The Conversion of Total Column Ozone Data to Numerical Weather Prediction Model Initializing Fields, with Simulations of the 24–25 January 2000 East Coast Snowstorm

Abstract

Abstract Satellites are uniquely capable of providing uniform data coverage globally. Motivated by such capability, this study builds on a previously described methodology that generates numerical weather prediction (NWP) model initial conditions (ICs) from satellite total column ozone (TCO) data. The methodology is based on three principal steps: 1) conversion of TCO to mean potential vorticity (MPV) via linear regression, 2) conversion of two-dimensional MPV to three-dimensional potential vorticity (PV) via vertical mapping onto average PV profiles, and 3) inversion of the three-dimensional PV field to obtain model-initializing height, temperature, and wind fields in the mid- and upper troposphere. The overall accuracy of the process has been significantly increased through a substantial reworking of the details of this previous version. For instance, in recognition of the fact that TCO ridges tend to be less reliable than troughs, the authors vertically map an MPV field that is a synthesis of ozone-derived MPV troughs and analysis MPV ridges. The vertical mapping procedure itself produces a more physical three-dimensional PV field by eliminating unrealistically strong features at upper levels. It is found that the ozone-influenced upper-level initializing fields improve the quantitative precipitation forecast (QPF) of the 24–25 January 2000 East Coast snowstorm for two of the three (re)analyses. Furthermore, the best QPF involves ozone-influenced upper-level initializing fields. Its high threat scores reflect a superior placement, amplitude, and structure. This best QPF is apparently superior to a forecast of the same case where TCO data were assimilated using four-dimensional variational data assimilation.