Probabilistic Temporal Extension of a New Parametric Tropical Cyclone Wind Field Model

Existing parametric models of tropical cyclone wind fields have stopped short of formal statistical treatment of uncertainty, but modeling parameter noise is needed for fully probabilistic, spatio-temporal representations of wind field evolution. This talk describes the construction of a multivariate error structure for a new azimuthally asymmetric wind field parametrization. We show a solution for fitting such time series modeling despite the lack of TC wind field observations at regular time intervals. This explicit statistical modeling allows the wind field parameterization to more fully represent the natural variability of winds in risk analysis settings. Variations across an ensemble of modeled space-time wind fields reflect both the calibration data uncertainty, as well as the signal-to-noise ratio of the modeled environmental influences on wind field structure. The statistical model also enables rigorous uncertainty quantification in explorations of wind field climatology, for example in studies of climatological structural asymmetry. Finally, the error structure allows formal assimilation of surface observations in estimation of historical TC wind fields.