Department of Botany and Zoology

Permanent URI for this community

https://scholar.sun.ac.za/handle/10019.1/219

Browse

Browsing Department of Botany and Zoology by browse.metadata.advisor "Beckett, Heath"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Physical patterns and post-hoc measures of wildfire behaviour in grassland
    (Stellenbosch : Stellenbosch University, 2022-12) Danckwerts, Matthew; Midgley, Guy F.; Beckett, Heath; Stellenbosch University. Faculty of Science. Dept. of Botany and Zoology.
    ENGLISH ABSTRACT: Wildland fires are an intrinsic feature of seasonal grassland and savanna ecosystems, and are thus guaranteed to occur at some frequency in these systems. In addition, fire managers and practitioners in southern Africa use fire as a management tool in these rangeland systems to achieve an array of management goals for ecological, agricultural, and safety purposes. The occurrence of uncontrolled fires potentially attributable to human agency and accompanied by damage to property and associated litigation is an inevitable outcome in these wildland fire systems. The fields of fire litigation and forensics are, however, under-developed for such wildfire systems in which the inference of deliberate, accidental, and natural contributors to events is critical. Post-hoc physical indicators of wildfire pattern and spread, as well as fire behaviour prediction models are used by experts to determine origin, spread, and behaviour of fires in forensic wildland fire investigations conducted in contemplation of litigation. A review of post-hoc wildfire behaviour indicators relevant to forensic wildfire investigations, detailing application, reliability, and limitations, identified a suite of indicators of potential value in such circumstances. Selected post-hoc indicators and prediction models of fire behaviour were then tested for their suitability and utility across a range of fuel and weather conditions in montane grassland systems of the Greater Winterberg region, Eastern Cape, South Africa. Real-time fire behaviour data were gathered from 143 fires across seven sites and compared with corresponding post-fire indicator expression and prediction model outputs. Slope proved to be a critical driver of fire behaviour in the mountainous Eastern Cape grassland systems. Fire behaviour models (encompassing a wide range of fuel model configurations constructed for southern African grassland systems), did not suitably predict (or retrodict) actual fire conditions, generally under-predicting rate of spread, fire intensity, & flame length values. McArthur models, derived from field-based data, proved to be the most suitable predictors of fire behaviour in Eastern Cape montane grasslands. It is recommended that current fuel model and input assumptions which exist for southern African grassland and savanna systems be re-evaluated. Residual ash colour, curling, spalling, and residual ash organic carbon content were found to be unreliable forensic indicators of wildfire pattern and spread. Foliage freeze, cupping of grass tussocks, consumption depth of grass tussocks, amount of residual unburnt plant litter, protection, sooting, & staining were considered reliable, but in some cases may be site specific and need to be applied in conjunction with each other to draw accurate and reliable conclusions of fire pattern and spread. The presence of undercut culms and leeside charring on pole-type fuels are almost unequivocal indicators of back fires and wind direction, respectively. These findings suggest that certain post-hoc wildland fire pattern and spread indicators must be re- evaluated, and practitioners are advised to adopt an adaptive approach to indicator interpretation, applying fire behavioural science of the processes and drivers of indicator formation and expression, while collectively incorporating several indicators in conjunction with each other before forming conclusions on origin, cause, and spread of ‘runaway’ wildfires.