Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

När har all Sveriges skog brunnit ned? En studie över svenska skogsbränder under åren 1998 till 2018.

Byström, Johan LU and Fjellström, Jakob LU (2020) STAH11 20192
Department of Statistics
Abstract
This thesis uses a data material from the Swedish Civil Contingencies Agency (MSB) to analyze forest fires in Sweden between 1998 and 2018. The thesis also uses weather data (precipitation and average temperature) as explanatory variables in two different regression models. The purpose of this thesis is to make predictions for the number of forest fires and the total burned down hectare of forest for the period 2019 to 2021. This is a research field where, for Swedish conditions, there is a lack of research and this thesis aims to fill that gap. Since the number of forest fires is a discrete variable a GLM model, that incorporates auto-regressive terms, is used and the observations are assumed to follow a Negative Binomial Distribution. To... (More)
This thesis uses a data material from the Swedish Civil Contingencies Agency (MSB) to analyze forest fires in Sweden between 1998 and 2018. The thesis also uses weather data (precipitation and average temperature) as explanatory variables in two different regression models. The purpose of this thesis is to make predictions for the number of forest fires and the total burned down hectare of forest for the period 2019 to 2021. This is a research field where, for Swedish conditions, there is a lack of research and this thesis aims to fill that gap. Since the number of forest fires is a discrete variable a GLM model, that incorporates auto-regressive terms, is used and the observations are assumed to follow a Negative Binomial Distribution. To make predictions for the total burned down hectare of forest a multiple linear regression model is used. Our results suggest minor deviations in future forest fires from the overall pattern over the last two decades. However, our predictions carry some uncertainty since our prediction intervals are quite wide. (Less)
Please use this url to cite or link to this publication:
author
Byström, Johan LU and Fjellström, Jakob LU
supervisor
organization
course
STAH11 20192
year
type
M2 - Bachelor Degree
subject
keywords
GLM, Linear Regression, Time-Series, Negative Binomial Distribution
language
Swedish
id
9001999
date added to LUP
2020-02-28 13:50:04
date last changed
2020-02-28 13:50:04
@misc{9001999,
  abstract     = {{This thesis uses a data material from the Swedish Civil Contingencies Agency (MSB) to analyze forest fires in Sweden between 1998 and 2018. The thesis also uses weather data (precipitation and average temperature) as explanatory variables in two different regression models. The purpose of this thesis is to make predictions for the number of forest fires and the total burned down hectare of forest for the period 2019 to 2021. This is a research field where, for Swedish conditions, there is a lack of research and this thesis aims to fill that gap. Since the number of forest fires is a discrete variable a GLM model, that incorporates auto-regressive terms, is used and the observations are assumed to follow a Negative Binomial Distribution. To make predictions for the total burned down hectare of forest a multiple linear regression model is used. Our results suggest minor deviations in future forest fires from the overall pattern over the last two decades. However, our predictions carry some uncertainty since our prediction intervals are quite wide.}},
  author       = {{Byström, Johan and Fjellström, Jakob}},
  language     = {{swe}},
  note         = {{Student Paper}},
  title        = {{När har all Sveriges skog brunnit ned? En studie över svenska skogsbränder under åren 1998 till 2018.}},
  year         = {{2020}},
}