Complete our Visitors’ Survey and WIN! Click here.
Spatial-statistical characterisation of wind fields over complex terrain for bushfire modelling applications.
By statistically characterising wind fields over complex terrain, we can develop a better understanding of wind fields and enable the construction of probabilistic wind models for fire prediction systems.
Rachael Griffiths (PhD Candidate, UNSW Canberra)
Jason Sharples (UNSW), Leesa Sidhu (UNSW) and Graham Thorpe (Vic U)
The aim of this research is to statistically characterise wind fields over complex terrain in order to develop a better understanding of wind fields and enable the construction of probabilistic wind models for fire prediction systems.
Given the inherent uncertainties in bushfire modelling, future developments are heading towards a probabilistic approach to prediction, utilising ensemble methods.
A probabilistic characterisation of wind fields will not only better suit these new models, but is also more appropriate in capturing the uncertainties in wind flow than the deterministic schemes that are currently used.
This PhD project leads on from the work of Sharples et al. (2010) on wind characteristics over complex terrain, which characterised wind fields using bivariate distributions representing wind response at a number of points in the landscape.
Statistically, there is much work to be done to characterise these distributions with physically interpretable parameters and this PhD project also addresses a number of technical issues surrounding the statistics of surface and distribution comparisons.
In 2014, wind data was collected in Flea Creek Valley, Brindabella National Park. Using this data to construct bivariate wind response distributions, we have seen that lee-eddies are persistent in areas where current wind modelling techniques (used in fire modelling) do not capture them.
Through initial comparison with data collected along the same valley transect in 2007, it is also evident that the wind responses have remained very similar over time, despite years of re-growth. Unfortunately, we are unable to quantitatively capture the changes in vegetation that have occurred across Flea Creek valley over the past decade.
A wind response study at the National Arboretum would enable the detailed quantification of vegetation growth and structure as well as terrain features that lead to variations in wind flows. This data can then be used in developing the probabilistic wind field models we hope to achieve from this PhD project.
This work will begin to enrich fire modelling systems with greater understanding of inherent variability and uncertainties in wind flow (and fire behaviour) that are currently not captured.
The Arboretum study would help to answer two key research questions:
Two further research questions can be investigated through this Arboretum study:
Eleven portable automatic weather stations (PAWS) will be used to collect a range of data, including wind speed and direction, temperature, relative humidity, rainfall, solar radiation and UV. Data are collected at 1 minute intervals. Figure 1 shows one PAWS set up; the anemometer sits at 5m, with the rain bucket and sensor array at 2m above ground. The footprint consists of small rectangular base and three guy ropes.
Figure 1 Portable Automatic Weather Station (PAWS)
It is proposed that the stations are set up in an array within and adjacent to the pine forest block at the National Arboretum, near the lookout point, such that the four research questions can be addressed:
Figure 2 gives an example of station design throughout the pine forest plot.
Figure 2 Potential experimental design with PAWS at the National Arboretum. Red circles indicate 'ridgeline' stations.
In order to collect the full distribution of wind response, it is important that the stations are situated at each site for a reasonable length of time and capture the full range of seasonal flows.
It is expected that the stations could be set up in April 2015, with data collected through to the early summer (November 2015). If possible, it would be ideal to collect wind data throughout the summer months as well; this would give an insight into wind behaviours directly relevant to the fire season.