Improved Hurricane Track Forecasting from the Continuous Assimilation of High Quality Satellite Wind Data

• Published in: Monthly weather review 1998-05, Vol.126 (5), p.1248-1257
• Main Authors: Leslie, L M, LeMarshall, J F, Morison, R P, Spinoso, C, Purser, R J, Pescod, N, Seecamp, R
• Format: Article
• Language: English
• Subjects: Hurricane tracking; Hurricanes
• ISSN: 0027-0644; 1520-0493
• DOI: 10.1175/1520-0493(1998)126<1248:ihtfft>2.0.co;2

Despite recent improvements in the accuracy of hurricane track forecasts, mean position errors still remain unacceptably large. For example, recurvature is captured poorly by forecast models and can produce excessively large position errors. This study addresses the problem of hurricane track forecasting in three ways. First, the initial conditions for the forecast model are augmented by a dense coverage of high spatial and temporal resolution satellite-derived wind vectors. Second, to gauge the extent to which this additional four-dimensional detail of the atmospheric structure can be exploited, three distinct types of data assimilation methods are examined. These are 1) conventional (intermittent, cycled) 6-h assimilation, 2) nudging over a 12- or 24-h period up to the initial time, and 3) recently developed barotropic and four-dimensional variational assimilation schemes, also over a 12- or 24-h period. The nudging and variational methods are continuous assimilation procedures and incorporate satellite-derived winds, typically at 6-h frequencies, but up to hourly frequencies in trials over the Australian region. Over the Atlantic basin, only a 6-h frequency of high-density satellite-derived wind vectors was available for the 1995 season. Third, a very high-resolution (15 km) semi-implicit, semi-Lagrangian model provided forecasts out to 72 h. The present study describes the direct application to the Atlantic basin, for the specific case of Hurricane Opal, of data assimilation and prediction procedures developed for tropical cyclones over the Australian Pacific basin. A series of forecasts was made, from the two initial times 0000 and 1200 UTC 2 October 1995, respectively. In these cases, where CLIPER (climatology and persistence) and other conventional forecast guidance was poor, the nudging and variational assimilation procedures, which were those that best utilized the high spatial and temporal resolution satellite-derived winds, produced greatly improved forecasts.