Mesoscale wind fields have been determined for a mature hurricane with high spatial and temporal resolution, continuity, and coherency. These wind fields, near the tropopause in the inner core and at low levels inside the eye, allow the evolution of mesoscale storm features to be observed. Previously, satellite-derived winds near hurricanes have been determined only at some distance from the eye over a typical time period of 1–2 h. Hurricane reconnaissance aircraft take 30 min to 1 h to complete an inner-core pattern. With the long observation periods of these previous methods, steady-state conditions must be assumed to give a complete description of the observed region.
With the advent of 1-min interval imagery, and fourfold improvement of image dynamic range from NOAA's current generation of GOES satellites, there is a new capability to measure inner-core tropical cyclone wind fields near the tropopause and within the eye, enabling mesoscale dynamical processes to be inferred. These measurements give insights into the general magnitude and structure of the hurricane vortex, along with very detailed measurements of the cloud-top wind's variations in response to convective outbursts. This paper describes the new techniques used to take advantage of the GOES satellite improvements that, in turn, allowed the above innovations to occur.
The source of data for this study is a nearly continuous 12-h sequence of 1-min visible images from NOAA GOES-9 on 6 September 1995. These images are centered on Hurricane Luis with maximum winds of 120 kt (CAT4) when it was 250 km northeast of Puerto Rico. A uniform distribution of long-lived cirrus debris with detailed structure is observed in the central dense overcast (CDO), which has been tracked using the 1-min images. The derived wind field near the tropopause at approximately 15 km in the CDO region has a strong closed circulation with speeds up to 25 m s−1, which pulses in response to the convective outbursts in the eye wall. Cloud displacements are computed at every pixel in every image, resulting in a quarter-million u–v winds in each of 488 hurricane images observed at 1- to 4-min intervals over 12 h. For analysis and presentation, these ultradense wind fields are reduced to 8- or 16-km grids using a 7-min time base by smoothing displacement vectors in space and time.
Cloud structures were tracked automatically on a massively parallel processing computer, but with manual spot-checking. Manual tracking has been used to follow CDO structure over long time periods, up to 90 min for a small test sample. Cloud tracking for the wind fields presented here is accomplished using a Massively Parallel Semi-Fluid Motion Analysis (MPSMA) automatic technique. This robust deformable surface-matching algorithm has been implemented on the massively parallel Maspar supercomputer. MPSMA automatic tracking typically follows a feature for 7 min. For this time base the error of these winds is estimated to be 1.5 m s−1. However, systematic navigation and height assignment errors in the moderately sheared hurricane environment must still be considered. Spatial and temporal smoothing of the wind field have been performed to reduce systematic navigation errors and small-scale turbulent noise. The synthesis used here to compute the wind fields gives an order of magnitude reduction in the amount of data presented compared to the amount of data processed. Longer tracking could give higher accuracy but would smooth out the smaller-scale spatial and temporal features that appear dynamically significant.
The authors believe that the techniques described in this paper have great potential for further research on tropical cyclones and severe weather as well as in operational use for nowcasting and forecasting. United States and foreign policymakers are urged to augment the GOES, GMS, FY2, and Meteosat geostationary satellite systems with dual imaging systems such that 1-min observations are routinely taken.
*NASA/Goddard Space Flight Center, Greenbelt, Maryland.
+Department of Computer and Information Science, University of Delaware, Newark, Delaware.
#Hurricane Research Division, NOAA/Atlantic Oceanographic and Meteorological Laboratories, Miami, Florida.
@Laboratory for Atmospheres, NASA/Goddard Space Flight Center, Greenbelt, Maryland.