A case study for Skukuza [Elektronische Ressource] : estimating biophysical properties of fires using EOS-MODIS satellite data / vorgelegt von Tobias Landmann

Mfienem

Informations

• Publié par: georg-august-universitat_gottingen
• Publié le: 01 janvier 2003
• Nombre de lectures: 26
• Langue: Deutsch
• Poids de l'ouvrage: 4 Mo

Extrait

A case study for Skukuza: Estimating biophysical
properties of fires using EOS-MODIS satellite data




Dissertation
zur Erlangung des Doktorgrades
der Mathematisch-Naturwissenschaftlichen Fakultäten
der Georg-August-Universität zu Göttingen





vorgelegt von

Dipl. Geogr. Tobias Landmann



aus Elim (Südafrika)









Göttingen 2003





























D 7

Referentin/Referent: Prof. Dr. G. Gerold

Korreferentin/Korreferent: Prof. Dr. M. Kappas

Tag der mündlichen Prüfung: 11.02.2003








Vorwort

Savannenbrände in Afrika, speziell in Südafrika, tragen zu einem großen Anteil an der
jährlichen feuerbedingten Aerosol- und Gasemissionen in die Atmosphäre bei. Sowohl von
der Häufigkeit als auch von der verbrannten Biomasse her (35 % der globalen jährlichen
Biomasseverluste durch Feuer) besitzt Südafrika eine Schlüsselrolle in Untersuchungen zum
feuerbeeinflussten Savannenökosystem, für das Feuermonitoring und Feuermanagement im
Zusammenhang mit den Savannen-Nationalparks und für die Analyse und Modellierung der
Emissionen in die Atmosphäre. Eine wesentliche Grundlage dafür stellt die weitflächige
satellitengestützte Analyse der brennbaren Biomasse, Zahl und Lokalität der Feuer sowie
Erfassung der Feuerintensität dar.
Unter Beteiligung am SAFARI-2000-Programm (Southern Africa Regional Science Initiative)
konnten mit dieser Dissertation am Beispiel des Krüger-Nationalparks in Südafrika sowohl im
Feld leicht einsetzbare Methoden zur Analyse der brennbaren Biomasse wie auch neue
Algorithmen zur pixelgenauen Erfassung von Feuerintensität und Brandflächen über MODIS-
und Landsat-ETM-Daten entwickelt werden. Ihre Anwendbarkeit für bessere regionale
Emissionsmodelle - wie für ein effektives Feuermanagement im Nationalpark - wird in dieser
Arbeit aufgezeigt und ist auf andere Savannenbrände übertragbar.
Dank der guten Unterstützung über die Gottlieb Daimler und Karl-Benz-Stiftung, der
südafrikanischen Forschungseinrichtungen und Kollegen (CSIR Environmentek) und der
Universität Göttingen konnte diese Dissertation erfolgreich abgeschlossen werden.


Göttingen, den 1. November 2003








I
Prologue

Biomass burning in Africa, explicitly in Southern Africa, contributes excessively to the
annual aerosol loading and pyrogenic gas emissions within the African atmospheric system.
The frequency of burning and the extensive size of burned areas in Southern Africa (wildfires
in Southern Africa account for 35% of global annual fire biomass fuel consumptions) make
South Africa suitable for fire related research. Research regarding the fire effects on
ecosystem function, the development of spatial fire monitoring mechanisms and fire
management practises in protected areas is necessary to develop, improve and analyse models
that predict regional pyrogenic emissions. Spatial fire information from remote sensing can be
used as tool to rigorously and effectively detect and analyse biomass fuel available for
burning, fire count locations as well as fire severity.
This dissertation was accomplished within the context of the 2000 Southern Africa Regional
Science (SAFARI-2000) Initiative, using the Kruger National Park in South Africa as a
‘casestudy’ savanna ecosystem. Methods to analyse pre-burn biomass fuel loads, spatially
explicit algorithms to detect fire severity, and fire scar mapping and detection methods using
Landsat ETM+ and MODIS satellite imagery were developed. The applicability of these
methods and algorithms to effectively improve regional emissions models and fire
management practices in protected areas are shown. The new methods developed within the
context of this dissertation may be applicable to similar savanna ecosystems. This dissertation
could be completed successfully and thankfully with the support of the Gottlieb Daimler and
Karl-Benz Foundation, South African Research Institutions (especially CSIR Environmentek)
and the University of Goettingen in Germany.


st Göttingen, 1 of November 2003


II
Abstract

Context Abstract. The Southern African Regional Science Initiative (SAFARI-2000) program
provides the working context of this study. SAFARI-2000 aims to explore study and address
linkages between land-atmosphere processes and the relationship between biogenic,
anthropogenic emissions and the consequences of their deposition to the functioning of the
biogeophysical and biogeochemical systems (SWAP et al., 2001). The interdisciplinary nature
of SAFARI allows an integrative data analyses approach. Through collaborative SAFARI
research, fire study sites in South Africa were identified for in situ validation and refinement
of fire information from the MODerate resolution Imaging Spectroradiometer (MODIS) using
high-resolution satellite data and field data observations at representative validation sites.
Biomass burning contributes significantly to global budgets of many atmospheric gases, the
rising levels of which are implied with potential climate forcing factors and global change.
Global satellite estimates of biomass burning with high accuracies are thus required to provide
exact figures for gas fluxes derived from this source. Only satellite data combined with well-
defined numeric models can provide these data.

Summary Abstract. Biomass burning in Southern African savannas has the potential to emit
large amounts of trace gases and aerosols to the atmosphere. There are large uncertainties in
methods that quantitatively measure sub-pixel fire effect, spatial explicit fuel biomass
parameters, biomass consumption rates and combustion efficiencies rigorously and effectively
over large physiological diverse savanna landscapes. Savannas in Southern Africa are
characterized by a highly differential nature of fire behavior mainly due to fragmented land
cover. This thesis will utilize the improved remote sensing capabilities provided by Earth
Observing System (EOS) MODerate resolution Imaging Spectroradiometer (MODIS) and
Landsat ETM+ to measure fire effects at the Skukuza-area in the Kruger National Park
(KNP), South Africa. In situ aboveground fuel biomass data and fire spectral properties were
collected at the KNP site and investigated to feed combustion completeness, sub-pixel area
fractions burnt and fire severity models from contemporary 30-meter Landsat ETM+ data,
using appropriate wavelengths. Implications are made for automated and effective fire scar
monitoring techniques in multi-temporal MODIS data sets. Further, the study illustrates how
spatial explicit Landsat ETM+ fuel biomass satellite calculations that account for the fuel type
complexity in each pixel, combustion completeness and area burnt from contemporary
MODIS overpasses as well as in situ emission factors can be used to accurately estimate gas
III
and particulate emissions for the KNP-area. This method is shown to reduce some
uncertainties in local emission estimates. Finally the results show that MODIS and Landsat
ETM+ spectral and spatial properties of fires can effectively be used to corroborate fire
management policies in the KNP.
IV
TABLE OF CONTENTS

PROLOGUE................................................................................................ I-II
ABSTRACT...III
TABLE OF CONTENTS............................................................................... V
LIST OF FIGURES......................IX
LIST OF TABLES .......................................................................................XV
LIST OF MAPS.......................... XVI

SECTION 1: INTRODUCTION........................................................................................ 1
1.1 DETERMINANTS OF FIRES IN SOUTHERN AFRICA............................. 2
1.1.1 Regional fire patterns 2
1.1.2 Landscape and landform fire patterns..................................................................... 4
1.2 SCIENCE RATIONAL FOR REMOTE SENSING FIRE INFORMATION........ 5
1.2.1. Fire regimes as sources or sink.............................................................................. 6
1.2.2. Atmospheric chemistry effects 6
1.2.3. Ecosystem effects ................................................................................................. 7
1.3 THE HERITAGE OF FIRE REMOTE SENSING....... 8
1.3.1 Past, current and future fire detection sensors......................... 8
1.3.2 Current MODIS status ................................................................ 11
1.4 STUDY SITES ............................................... 12
1.5 SATELLITE DATA CORRECTIONS....