|
|
CONFERENCE PROCEEDINGSAustralian Bushfire Conference, Albury, July 1999 |
copyright 1999 |
The forecasting of wind changes in support of bushfire and controlled burning operations is a major forecast challenge for meteorologists. In the author’s professional lifetime these forecasts have evolved from being based on almost entirely land-based surface and upper-air observations, with subjective extrapolation using conceptual models of atmospheric development patterns, to having available much more comprehensive observational and guidance data-bases. These now include hourly geostationary satellite imagery and 6-hourly soundings from orbiting satellites, and guidance from increasingly complex and accurate numerical weather prediction (NWP) model forecasts.
NWP models are now approaching the resolutions and accuracies needed to explicitly resolve the structure and evolution of wind change lines and topgraphically and thermally forced mesoscale circulation systems, and to forecast these systems with some accuracy. This offers the opportunity for more detailed and accurate forecasts, and the opportunity to gain greater insight into the structure and mechanisms of circulation systems at scales much finer than those of the observational network. The challenge for the model developers is to develop techniques for initialising mesoscale models at their scales of resolution, and to properly validate, refine, and develop these models, while the challenge for forecasters is to learn how to interpret the detail contained in these model forecasts.
In this talk I will briefly describe the components which comprise an operational numerical modelling system, the Bureau of Meteorology’s current operational suite of NWP models, planned developments, and present examples of the output from very high resolution modelling studies of some significant fire weather events to demonstarte the current state and future potential of these systems to forecast mesoscale weather systems important to fire agencies.
The components of a NWP system comprise:
A NWP model numerically integrates the momentum, thermodynamic, and moisture equations to produce a forecast of these variables at a future time, usually on a 3-dimensional grid. Most models include what are termed the "dynamics" – that component of the model which integrates the equations – and the "physics" in which those processes (usually of smaller scale than resolvable by the model grid) that are not explicitly included in the forecast equations are "parameterised" in order to include their effects. These parameterised processes include radiation and its interaction with the earth and atmosphere, convection, surface processes, and turbulent processes. In order for an NWP model to produce an accurate forecast, not only do the model numerics need to be well formulated, and the physical parameterisations represent as accurately as possible the processes occurring in the atmosphere, but also the initial state must be defined as accurately as possible. This requires not only that the three dimensional state of the momentum and thermodynamics of the atmosphere be specified, but also requires an accurate depiction the Earth’s topography, surface characteristics (roughness, vegetation type and fraction, soil type, moisture availability, snow cover, sea-ice extent, etc) and sea surface temperature.
The data input typically comprise surface observations from land and ocean based platforms, rawindsonde profiles, temperature, humidity and wind data observed or diagnosed from sensors on board orbiting and geostationary satellites, and observations from aircraft. These data are not on the regular grid of the model, but are irregularly distributed in time and space. In order to interpolate this information onto the regular grid required by the forecast model, a "data assimilation" step is carried out. During this process the model state is corrected to "fit" the observations to within their observational error, so that the resulting fields retain the balance and information content of a short-term forecast based on previous information, but also reflect the information content of the most recent observations. (In this context, the short-term forecast used to provide the guess field can be regarded as data.) Observations are subject to complex quality control procedures, usually including gross error checks followed by some form of "buddy checking". The systems also include techniques for elimination of redundant data (highly correlated observations) and complex data selection algorithms so that dense observations of one type do not overwhelm complementary data of another type.
Unfortunately, the data available for use in operational NWP systems is distributed irregularly in space and time. The density of surface observations varies from land to sea, data from orbiting satellites arrives in dense swathes, other data are tied to shipping or aircraft routes, or the availability of suitable tracers (eg Cloud Drift Winds, CDW). As the distribution of data, and types of data, are so inhomogeneous, it is necessary to include mass-wind coupling in the analysis so that the mass field is adjusted in the presence of wind-only data, and vice-versa. This ensures that information is inserted in a balanced way into the forecast model, and thus retained during the model’s integration. The appropriate statistics for background error correlations and observational error characteristics must also be included.
Until recently, most major NWP centres used forms of multivariate statistical interpolation to correct the short-term forecasts (also known as the background field), with data insertion typically every 6 hours (intermittent data insertion). Such a system is used in the Bureau of Meteorology. Recently, though, there has been a trend towards 4-dimensional variational analysis. In such a system, the model state is variationally adjusted to minimise the difference between the model state and the observations, subject to the strong constraint that the adjusted state is consistent with the model equations. Variational analysis has the advantage that variables such as radiance in a given spectral band, as is measured by satellite, can be treated as an observation, rather than needing to first have a temperature "retrieved" from these data, while the addition of the time dimension allows the time-tendency of the model state to match the implicit time tendency of the data.
Initialisation schemes are used after the data insertion phase for intermittent insertion assimilation systems to "balance" the analysis fields – that is, to remove dynamic inconsistencies between the mass and the wind fields which may lead to spurious gravity waves being generated early in the model forecasts. The removal of these gravity waves is desirable so that subsequent data insertions are not adversely affected, and to reduce the spurious prediction of rainfall early in the forecast period. Such schemes can be normal mode initialisation schemes, use digital filters or use some form of Newtonian relaxation . The variational assimilation schemes include constraints based on the equations within the model equations, and so initialisation is integral to these analysis schemes.
The Bureau of Meteorology operates a suite of NWP models ranging from a global system, GASP, which produces forecasts to 7-days twice per day, through the regional models for mid-latitudes (LAPS) and the tropics (TLAPS) (twice daily to 48 hours), and the high resolution meso-LAPS system (twice daily to 36 hours). The regional systems have their lateral boundary conditions specified by the global model forecast, while the mesoscale system is nested inside the regional system. By the second half of 1999, the regional systems will operate on a regular latitude-longitude grid of 0.375o, with 29 levels from the surface to around 20 km, while the mesoscale system will cover all of Australia with a grid resolution of 0.125o. In addition, the Bureau of Meteorology Research Centre is routinely running versions of the same model with a grid spacing of 0.05o, centred over Sydney and Melbourne.
Since the meso-LAPS model commenced its twice-daily forecasts with a resolution of 0.25o in 1997, forecasts of the structure and timing of wind shear lines has shown a leap forward. Part of the gain has been in the improved quality of the model numerics, but the role of more complex and realistic treatment of land surface schemes, and simply the fact that the finer grid resolution enables the model to resolve smaller scale features have aided these advances. Further, the presentation of low-level wind flow and temperature field forecasts, rather than the more traditional mean-sea-level pressure forecasts, has allowed improved interpretation of model output, as will be shown for the case of the shear line across western Victoria for 15 December 1995. Many very accurate forecasts of frontal passage, such as that of 6 March 1996 are recorded.
These forecasts also show the influences of land-sea heating contrast, and, providing, the scale of the systems are resolvable on the model’s grid, and always subject to the caveat that the initial conditions are sufficiently accurate, then forecasts of systems such as the coastal front which affected the Berringa fire in February 1995 can be forecast remarkably accurately. This forecast demonstrated the scale and structure of this wind change in a way that could not be done from the observational network, and provided a framework in which an "anomalous" observation could be interpreted. Sensitivity tests varying the soil moisture parameter in the model also showed that such a wind change would be more likely to occur in a drought year.
There are a wide range of frontal systems which can occur, and while the larger-scale systems, with strong control from the synoptic-scale flow, are generally fairly well forecast by the 0.25o meso-LAPS model, there are other situations where the forecasts of the wind changes are less realistic. Such a case is that of the cool change that affected the Dandenong Ranges bushfire on 21 January 1997. In this case the meso-LAPS forecast was unable to resolve the sharpness or phase of the coastally-surging cold front However, research simulations using the same broad-scale initial conditions with a 5km resolution version of the model produced a far more realistic simulation of the event. This indicates that in this case, the finer grid resolution was able to resolve the sharp thermal gradients that developed along the Victorian coastline, and the small-scale, but intense circulation systems that then developed.
As the implementation phase of the current suite of operational upgrades is completed, work is already underway to develop techniques for the detailed validation of the surface parameter forecasts from the 0.125o and 0.05o mesoscale models, as verification techniques applicable to synoptic scale forecasts are not necessarily applicable to the smaller scale systems we are now forecasting. This will provide forecasters with measures of uncertainty in our point forecasts of wind and temperature, allow monitoring of longer-term trends in mesoscale NWP model accuracy, and feed back into model development. Increasing efforts are being made to improve the initial state specification of these mesoscale models via alternative data insertion strategies and the incorporation of new data sources. Models are being improved by the inclusion of improved representation of physical processes, improved numerical techniques, and from the incorporation of improved topography and land surface data sets. All these will lead to more accurate forecast guidance being provided to forecasters.
Published by School of Environmental & Information Sciences Charles Sturt University