented a detailed analysis of the boundary layer of Hurricane Georges (1998), based mainly on the newly available high-resolution GPS dropsonde data. Here, similar techniques and data are used to study Hurricane Mitch (1998). In contrast to Hurricane Georges, the flow in the middle to upper boundary layer near the eyewall is found to be strongly supergradient, with the imbalance being statistically significant. The reason for the difference is shown to be the different radial structure of the storms, in that outside of the radius of maximum winds, the wind decreases much more quickly in Mitch than in Georges. Hurricane Mitch was close to inertially neutral at large radius, with a strong angular momentum gradient near the radius of maximum winds. Kepert and Wang predict strongly supergradient flow in the upper boundary layer near the radius of maximum winds in this situation; the observational analysis is thus in good agreement with their theory. The wind reduction factor (i.e., ratio of a near-surface wind speed to that at some level further aloft) is found to increase inward toward the radius of maximum winds, in accordance with theoretical predictions and the analysis by Franklin et al. Marked asymmetries in the boundary layer wind field and in the eyewall convection are shown to be consistent with asymmetric surface friction due to the storm’s proximity to land, rather than to motion. The boundary layer flow was simulated using Kepert and Wang’s model, forced by the observed storm motion, radial profile of gradient wind, and coastline position; and good agreement with the observations was obtained.

A major limitation of the Ensemble Kalman Filter (EnKF) is that the finite ensemble size introduces sampling error into the background covariances, with severe consequences for atmospheric and oceanographic applications. The negative effects of sampling error are customarily limited by covariance localisation, which earlier studies have suggested may introduce imbalance into the system. The deleterious effects of localisation upon balance are confirmed and detailed here, with localisation producing analyses with weaker geostrophic balance and stronger divergence than are obtained using the unlocalised covariances. These imbalances reduce as the localisation radius is increased, but are argued to be large for typical settings. An improved method for calculating local covariances from an ensemble is presented, in which the localisation is performed in streamfunction–velocity potential (ψ–χ), rather than wind component, space. Analyses using this method better preserve the balances contained within the unlocalised covariance model. This transformation further allows the option of intervariable localisation, in which the cross‐covariances involving χ, which are weak and therefore particularly subject to sampling error, are set to 0 instead of being calculated from the ensemble. The various localisations are compared in a series of identical‐twin experiments, with the new localisations producing analyses that are better balanced and significantly more accurate than the usual approach. The localisation with the χ cross‐covariances set to 0 is shown to be superior for the smaller ensemble sizes but not for the larger, implying that the larger ensembles are capable of resolving some of the true χ cross‐covariance in the test system. Copyright © 2009 Royal Meteorological Society