Modelling spatial temporal patterns and drivers of urban residential fire risk

Ardianto, R 2018, Modelling spatial temporal patterns and drivers of urban residential fire risk, Doctor of Philosophy (PhD), Business IT and Logistics, RMIT University.

Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Ardianto.pdf Ardianto.pdf Click to show the corresponding preview/stream application/pdf; Bytes
Title Modelling spatial temporal patterns and drivers of urban residential fire risk
Author(s) Ardianto, R
Year 2018
Abstract Fire risk, in general, is the probability of a fire occurrence and its potential consequences (e.g. injuries/deaths or financial losses). An exposure to the source of fire ignition such as a live flame or a spark that is further fuelled by the presence of combustible materials, faulty electrical wiring or cooking devices, directly contributes to fire risk. It also hinges on an individual’s perception of fire risk, exhibited in situ behaviour such as alcohol drinking habits and the preparedness to respond to threat from fire. More broadly, fire risk is influenced by the size and characteristics of the population at risk or exposed to a fire hazard, and the levels of community resilience, which reflect the sustained ability to utilize available resources to respond to, withstand, and recover from adverse situations. Fire risk, therefore, is difficult to examine as it is driven by a multitude of interwoven factors. There are numerous studies that have applied a range of methods to model fire risk. These methods undoubtedly provide a useful baseline to enhance emergency response and improve resource allocation. Nonetheless, the accuracy, reliability and robustness of these models can be further improved by considering fire risk as a stochastic phenomenon. Furthermore, the role, which space and time plays in shaping fire risk a stochastic process is often overlooked.

The aim of this study was to develop stochastic models to estimate the likelihood of residential fire occurrence and to identify key urban characteristics underpinning fire patterns over time and across space. This study addressed four interrelated key research questions: (i) How does a residential fire pattern occur over time and space? (ii) Can the probability of a residential fire occurrence be predicted as a stochastic process? (iii) How do the urban characteristics impact on residential fire risk? (iv) What spatially-integrated strategies can be developed to mitigate fire risk in an urban setting?

The application of the Markov chain model and Geographically Weighted Regression (GWR) is a key contribution of this thesis because of the novelty of the methodology in quantifying residential fire risk which not only potentially improves the accuracy and reliability of fire risk modelling, but also enriches our understanding of behaviour associated with fire risk in relation to space, time, and situated context at the local level. These models are constructed using residential fire data for the Melbourne Australia, spanning a ten-year period from June 2005 to May 2015.

The key findings demonstrate that, first, the incidence of residential fires across Melbourne during a 10-year period show a higher degree of fluctuation with a strong seasonal variation according to the months of the year. June-August is recorded as having the highest rate whereas March-April and November recorded the lowest rate of fire occurrence. Second, the mapping of the probability of fire occurrence across the Melbourne metropolis shows a city-centric spatial pattern where inner-city sub-regions are relatively more vulnerable to fire than are the outer sub-regions. Third, the time threshold that affects the fire risk levels within a neighbourhood that has had a fire is about two months. When this period of low fire risk elapses, the probability of a fire increases to the normal baseline, equivalent to that in areas with no fire. Fourth, a fire that occurs in a distant area has no significant effect on mitigating fire risk within the neighbourhood. When a distance threshold of 5 km is reached, either one fire or no fire in the past significantly increases the probability of fire. Areas with two or more fires within the neighbourhood are likely to reduce the chance of a fire in the initial period. Fifth, the key findings also reveal that the distribution of residential fires across Melbourne has a complex pattern and is associated with both temporally and spatially-varying neighbourhood attributes. The effect of socio-spatial characteristics such as language, residential mobility, home ownership, type of dwellings, and dwelling density, in relation to residential fires risk tends to be inconsistent across urban areas. Different areas have different contextual situations which influence the level of residential fire risk.

These findings provide new empirical evidence useful for fire agencies seeking to establish appropriate strategies to mitigate adverse impacts of fire on communities. It can also help to identify high fire risk areas and to geo-target when and where to disseminate fire safety information to increase residents’ awareness of fire risk.
Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Business IT and Logistics
Subjects Urban Analysis and Development
Logistics and Supply Chain Management
Stochastic Analysis and Modelling
Keyword(s) Residential Fire Risk
Markov chain
Geographically Weighted Regression
Memory effect
Local learning
Version Filter Type
Access Statistics: 153 Abstract Views, 45 File Downloads  -  Detailed Statistics
Created: Fri, 08 Jun 2018, 14:15:45 EST by Denise Paciocco
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us