The Effect Of Land Use Practices on the Spatial and Temporal Characteristics of Savanna Fires in Namibia [Elektronische Ressource] = Der Effekt der Landnutzung auf die räumlichen und zeitlichen Merkmale der Savannenbrände in Namibia / Johan le Roux. Betreuer: Cyrus Samimi

Johan Le Roux

Sujets

• Biologie

Informations

• Publié par: friedrich-alexander-universitat_erlangen-nurnberg
• Publié le: 01 janvier 2011
• Nombre de lectures: 4
• Langue: English
• Poids de l'ouvrage: 4 Mo

Extrait

THE EFFECT OF LAND USE PRACTICES ON THE
SPATIAL AND TEMPORAL CHARACTERISTICS OF
SAVANNA FIRES IN NAMIBIA

DER EFFEKT DER LANDNUTZUNG AUF DIE RÄUMLICHEN UND ZEITLICHEN
MERKMALE DER SAVANNENBRÄNDE IN NAMIBIA.





Der Naturwissenschaftlichen Fakultät
der Friedrich-Alexander-Universität Erlangen-Nürnberg
zur
Erlangung des Doktorgrades Dr. rer. nat.







Vorgelegt von

Johan le Roux
aus Oranjemund


Als Dissertation genehmigt
von der Naturwissenschaftlichen Fakultät
der Friedrich-Alexander-Universität Erlangen-Nürnberg





















Tag der mündlichen Prüfung: 29. Juni 2011


Vorsitzender der
Promotionskommission: Prof. Dr. Rainer Fink
Erstberichterstatter: Prof. Dr. Cyrus Samimi
Zweitberichterstatter: Prof. Dr. Rupert Bäumler
Summary

Namibia is often branded as the most arid country in sub-Saharan Africa. Despite this
apparent aridity, a strong east-west rainfall gradient combined with typical sub-tropical
seasonality in rainfall, produces fine biomass fuels which are sufficient in quantity and
more than dry enough in quality to support widespread savanna fires every year.

These fires which have been burning in Africa for centuries are perceived by many as a
scourge – something which should be suppressed at all cost. This outlook has been
perpetuated by a succession of mostly foreign administrators and consultants, who are
striving to enshrine their well-intended but somewhat misguided views in policy under
the guise of integrated fire management.

Burned area mapping in Namibia dates back to the mid 1990’s when a NOAA AVHRR
HRPT direct broadcast receiving station was installed at the Etosha Ecological Insti-
tute. While burned area data has been freely available, it only served to add fuel to the
debate, because apparently large areas of Namibia’s woodlands burned “every year”.
What was lacking is an analysis of the data that would provide answers to the ques-
tions of where does it burn? Where does it burn every year? At what time of year does
it burn? And many others.

In order to answer these questions, the data needed to be re-processed. Not only to fill
spatial and temporal gaps in the existing datasets, but also to produce burned area
products that could be used to address the all-important issue of fire intensity/severity.
This was not possible with the existing national datasets which had been aggregated to
a single burned area map for a particular year. The AVHRR archive was therefore
back-processed to produce monthly burned area maps of Namibia for the 10 year peri-
od from 1994 to 2003. In order to relate the fire season characteristics to the plant ma-
terial that it relies on for fuel, a corresponding time series of NDVI images were pro-
cessed to extract phytophenological milestones. In order to reduce the data volume
and to align it with existing biophysical datasets for Namibia, the burned area and NDVI
products were aggregated by Quarter Degree Square (QDS). The milestone extraction
from NDVI and temporal data extraction from burned area, was made possible by the
development of two spreadsheet tools.
The data tables produced by the spreadsheet tools were combined and analysed in a
GIS, to produce a series of maps that characterise Namibian fire regimes – temporally
as well as spatially.

The results show that only 38% of the QDSs in Namibia were affected by fire during the
10 years covered by the study. Since the arid south-eastern parts of the country do not
burn, the FAA is confined to the central and north-eastern areas. While a small propor-
tion of this area starts burning within a month or two after the peak of the growing sea-
son, over much of the area the lag is between 4 and 6 months, giving rise to a peak in
the burning season during the months of August and September. The burning season
generally extends over a 2-6 month period, although some parts – notably in land use
category (LUC-1) 1 agriculture and tourism on freehold land – the season is controlled
and last for 1 month only. Over much of the FAA, the fire regime can be described as
mild, with only 6% of the FAA subjected to a severe fire regime.

In order to evaluate the role of land-use on the fire regimes, Spearman’s rank correla-
tion coefficients were calculated for fire return period, fire season duration, peak month
for the fire season, and area burned (AB) paired with human population density, live-
stock density, landscape fragmentation, bush density, rainfall (RF), maximum NDVI
value (Max NDVI), Mean NDVI value, NDVI greening up rate, peak month of NDVI, and
sum of NDVI values (Total NDVI).

For the fire affected area as a whole, there are significant correlations between all of
the variables, except for fire season peak and the NDVI variables. This confirms a de-
gree of independence between the time of year when the most burning takes place,
and the production of the biomass that is consumed. There is also no significant corre-
lation between the NDVI green-up rate and the extent of burning. When considering the
four spatio-temporal fire characteristics - Fire Return Period (FRP), Fire Season Dura-
tion (FSD), Fire Season Peak (FSP) and area burned, the strongest correlations are
between:
• FRP and RF(-); FRP and Mean NDVI(-); FRP and Total NDVI(-), showing that an
increase in rainfall (and therefore NDVI) is associated with more frequent burning.
• FSD and Bush Density(-); FSD and RF(+); FSD and Max NDVI(+); FSD and Mean
NDVI(+), showing that an increase in rainfall (and therefore NDVI) is associated with a lengthening of the fire season which is mitigated or inhibited by an increase in
bush density.
• FSP and Human Population Density(-), showing that an increase in human popula-
tion density is associated with an earlier peak to the burning season. Although this
was the strongest relationship, it was still statistically weak while the remainder were
very weak or not significant.
• AB% and Landscape Fragmentation(-); AB% and Livestock Density(-); AB% and
Bush Density(-); AB% and Rainfall(+), showing that an increase of the area burned
is associated with an increase in rainfall, but that this effect is reduced by landscape
fragmentation, bush density and livestock grazing pressure.

There is a marked difference in the correlations between the spatio-temporal fire char-
acteristics and the land use/environmental parameters within different land use catego-
ries. One category in particular, LUC-1 – agriculture and tourism on Freehold Land,
differs radically from the rest. The reason for this is not so much a case of land use
directly affecting the fire environment, but it provides the farmers with the ability to con-
trol fire events - an ability that stems from a highly developed road network, combined
with a mobile community who perceives every fire as a threat and is prepared to come
together to fight it at all costs.
Zusammenfassung

Namibia gilt als das trockenste Land im subsaharischen Afrika. Trotz der Aridität,
einem starken Ost-West-Gradienten und der typischen subtropischen Saisonalität wird
genügend Biomasse produziert, die auch trocken genug ist, um jährlich weit verbreitete
Savannenbrände zu ermöglichen.

Feuer, die in Afrika seit Jahrhunderten betreffen, werden von Vielen als Plage ange-
sehen, die es um jeden Preis zu unterdrücken gilt. Diese Ansicht wird durch meist
fremde Administratoren und Berater fortgeführt. Sie bemühen sich, ihre gut gemeinten
aber teilweise fraglichen Ansichten als integriertes Feuermanagement zu bewahren.

In Namibia reicht die Kartierung von Feuerflächen bis in die Mitte der 90iger Jahre
zurück, als im Etosha Nationalpark eine Empfangsstation für NOAA AVHRR HRPT
installiert wurde. Obwohl Daten zu Brandflächen frei verfügbar sind, wurde nur die
Debatte um Feuermanagement angeheizt, da offensichtlich große Bereiche der
Gehölzbestände Namibias jährlich brennen. Was fehlt sind Analysen der Daten, die
Antworten auf die zentralen Fragen geben. Diese sind neben anderen: Wo brennt es?
Wo brennt es jährlich? Zu welcher Jahreszeit brennt es?

Um diese Fragen zu beantworten, mussten die Daten zuerst vorprozessiert werden.
Dabei wurden nicht nur räumliche und zeitliche Lücken gefüllt, sondern auch
Datenprodukte über Feuerflächen erstellt, die im Folgenden dazu dienten, sich der
wichtigen Frage der Feuerintensität zu widmen. Dafür kam nicht nur der existierende
nationale Datensatz zur Verwendung, aus dem für jedes Jahr eine Brandflächenkarte
aggregiert wurde, sondern auch das AVHRR Archiv für eine 10-Jahresperiode, rück-
prozessiert für die Jahre 1994 bis 2003, um monatliche Brandflächenkarten für
Namibia zu erstellen. Eine korrespondierende Zeitreihe des NDVI diente der Ermittlung
der phänologischen Eckdaten,