Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Pays Bas → Human influence on the spatial pattern of fire regime in the South African savannah

Wageningen University (2009)

Human influence on the spatial pattern of fire regime in the South African savannah

Hojas Gascón, L.

Titre : Human influence on the spatial pattern of fire regime in the South African savannah

Auteur : Hojas Gascón, L.

Université de soutenance : Wageningen University

Grade : Master of Science (MS) 2009

Fire is a natural factor in the savanna biome, whereby temporal and spatial occurrence is largely shaped by the climatic conditions. However, in the African savannas it is believed that human-induced fires are the main source of ignition. The objective of this research is to assess the influence of the spatial pattern of human development in fire regime in the South African savannah. The dependent variables, fire frequency and fire size, were extracted from the latest spatial burned area data set produced for seven global fire seasons (2000-2007) with a moderate spatial resolution (1km2) and a high temporal resolution (daily intervals). The explanatory variables were calculated with the Euclidean distance or at distances ranging from 1 up to 20 km from the pixels with fire presence. Moreover, two types of neighbourhoods were applied in the analysis of explanatory variables, namely rectangular and annulus neighbourhoods. The data analysis involved the identification of relationships between the explanatory variables with fire frequency and fire size as a first step. Secondly, the selection of the neighborhood shape and distance of measurement of the explanatory variables that showed better correlations. Lastly, the level of impact of the explanatory variables on fire frequency and size was estimated. The correlations and regressions analysis show that, in general, human variables correlate better with fire frequency when measured with annulus neighbourhoods, while biophysical variables show better correlations with rectangular neighbourhoods. Biophysical variables (herbaceous and tree covers, and topographic roughness) appear to be better explanatory variables than human variables of fire frequency. This suggests that fire maintenance and spread is ultimately a function of fuel availability and terrain conditions. Population and cattle densities however did show high levels of significance in patterns of fire frequency. In the vicinity of urban settlements, fire frequency tends to be higher in comparison to cases in rural settlements. Urban settlements usually provide more potential ignitions and suppression measures. On the other hand, the most significant variables in explaining fire size were topographic roughness, cattle and goat densities. It can be deduced that flatness of the terrain and fuel continuity appear more important than ignitions and fuel availability in determining fire size. At long distances from roads, fires tend to be larger. Roads generally inhibit fire spread and give access to fire extinction measures. For some variables, the correlation with fire frequency and size do not vary significantly over the distance, but for others correlations improve substantially. This is the case in correlations of human variables with fire frequency, and to a lesser extent in correlations of environmental variables with fire size. Multiple regressions suggest that fire frequency is higher in areas with (i) high herbaceous vegetation cover in the surrounding area ; (ii) high tree cover and topographic roughness in the closest area ; (iii) high human and cattle densities at 20km and low goat density at 1km ; and (iv) at short distances to national roads or urban areas. Fires are larger in areas with (i) low topographic roughness in the surrounding area ; (ii) low cattle and goat densities (especially at 4km distance) ; and (iii) at large distances from roads. These models only explain 12% and 11% of the variance of fire frequency and size respectively. The low impact level of human variables suggests that fire occurrence and area burned in the South African savannah are mainly determined by natural factors.

Mots clés : fires / wildfires / fire behaviour / fire causes / savanna woodlands / spatial distribution / human behaviour / southern africa / human impact


Version intégrale (1 Mb)

Page publiée le 26 janvier 2015, mise à jour le 14 octobre 2018