The influence of topographical variability on wildfire occurrence and propagation

Thesis (DPhil)--Stellenbosch University, 2024.

ENGLISH ABSTRACT: Wildfires have increasingly become a point of concern, especially with notable incidents like the 2017 Knysna fire. These naturally occurring phenomena, despite their disruptive nature, are crucial for the sustainability of certain ecosystems. At the heart of understanding wild-fires lies the relationship between climate, vegetation, topography, and human land use, with topography standing out as a significant determinant. This thesis delves into the fundamen-tal role of topography, emphasizing its effect on the ignition, propagation, and behaviour of wildfires. Utilizing Digital Elevation Models (DEMs), the research extracts invaluable topographic data aiming to augment the understanding of wildfires, especially in mixed natural forest and fyn-bos ecosystems. Existing fire models have shown certain shortcomings, often overlooking crucial localized wind data, which has profound implications for predicting fire behaviour. By bridging this gap, the study explores the potential of computational fluid dynamics in modelling surface winds based on topography for fire research. The research systematically addresses several key objectives: Mapping the current land-scape of topography-cantered wildfire research and investigating the utility of DEM-derived surface wind in refining fire propagation models, identifying and analysing historical fire patterns to pinpoint fire refugia in the Knysna/Tsitsikhama region, employing machine learning techniques, to determine if topographic variables extracted from DEMs can antici-pate fire refugia. The findings underscore the salience of topography in wildfires. Especially significant is the role of aspect in determining fire refugia, emphasizing that a combination of multiple variables offers the most accurate insights. Machine learning, notably the XGBoost model, showcases potential in identifying critical topographical features impacting fire behaviours. Furthermore, the research sheds light on the pivotal influence of wind chan-nels, formed by topographical features, in both the inception and spread of wildfires. In summary, this thesis underscores the integral role of topography in understanding wild-fires. It charts a roadmap for future research, emphasizing the importance of high-quality validation data, a more comprehensive mapping of fire refugia, and an acknowledgement of the influence of human activity on fire regimes. By building on the methodologies and in-sights presented, there lies an opportunity to advance sustainable wildfire management so-lutions that benefit both ecosystems and human communities.

AFRIKAANSE OPSOMMING: Veldbrande het toenemend 'n punt van kommer geword, veral met noemenswaardige voor-valle soos die 2017 Knysna-brand. Hierdie natuurlike verskynsels is, ten spyte van hul on-twrigtende aard, deurslaggewend vir sekere ekosisteme se volhoubaarheid. Die kern van be-grip van veldbrande lê die verband tussen klimaat, plantegroei, topografie en menslike grondgebruik, met topografie wat uitstaan as 'n beduidende determinant. Hierdie tesis delf in die fundamentele rol van topografie, en beklemtoon die effek daarvan op die ontsteking, voortplanting en gedrag van veldbrande. Deur gebruik te maak van digitale hoogtemodelle (DEM), onttrek die navorsing onskatbare topografiese data wat daarop gemik is om ons begrip van veldbrande te vergroot, veral in gemengde natuurlike woude en fynbos-ekosisteme. Bestaande brandmodelle het sekere tekortkominge getoon, wat dikwels belangrike gelokaliseerde winddata oor die hoof gesien het, wat diepgaande implikasies het op die voorspelling van brandgedrag. Deur hierdie gap-ing te oorbrug, ondersoek die studie die potensiaal van berekeningsvloeistofdinamika in die modellering van oppervlakwinde gebaseer op topografie vir brandnavorsing. Die navorsing spreek sistematies verskeie sleuteldoelwitte aan: Kartering van die huidige landskap van topografiese galop veldbrandnavorsing. Ondersoek die nut van DEM-afgeleide oppervlakwind in die verfyning van brandvoortplantingsmodelle. Die identifisering en ontleding van historiese brandpatrone om brandtoevlugsoorde in die Knysna/Tsitsikhama-streek vas te stel. Die gebruik van masjienleertegnieke, om te bepaal of topografiese ve-randerlikes van DEM's brandtoevlugsoorde kan verwag. Die bevindinge onderstreep die opvallendheid van topografie in veldbrande. Veral betekenisvol is die rol van aspek in die bepaling van brandtoevlugsoorde, wat beklemtoon dat 'n kombinasie van veelvuldige ve-randerlikes die mees akkurate insigte bied. Masjienleer, veral die XGBoost-model, toon po-tensiaal in die identifisering van kritieke topografiese kenmerke wat brandgedrag beïnvloed. Verder werp die navorsing lig op die deurslaggewende invloed van windkanale, gevorm deur topografiese kenmerke, in beide die ontstaan en verspreiding van veldbrande. Samevattend, hierdie tesis onderstreep die integrale rol van topografie in die verstaan van veldbrande. Dit skets 'n padkaart vir toekomstige navorsing, wat die belangrikheid van hoë-gehalte valideringsdata, meer omvattende kartering van brandtoevlugsoorde, en 'n erken-ning van die invloed van menslike aktiwiteit op brandregimes beklemtoon. Deur voort te bou op die metodologieë en insigte wat aangebied word, lê daar 'n geleentheid om volhoubare oplossings vir veldbrandbestuur te bevorder wat beide ekosisteme en menslike gemeen-skappe bevoordeel.

Doctoral