Abstract

Omega dropwindsonde (ODW) observations from three synoptic-flow experiments in environment of Hurricane Josephine have been analyzed in a research mode using an objective analysis procedure. The nominal times of the analyses are 0000 UTC 10, 11, and 12 October 1984. The filtered, three-dimensional analyses have been used as a basis for several diagnostic and prognostic calculations relating to the motion of the hurricane.

Examination of Josephine's environment revealed a strong variability of the flow with distance from the storm center and with pressure. Josephine moved at right angles to the azimuthally averaged wind at 500 mb; the vortex motion was more consistent with the flow near 700 mb. Forecasts made with a barotropic forecast model showed a high sensitivity of the forecast track to the vertical layer used in the initial analysis. These results demonstrate the potential value of vertical sounding information from the ODWs, and show that single-level midtropospheric information is not always representative of a hurricane's environment flow.

On each of the three days, the motion of Josephine deviated significantly from its environmental “steering,” as measured by an azimuthal average of the 300–850 mb mean flow over the 5°–7° radial band. This deviation from steering (the so-called “propagation” vector) was oriented with components parallel and to the left of the gradient of absolute vorticity in the asymmetric wind field. The magnitude of the propagation was proportional to the strength of the absolute vorticity gradient. These results are consistent with many barotropic modeling studies.

Abstract

Potential vorticity (PV) analysts for Hurricane Gloria of 1985 are derived from nested objective wind analyses of Omega dropwindsonde and airborne Doppler radar data. The analyses resolve eyewall-scale features in the inner vortex core and embed analyses of these features within the larger-scale environment. Since three-dimensional geopotential height fields required for evaluation of PV are not available in the core, they are derived using the balance equation. In the process of deriving the heights, the degree of gradient balance is evaluated. The 500-mb tangential winds in the core, averaged azimuthally on the four cardinal points, are close to gradient balance outside the radius of maximum wind.

The resulting depiction of PV is the first presented for a real hurricane. Due to data deficiencies immediately outside the Doppler region, as well as inside the eye, smoothing of the wind data using a filter with a minimum 25-km spatial scale is required to derive a balanced geopotential height distribution consistent with a statically stable vortex. The large-scale PV distribution evidences asymmetries in the middle and upper troposphere that appear to be associated with Gloria's translation to the northwest. Eyewall-scale PV in the core and PY of the azimuthally averaged vortex are also presented.

Abstract

A three-dimensional, nested analysis of wind fields in the environment of Hurricane Debby (1982) has been completed. The basic analysis tool uses a two-dimensional least-squares fitting algorithm combined with a derivative constraint that acts as a spatial low-pass filter on the analyzed field. Gridded results of horizontally analyzed fields are combined into vertical cross sections and then analyzed to produce vertical continuity. Consequently, a three-dimensional analysis is obtained.

The database for the analysis comes primarily from Omega dropwindsondes (ODWs), rawinsondes, and satellite-derived winds above 400 mb in the environment of Hurricane Debby near 0000 UTC 16 September 1982. Since these data come from many different sources, and thus are not evenly distributed in the horizontal or vertical, techniques have been developed to alleviate difficulties associated with inhomogeneous data. The analyzed wind fields provide an independent evaluation of satellite-derived winds at and below 400 mb.

General features of the environmental wind fields surrounding Debby are described. The wind analyses are then used to diagnose terms in the vorticity equation. The spatial orientation of a calculated dipole in the horizontal vorticity flux convergence term indicates that it is an approximate indicator of Debby's observed short-term motion.

Finally, to provide an initial assessment of the wind analysis quality, experimental track forecasts with a barotropic model are performed with the layer-mean wind fields and operationally available data outside the analysis domain. Initial errors in the forecast tracks are directly related to the orientation of the diagnosed vorticity flux convergence dipole. The research wind analysis results in a substantial reduction in track error for short-term (12 h) forecasts compared to analyses from operationally available data. This reduction is due to an improved representation of the wind fields in the near-storm environment.

Abstract

A scarcity of observations in the hurricane environment is one factor believed to be limiting the improvement in hurricane track forecast accuracy. Since 1982, the Hurricane Research Division (HRD) of the NOAA Atlantic Oceanographic and Meteorological Laboratory has conducted 14 experiments to determine the wind and thermodynamic fields within about 1000 km of tropical cyclones in the Atlantic basin. During these synoptic-flow experiments, Omega dropwindsondes (ODWs) are released from the two NOAA WP-3D research aircraft over a 9–10-h period in the hurricane environment. The ODWs measure pressure, temperature, humidity, and wind as they descend from flight level (about 400 mb) to the surface. These data are then transmitted in real time to the National Hurricane Center (NHC) and the National Meteorological Center (NMC).

Recently, a barotropic, nested, spectral hurricane track forecasting model, VICBAR, has been developed at HRD and tested quasi-operationally during the 1989 and 1990 hurricane seasons. Forecasts from this model have compared favorably with other models run at NHC and NMC. In this study, the VICBAR model is used to evaluate the impact of ODW data on track forecast error for the 14 HRD synoptic-flow experiments.

The ODW data produced highly consistent reductions in track forecast errors in this sample of cases. Forecast improvements due to single-level midtropospheric (aircraft) data were significantly smaller than those due to the ODWs. At the important verification times of 24–36 h (prior to landfall), when the decision to issue a hurricane warning is being made, the ODWs reduced the model mean forecast error by 12%–16%. These improvements, statistically significant at the 99% level, are comparable to the total improvement in normalized NHC official 24-h forecast error occurring over the, past 20–25 years.

Abstract

A three-dimensional analysis of temperature and relative humidity in the environment of Hurricane Debby (1982) has been completed. Observations from Omega dropwindsondes (ODWs) within 1000 km of the storm have been combined with rawinsondes over the continental United States and the Caribbean and with observations from surface ships and aircraft data where possible.

The temperature and relative humidity analyses, together with wind analyses from a previous study, form a dataset that can be used an an initial condition in a multilevel prognostic model when combined with analyses over area larger than our analysis domain. In this paper a series of diagnostic tests has been applied to the dataset to evaluate its performance without using a prognostic model. These tests include horizontal maps of the moist convective instability, calculation of the heat and moisture budgets in the vicinity of Bermuda, which was 350 km to the northeast of the storm center, and diagnosis of precipitation from these budgets and from the Arakawa-Schubert cumulus parameterization.

Results show that the horizontal distribution of moist convective instability is strongly affected by the low-level moisture field upstream of the main inflow region to the storm. The total surface heat flux, estimated with a bulk aerodynamic method, matches the vertically integrated eddy flux of moist static energy to within observational errors. Precipitation estimates from the budgets give rates of approximately 20 mm day⁻¹, which are consistent with an estimated rate from radar. Partition of the rainfall rate into convective scale and resolvable scale (stratiform) shows about equal contributions.

Our results lead us to believe that, within the limitations determined by the horizontal distribution of the observations, the final dataset for Hurricane Debby provides a realistic depiction of the various physical processes that were occurring in Debby's environment. Future work will include data sensitivity experiments with a three-dimensional forecast model.

Abstract

A set of nine synoptic-flow cases, incorporating Omega dropwindsonde observations for six tropical storms and hurricanes, is used to deduce the three-dimensional distribution of potential vorticity (PV) that contributed to the deep-layer mean (DLM) wind that steered the cyclones. A piecewise inversion technique, the same as that previously applied by Shapiro to Hurricane Gloria of 1985, is used to derive the DLM wind induced by pieces of anomalous PV restricted to cylinders of different radii centered on each cyclone. The cylinder of PV that induces a DLM wind that best matches the observed DLM wind near the center of each cyclone is evaluated.

It is found that the results can be loosely placed into two categories describing the spatial scale of the PV anomalies that influenced the cyclone’s motion. Four of the cases, including Hurricane Gloria, had “local” control, with a good match (to within ∼40%) between the observed DLM wind near the cyclone center and the DLM wind attributable to a cylinder of PV with a given radius ⩽1500 km. Further decomposition of the PV anomaly into upper (400 mb and above) and lower levels (500 mb and below) indicates the dominance of upper-level features in steering two of the cyclones (Hurricanes Gloria of 1985 and Andrew of 1992), while Hurricane Debby of 1982 was steered by more barotropic features. These results supplement those found in other studies.

Five of the cases, by contrast, had “large-scale” control, with no cylinder of radius ⩽2000 km having a good match between the induced and observed DLM wind. Hurricanes Emily of 1987 and 1993 fell into this category, as did Hurricane Josephine of 1984. Implications of the results for guiding in situ wind measurements to improve hurricane track forecasts are discussed.

Abstract

The validity of the traditional balance approximation for the asymmetric flow above the boundary layer generally in hurricanes is examined here. Scaling considerations of the divergence equation show that the validity of the balance approximation hinges on the smallness of the nondimensional product (δ ^′ _n / ζ ^′ _n )×(n υ / η r). The first term represents the ratio of asymmetric horizontal divergence to asymmetric vertical vorticity for azimuthal wavenumber n, while the second term represents a Rossby number based upon the azimuthal mean tangential wind and absolute vertical vorticity of the hurricane vortex. Wind observations of Hurricane Gloria (1985) indicate that this product is not at all small in the near-vortex region (several hundred kilometers beyond the radius of maximum tangential winds) where asymmetric convergence forced by surface friction and cumulus convection is typically large. Although the Gloria observations represent only a single case, there are dynamical reasons to expect this product to be O(1) just above the hurricane boundary layer in steadily translating hurricanes. The meteorological relevance of these results to the problem of balance dynamics in hurricanes is briefly discussed.

Abstract

The major sources of error in Omega-derived wind estimates are examined and illustrated. Sample dropwindsondes and local Omega signals are used to illustrate the effects of several types of phase propagation anomalies. A stationary test sonde and synthetic Omega phases are used to determine the accuracy of three Omega phase-smoothing algorithms and their associated error estimates and to determine the impact of base station motion for sondes released from aircraft.

Omega windfinding errors can be classified as either “internal” or ”external” errors. Internal errors are associated with signal quality and transmitter-sonde geometry, while external errors are caused by anomalous phase propagation. Estimates of wind error (wind uncertainties) are provided by the equations of Omega windfinding. These uncertainties, however, estimate only the effects of internal errors. Precise assessment of errors caused by anomalous phase propagation requires the measurement of phase data by a stationary receiver. Such measurements show that errors from external sources range from about 1 m s⁻¹ for diurnal changes in ionospheric height to 20–30 m s⁻¹ for sudden ionospheric disturbances. Methods for dealing with these problems in sonde postprocessing are described.

Data from a stationary test sonde show that the effect of aircraft maneuvers on real-time Omega wind estimates is substantial; during turns, errors in real-time wind estimates increase by over 50%. The comparison of phase-smoothing algorithms shows that cubic-spline smoothing produces wind estimates 20–50% more accurate than those obtained with other methods. Hence, it is recommended that this smoothing algorithm be used in dropwindsonde postprocessing. It is estimated that such postprocessing will reduce errors by 60% during aircraft turns and by 30% at other times.

The National Center for Atmospheric Research (NCAR), in a joint effort with the National Oceanic and Atmospheric Administration (NOAA) and the German Aerospace Research Establishment, has developed a dropwindsonde based on the Global Positioning System (GPS) satellite navigation. The NCAR GPS dropwindsonde represents a major advance in both accuracy and resolution for atmospheric measurements over data-sparse oceanic areas of the globe, providing wind accuracies of 0.5–2.0 m s⁻¹ with a vertical resolution of ~5 m. One important advance over previous generations of sondes is the ability to measure surface (10 m) winds. The new dropwindsonde has already been used extensively in one major international research field experiment (Fronts and Atlantic Storm Track Experiment), in operational and research hurricane flights from NOAA's National Weather Service and Hurricane Research Division, during NCAR's SNOWBAND experiment, and in recent CALJET and NORPEX El Niño experiments. The sonde has been deployed from a number of different aircraft, including NOAA's WP-3Ds and new Gulf stream IV jet, the Air Force C-130s, NCAR's Electra, and a leased Lear-36. This paper describes the characteristics of the new dropwindsonde and its associated aircraft data system, details the accuracy of its measurements, and presents examples from its initial applications.

In 1997, the Tropical Prediction Center (TPC) began operational Gulfstream-IV jet aircraft missions to improve the numerical guidance for hurricanes threatening the continental United States, Puerto Rico, and the Virgin Islands. During these missions, the new generation of Global Positioning System dropwindsondes were released from the aircraft at 150–200-km intervals along the flight track in the environment of the tropical cyclone to obtain profiles of wind, temperature, and humidity from flight level to the surface. The observations were ingested into the global model at the National Centers for Environmental Prediction, which subsequently serves as initial and boundary conditions to other numerical tropical cyclone models. Because of a lack of tropical cyclone activity in the Atlantic basin, only five such missions were conducted during the inaugural 1997 hurricane season.

Due to logistical constraints, sampling in all quadrants of the storm environment was accomplished in only one of the five cases during 1997. Nonetheless, the dropwindsonde observations improved mean track forecasts from the Geophysical Fluid Dynamics Laboratory hurricane model by as much as 32%, and the intensity forecasts by as much as 20% during the hurricane watch period (within 48 h of projected landfall). Forecasts from another dynamical tropical cyclone model (VICBAR) also showed modest improvements with the dropwindsonde observations. These improvements, if confirmed by a larger sample, represent a large step toward the forecast accuracy goals of TPC. The forecast track improvements are as large as those accumulated over the past 20–25 years, and those for forecast intensity provide further evidence that better synoptic-scale data can lead to more skillful dynamical tropical cyclone intensity forecasts.