Research findings

The HighFire Risk Project



Citation Sharples, J.J., McRae, R.H.D. & Weber, R.O. (2009). An empirical probabilistic study of wind direction over complex terrain. 18th World IMACS/MODSIM Congress, Cairns.

An empirical probabilistic study of wind direction over complex terrain.

Author(s) Jason J. Sharples, Richard H. D. McRae & Rodney O. Weber.
Abstract Understanding spatially distributed wind fields over complex terrain is important for a variety of applications including pollutant dispersion modelling and light and recreational aviation as well as the interdependent problems of fire spread modelling and bushfire risk management. Directional changes in surface winds are particularly important in this latter context. To accurately predict the spread of a fire, models often incorporate methods to account for modifications in the surface wind field driven by interactions with the terrain. The simplest of these methods assume a single wind speed and direction over the whole region of interest. While such an assumption may be valid over flat or slightly undulating terrain and may even be able to account for large-scale fire spread patterns in more mountainous terrain, finer-scale processes that can have a significant role on fire severity and in the risk of fire escalation will not be properly represented. These include finer-scale processes such as the interaction of bushfire with channelled flows and lee-slope eddies, which can have a significant effect of the ecological and hydrological aspects of bushfire risk, for example, as well as the risk a bushfire poses to life and property.
More sophisticated methods include deriving a terrain-modified wind field using computational fluid dynamics (CFD), mass consistency or empirically derived look-up tables. These methods are inherently deterministic in nature, producing a unique terrain-modified wind field for a given ambient wind vector. While these methods can offer significant improvements in predicting the evolution of fire perimeters, they are typically more computationally intensive and can fail to recognise the nonlinear and thermal effects that the terrain can have on local winds. Moreover, such methods can be difficult to validate over large tracts of land.
In this paper we discuss an alternative method for understanding the directional aspect of wind-terrain systems. Terrain-modified winds are analysed using joint probability distributions derived from wind speed and direction data collected in the Tidbinbilla Nature Reserve in the southwest of the Australian Capital Territory. Empirical distributions of wind direction are used to identify and characterise the dominant states of the particular wind-terrain system, in this case a steep east-facing slope. By considering the modal structure of joint distributions relating the conditional probability of different terrain-modified wind responses over the spectrum of ambient (input) wind vectors, several important wind-terrain interactions are identified and discussed. By analysing the temporal characteristics of data comprising particular modes a number of particular processes including thermally forced winds, lee-slope eddies and dynamic channelling are identified. The method discussed offers a way of characterising terrain-modified winds in terms of likelihood, which may be better suited to risk-based approaches to bushfire management in high-country or mountainous landscapes. The analyses also reveal the inherently stochastic nature of the wind-terrain system, and thus raise some doubts about the suitability of deterministic approaches to modelling terrain-modified surface winds in rugged terrain.
Keywords Wind direction, complex terrain, channelling, thermal winds, lee-slope eddy, fire weather, fire spread, bushfire risk management, high-country fire.