Assimilation of Tropical Cyclone Observations: Improving the Assimilation of TCVitals, Scatterometer Winds, and Dropwindsonde Observations

Abstract

Abstract The standard statistical model of data assimilation assumes that the background and observation errors are normally distributed, and the first- and second-order statistical moments of the two distributions are known or can be accurately estimated. Because these assumptions are never satisfied completely in practice, data assimilation schemes must be robust to errors in the underlying statistical model. This paper tests simple approaches to improving the robustness of data assimilation in tropical cyclone (TC) regions. Analysis–forecast experiments are carried out with three types of data—Tropical Cyclone Vitals (TCVitals), DOTSTAR, and QuikSCAT—that are particularly relevant for TCs and with an ensemble-based data assimilation scheme that prepares a global analysis and a limited-area analysis in a TC basin simultaneously. The results of the experiments demonstrate that significant analysis and forecast improvements can be achieved for TCs that are category 1 and higher by improving the robustness of the data assimilation scheme.