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

Accueil du site → Master → Pays Bas → Analyzing wildfires and wildfire ignition types in Southern Africa using the L2JRC burned area product

Wageningen University (2009)

Analyzing wildfires and wildfire ignition types in Southern Africa using the L2JRC burned area product

Salet, F.

Titre : Analyzing wildfires and wildfire ignition types in Southern Africa using the L2JRC burned area product

Auteur : Salet, F.

Université de soutenance : Wageningen University

Grade : Master of Science (MS) 2009

Résumé
The research presented in this thesis focuses on the wildfires and wildfire ignition types in Southern Africa. The fire data used in the thesis is derived from the L3JRC product, which consists of detected burnt areas for seven fire years, from 2000 till 2006, with a moderate spatial resolution and a high temporal resolution. Three analyses are performed on the L3JRC product in order to derive spatiotemporal patterns about wildfire in the study area. For the analyses the theory of Exploratory Data Analysis is used. Behaviour of wildfire is made visible by means of map representations. Subsequently there are in each analyses cluster patterns described by investigating the map visualisations. The first analysis aims on the comparison of yearly fire frequency. The second analysis performs a correlation of burned area with two other variables, namely herbaceous vegetation cover and human influence index. The last analysis looks at the variation of monthly fire frequency. One of the results comprehends that fires are mainly occurring in a large cluster between 5° and 15° latitude south, and there are two large clusters that have no or almost no fire, i.e. an area near the equator and a large area in and around the Kalahari. Also we found that the correlation between fire and herbaceous vegetation cover is moderate positive, and the correlation between fire and human influence is negative and small. Concerning the causes of wildfire we distinguished two ignition types : natural caused fire, which means ignited by lightning, and anthropogenic caused fire, which means ignited by human. In order to assign the two ignition types to wildfires that are represented in the L3JRC product, the so called Ignition Type Model is created. This model aims on assigning ignition types by considering the location in space and time of each wildfire. The Ignition Type Model consists of four steps. In the first step the burned areas of the L3JRC are converted to individual wildfires, and further the starting point of each wildfire is selected. In the second step thunderstorms, which are derived from the Lightning Imaging Sensor product, are selected that might reproduce strikes that ignite wildfire. In the third step of the model these thunderstorms are allocated to wildfires if they comply with certain parameters for distances in space and time. The fourth and last step of the Ignition Type Model implies the allocation of wildfires with human object, like roads and urban areas, based on certain distance parameters. The Ignition Type Model resulted in 1690546 wildfires. 19.3% of these fires are allocated to either lightning or human objects and therefore received an ignition type. In total 34215 wildfires are assigned as natural caused and 292695 wildfires are assigned as anthropogenic caused. Based on these results the percentage of natural caused and anthropogenic caused wildfires in Southern Africa is calculated as respectively 31.7% and 68.3%. Two analyses are performed in order to derive spatiotemporal characteristics from the two ignition type datasets that are the result of the Ignition Type Model. The first analysis aims on the monthly fire frequency of both natural caused and anthropogenic caused wildfire. The second analysis performs an investigation of natural caused fire and the land cover type on which these fires occurred. These two analyses revealed that there are differences in spatial clustering of both ignition types. Both the spatial and temporal clustering of natural caused fires is smaller. We also found that 82% of all lightning caused fires are located in woodland areas.

Mots clés  : fires / firebreaks / fire danger / analysis / wildfires / burning / ignition

Présentation

Version intégrale (2 Mb)

Page publiée le 26 janvier 2015, mise à jour le 18 octobre 2018