Tropical Prediction Using Dynamical Nudging, Satellite-defined Convective Heat Sources, and a Cyclone Bogus

View More View Less
  • 1 Bureau of Meteorology Research Centre, Melbourne, Australia
© Get Permissions
Restricted access

Abstract

Some notable problems in tropical prediction have been (i) the sensitivity to, and inaccuracies in, the four-dimensional structure of parameterized convective heating, (ii) the inability of conventional data networks to adequately define tropical cyclone structures, and (iii) the so-called spinup problem of numerical models. To help overcome some of these deficiencies, a diabatic nudging scheme has been developed for the Bureau of Meteorology Research Centre (BMRC) limited-area tropical prediction system.

A target analysis for the nudging is first obtained from statistical interpolation of all observational data, using, as first-guess field, output from a global assimilation and prediction system. Tropical cyclones are optionally inserted via bogus wind observations. From 12 or 24 h prior to the base time of the forecast, the prediction model is nudged toward the target analysis. During nudging the “observationally reliable” rotational wind component is preserved and the heating from the Kuo scheme is replaced by a heating function determined from 6-h satellite-observed cloud-top temperatures. The system introduces realistic tropical cyclone structures into the initial condition, defines a vertical-motion field consistent with the satellite cloud imagery, enhances rainfall rates during the early hours of the forecast, reduces the occurrence of spurious rainfall maxima, and improves mass-wind balance and retention of cyclone circulations during the model integration.

Examples of system performance from enhanced observational datasets and from real-time forecasting are presented. Encouraging results for short-term prediction of both tropical cyclone behavior and rainfall events are documented.

Abstract

Some notable problems in tropical prediction have been (i) the sensitivity to, and inaccuracies in, the four-dimensional structure of parameterized convective heating, (ii) the inability of conventional data networks to adequately define tropical cyclone structures, and (iii) the so-called spinup problem of numerical models. To help overcome some of these deficiencies, a diabatic nudging scheme has been developed for the Bureau of Meteorology Research Centre (BMRC) limited-area tropical prediction system.

A target analysis for the nudging is first obtained from statistical interpolation of all observational data, using, as first-guess field, output from a global assimilation and prediction system. Tropical cyclones are optionally inserted via bogus wind observations. From 12 or 24 h prior to the base time of the forecast, the prediction model is nudged toward the target analysis. During nudging the “observationally reliable” rotational wind component is preserved and the heating from the Kuo scheme is replaced by a heating function determined from 6-h satellite-observed cloud-top temperatures. The system introduces realistic tropical cyclone structures into the initial condition, defines a vertical-motion field consistent with the satellite cloud imagery, enhances rainfall rates during the early hours of the forecast, reduces the occurrence of spurious rainfall maxima, and improves mass-wind balance and retention of cyclone circulations during the model integration.

Examples of system performance from enhanced observational datasets and from real-time forecasting are presented. Encouraging results for short-term prediction of both tropical cyclone behavior and rainfall events are documented.

Save