Home | About Us | Contacts | Help | Feedback | SEARCH Global | Australia | NSW | Vic. | Qld | WA | SA | Tas. | ACT | NT | Ant. | Weather & Warnings | Hydrology | Climate | Numerical Prediction | About Services | Learn About Meteorology | Registered User Services |

Global Guide
Main Menu

Global Guide to Tropical Cyclone Forecasting:
CHAPTER 8: NUMERICAL TRACK PREDICTION MODELS

8.3 Operational Numerical Models

Operational numerical models provide track forecast guidance that is comparable with other techniques. These systems also have the potential for markedly increased forecast accuracy as current research is transformed into model improvements. The main features of a number of current systems are provided here to help forecasters interpret the forecast output. A detailed survey is provided by Elsberry (1993), and current information from forecast centres is provided by WMO (1993). The information is valid as of mid-1993. Because of the rapidly changing nature of operational numerical modelling, the information in this sections will change and readers should consult the relevant literature for current information.

8.3.1 Global Models

Global models have attained resolutions where the outer circulations of small-scale systems such as tropical cyclones are resolved. Table 8.1 lists the characteristics of several widely available global models in 1993. Note that all centres except the UKMO use spectral models.

Parameterisation of physical processes in various degrees of detail are carried out by all global models (Table 8.2). All models include cumulus-convection parameterisation, large-scale condensation, surface fluxes, a detailed planetary boundary layer, diagnostic clouds, radiative transfer, ground hydrology and horizontal and vertical diffusion. Improvements in the parameterisation of physical processes have had a significant impact on the performance of the models in the tropics (Tiedtke et al., 1988) and for tropical cyclones (Bender et al., 1987; Bender and Kurihara, 1987; Puri and Miller, 1990a,b).

All centres also use some form of data assimilation. ECMWF, JMA and NMC use intermittent assimilation (Section 8.2.3) with an analysis performed at 6-h intervals. The UKMO uses continuous data assimilation with observations inserted as they occur.

There are no comprehensive evaluations comparing all global models on homogeneous data sets, but the impression from recent studies (Goerss and Jeffries, 1993; Fiorino et al., 1993: Miller, 1992; Shun, 1992; Muroi and Sato, 1992) is that all global models are producing useful forecast guidance on tropical cyclone tracks (eg Table 8.3). Only the UKMO provides track forecasts in routine, twice daily tropical bulletins during the southern hemisphere tropical season to WMO RA5 members (Table 1.1).

8.3.2 Limited Area Models

Even with the improved performance for global models, most operational centres continue to use limited area models for tropical cyclone prediction, especially because of the higher resolution that can be obtained and the earlier availability of forecasts. In this section, we describe the basic features of limited area tropical cyclone prediction models used for tropical cyclone prediction at the JMA, FNOC, US NMC and Australian NMC (Tables 8.4, 8.5).

Detailed information on TYM is given by Iwasaki et al. (1987). An important component of the initial condition is that bogussing is used to define the circulation in the region around the reported position of the cyclone. Whenever a cyclone is expected to affect the area of responsibility, two 60h forecasts are made operationally with TYM, based on initial data at 0000 and 1200 UTC.

A number of case studies using TYM and other JMA models (Ueno, 1989) indicate that the size of the forecast domain, horizontal resolution and the convection scheme have a significant impact on the cyclone track forecasts. Moreover, studies indicate that careful specification of the asymmetric part of the bogus is important for track forecasts. TYM contains a clear tendency for cyclones to drift poleward relative to observed tracks.

As with the TYM, an important component of NORAPS is the use of bogussing in developing a tropical cyclone structure in the initial analysis. A notable feature is the use of a 12-h data assimilation cycle.

A distinguishing feature of the QLM is that the time derivative following the horizontal motion of an air parcel is evaluated by using the so-called quasi-Lagrangian method. The total acceleration and the nonlinear advective terms are therefore evaluated to a higher order of accuracy. As with the models at other centres, a bogus vortex is used in the QLM with the initial grided data (without the prescribed vortex) given by NMC's Aviation (AVN) operational initialised analysis. The recent performance of QLM in forecasting North Atlantic tropical cyclones (Table 8.6) provides a general indication of the capacity of regional models generally.

TAPS also uses bogus moisture data and a bogus tropical cyclone, which is inserted into the analysis as observations. Another feature is the use of dynamical nudging, as described in Section 8.2.3, during which diabatic heating estimated from Japanese GMS imagery is used to develop realistic vertical motion in the initial state.

Contents Chapter 9