Objectively estimating tropical cyclone intensity and wind structure using the advanced microwave sounding unit

Abstract

Estimating tropical cyclone (TC) intensity and structure is becoming increasingly important in light of population expansion along coastal regions. The two most commonly used techniques for estimating TC intensity, the Dvorak Technique and the Objective Dvorak Technique (ODT), utilize visible and infrared satellite imagery. However, both have limitations, as do observing techniques of TC wind structure. Satellite-borne passive microwave radiometers provide an opportune alternative for near real-time assessments of TC maximum sustained winds and wind radii. The first Advanced Microwave Sounding Unit (AMSU), aboard the NOAA-15 polar orbiting satellite, is the first NOAA instrument with sufficient resolution to do so. In this study, data derived from AMSU temperature, pressure, and wind retrievals are used to make objective intensity and wind radii estimates for tropical disturbances in the Atlantic and East Pacific basins. To approximate TC maximum sustained winds and azimuthally averaged wind radii of 34, 50, and 64 kt winds, algorithms are developed via correlations and multiple linear regressions from AMSU data from the 1999 tropical season; they then are tested independently on the 2000 tropical season data Additionally, the AMSU-derived estimates of the azimuthally averaged wind radii are used with a modified Rankine vortex model to assess the wind radii asymmetrically, specifically in the northeast (NE), northwest (NW), southeast (SE), and southwest (SW) quadrants of the TC. Validation data are from the National Hurricane Center (NHC) best track data for the intensity estimates, and from the NHC operational forecast advisories for the average and asymmetric wind radii estimates. Results show the objective AMSU algorithm is comparable to the ODT for estimating TC intensity in the Atlantic, with a root mean square error (RMSE) of 13 kts. The RMSE increases slightly to 16 kts for both basins combined. In general, the AMSU algorithm has a tendency to over (under) approximate the intensity of weak (strong) TCs. For the AMSU-estimated azimuthally averaged 34, 50, and 64 kt wind radii, the mean absolute errors (MAE) are 16, 17, and 8 nautical miles (nm), respectively. With respect to the average radii of each, these correspond to errors of 14.4 percent, 24.6 percent, and 17.8 percent. As with the intensity estimation algorithm, there is a tendency toward over (under) estimation of small (large) azimuthally averaged wind radii by the AMSU. Additionally, the wind radii estimates in the NE, SE, SW, and NW quadrants capture the asymmetric structure well, generally comparing favorably with the NHC operational advisory estimates. In some cases, the AMSU estimates may even be superior to NHC estimates, especially in the Eastern Pacific. Finally, the 1999 and 2000 AMSU data are combined to refine the TC intensity and azimuthally averaged wind radii estimation algorithms. The two-year based algorithms currently are being tested on AMSU data received from the Atlantic and East Pacific basins during the 2001 tropical season.