Peat Dynamics in the Angolan Highlands

Mauro Lourenco

830429

A thesis submitted in fulfilment of the academic requirements for the

degree of Doctor of Philosophy

School of Geography, Archaeology and Environmental Studies

University of the Witwatersrand, Johannesburg

March 2023

DECLARATION

I declare that this thesis is my own, unaided work, except where otherwise

acknowledged. It is being submitted for the Degree of Doctor of Philosophy in Science

at the University of the Witwatersrand, Johannesburg. It has not been submitted before

for examination for any degree at this or any other university.

________________________________

Mauro Lourenco

22 March 2023 at Johannesburg

iii

ABSTRACT

The Angolan Highlands is a war stricken, threatened, and under-studied area. The

region is hydrologically and ecologically important and supports extensive tropical

peatland deposits. Peatland preservation has been acknowledged to address climate

change, is sensitive to drought and fire, and is directly influenced by vegetation and

hydrological conditions. However, little research has been conducted in the Angolan

Highlands. This study addresses gaps in the literature through four key contributions.

The first is a critical review of peat definitions: the implications of disparate definitions

are detailed, and a new proposed definition for peatlands in the interest of climate

science is provided. The second is the first map of peatland extent in the Angolan

Highlands, containing details on the age and growth dynamics. The study presents a

conservative estimate of peatland extent that is much larger than previously estimated

for Angola and is a crucial first step in facilitating the preservation of this deposit. The

third contribution is the first historical assessment of drought and vegetation response

in the region. This contains a 40-year drought and 20-year vegetation history,

demonstrating that drought occurrence is increasing and there is a strong relationship

between precipitation and the peatland vegetation region. The fourth contribution is

the first assessment of the contemporary (2001-2020) fire regime of these peatlands,

and reveals that among all land cover classes, peatlands burn more frequently and at

a higher proportion. Investigation into the peat dynamics of the Angolan Highlands

indicate that they have critical importance and are naturally resistant to both droughts

and fire. Failure to preserve these deposits will have direct implications on the

communities, environment, and surrounding areas.

Keywords: Angolan Highlands, Carbon, Drought, Fire, Peat, Remote Sensing.

LIST OF WORKS

This study is organised in chapters, four of which are stand-alone research papers

that have been accepted for publication in peer-reviewed journals. These are

combined with an introduction, study region, methodology, general discussion, and

conclusion chapter to present a PhD thesis by publication. The four published papers

include:

1. Lourenco M, Fitchett JM and Woodborne S (2022) Peat definitions: A critical

review. Progress in Physical Geography: Earth and Environment, DOI:

10.1177/03091333221118353

Current status: Accepted for publication (21 July 2022), available online.

Impact factor for 2021-2022: 4.283, five-year impact factor: 5.023.

Project leaders Fitchett JM and Woodborne S (30% contribution) conceptualised the

scope of this manuscript. Lourenco M (70% contribution) collected and analysed the

data, produced the figure and tables in the text, and led the writing of the manuscript

and revisions with input from the project leaders.

2. Lourenco M, Fitchett JM and Woodborne S (2022) Angolan highlands peatlands:

Extent, age and growth dynamics. Science of The Total Environment, 810,

152315. DOI: 10.1016/j.scitotenv.2021.152315

Current status: Published.

Impact factor for 2021: 10.753.

Project leaders Fitchett JM and Woodborne S (35% contribution) conceptualised the

scope of this manuscript. Lourenco M (65% contribution) collected and analysed the

https://doi.org/10.1177/03091333221118353
https://doi.org/10.1016/j.scitotenv.2021.152315

data, produced the maps, figures, and tables in the text, and led the writing of the

manuscript and revisions with input from the project leaders.

3. Lourenco M, Woodborne S and Fitchett JM (2022) Drought history and vegetation

response in the Angolan Highlands. Theoretical and Applied Climatology, 151:

115–131 DOI: 10.1007/s00704-022-04281-4

Current status: Published.

Impact factor for 2021: 3.409, five-year impact factor: 3.518.

Project leaders Fitchett JM and Woodborne S (30% contribution) conceptualised the

scope of this manuscript. Lourenco M (70% contribution) collected and analysed the

data, produced the maps, figures, and tables in the text, and led the writing of the

manuscript and revisions with input from the project leaders.

4. Lourenco M, Woodborne S and Fitchett JM (2023) Fire regime of peatlands in the

Angolan Highlands. Environmental Monitoring and Assessment, 195(1): 1–17

DOI: 10.1007/s10661-022-10704-6

Current status: Published.

Impact factor for 2021: 3.307, five-year impact factor: 3.420.

Project leaders Fitchett JM and Woodborne S (25% contribution) conceptualised the

scope of this manuscript. Lourenco M (75% contribution) collected and analysed the

data, produced the maps, figures, and tables in the text, and led the writing of the

manuscript and revisions with input from the project leaders.

https://doi.org/10.1007/s00704-022-04281-4
https://doi.org/10.1007/s10661-022-10704-6

Conference presentations:

1. Lourenco M, Fitchett JM and Woodborne S (2021). Towards a peatland

Inventory for the Angolan Highlands using Google Earth Engine. Society of

South African Geographers and South African Association of

Geomorphologists Joint Conference, 6-8 September 2021.

2. Lourenco M, Woodborne S, and Fitchett JM (2022). Drought history and

vegetation response in the Angolan Highlands. Society of South African

Geographers Biennial Conference, 12-14 September 2022.

vii

ACKNOWLEDGEMENTS

First, I thank my two supervisors, Prof. Jennifer Fitchett, and Prof. Stephan

Woodborne for their individual and collective support, guidance, and technical advice

throughout my PhD. Through your supervision, you have helped me produce a thesis

containing four journal articles accepted for publication in international journals. I am

immensely proud of each of these contributions.

To Prof. Jennifer Fitchett, thank you for supporting my enthusiasm and curiosity when

working in Google Earth Engine. Thank you for your patience and the time taken to

help develop my interests into measurable and meaningful outputs. You helped me

become a better writer, academic and scientist. Thank you for funding my conference

presentations and pushing me to present my work to my fellow colleagues within the

school. You taught me to celebrate my academic achievements, to be more organised

and productive, and you encouraged me to enter new and exciting avenues of

research. Thank you for the countless messages of support, reassurance, and

guidance I received throughout. You are a fantastic supervisor, and I am incredibly

thankful for your supervision.

To Prof. Stephan Woodborne, thank you for helping me to focus on the bigger picture.

You helped me develop my critical thinking skills. You taught me to write concisely

and communicate my research effectively and with confidence. Thank you for the time

you took to create workshop spaces that included skills development, new techniques,

theories, and discussions of important science. Thank you for your asking the difficult

questions and challenging me to investigate further. Thank you for your insight, the

long phone call chats, and advice. You are an excellent supervisor, and I am incredibly

thankful for your supervision.

viii

I thank the anonymous reviewers for their valuable comments and inputs during the

review process of each of the four accepted papers.

I gratefully acknowledge the support of the National Geographic Okavango Wilderness

Project, including the financial support and invitation to attend the Lungui Bungu River

expedition in June 2022. I thank Dr. Rainer von Brandis for support, discussions, and

feedback during this PhD.

I thank Dr. Elhadi Adam for his help and guidance using Google Earth Engine in the

early stages of my PhD.

I thank Prof. Paida Mhangara and Prof. Diane Grayson for each offering me separate

sessional lecturing opportunities at Wits University during my PhD.

I thank Wits University for awarding me with the Post Graduate Merit Award and

funding my tuition fees over the course of my PhD.

I thank my colleagues in the School of Geography, Archaeology and Environmental

Studies for your support, encouragement, and insight.

I thank my family (my mom, dad, and brother), for supporting me and providing me the

opportunity to pursue my PhD. Thank you for your interest in my research and for your

unwavering belief in my capabilities. Thank you to my mom for your encouragement

and for reminding me that I need to put my education first. Thank you to my dad for

reminding me that a PhD does eventually come to an end, and that I need to stay

positive. Thank you to my big brother Marcio for your friendship, guidance, insight, and

advice.

To my Godfather Ricky and his wife Nadia, thank you for your support during my PhD.

To my extended family and friends that encouraged and celebrated with me

throughout my PhD, thank you.

I would like to offer special thanks to my grandfather, Vovo Lourenco, who is no longer

with us, you will forever have a special place in my heart.

Lastly, I would like to offer special thanks to my cousin, Hugo Infante, who although is

no longer with us, continues to inspire by the example, dedication and passion he left

behind.

TABLE OF CONTENTS

Declaration...................................................................................................................ii

Abstract.......................................................................................................................iii

List of works.................................................................................................................iv

Acknowledgements....................................................................................................vii

Table of Contents.........................................................................................................x

List of Figures.............................................................................................................xiii

List of Tables............................................................................................................xviii

List of Acronyms and Scientific Terms........................................................................xx

Chapter 1: Introduction............................................................................................1

1.1 Background…................................................................................................1

1.1.1 The Angolan Highlands……………………..…..……...................…..…...1

1.1.2 Peat……………………………………………………………………...……2

1.1.3 Remote Sensing……………………………………………...………...…...4

1.2 Rationale……………………………………………...…....................................5

1.3 Study Aim and Objectives..............................................................................6

1.4 Structure of the Thesis...................................................................................7

Chapter 2: Study Region..........................................................................................9

2.1 Introduction....................................................................................................9

2.2 Topography and Drainage...........................................................................11

2.3 Soils and Vegetation....................................................................................13

2.4 Contemporary Climate and Weather..…………...........................................19

2.5 Conclusion...................................................................................................25

Chapter 3: Methodology…………..........................................................................26

3.1 Introduction…..............................................................................................26

3.2 Remote Sensing………………….…............................................................26

3.3 Google Earth Engine……………..................................................................27

3.4 NGOWP Lungui Bungu River expedition…..................................................29

Chapter 4: Peat Definitions: A critical review..........................................................31

4.1 Brief synopsis..............................................................................................31

Chapter 5: Angolan highland peatlands: Extent, age and growth dynamics…........48

5.1 Brief synopsis..............................................................................................48

Chapter 6: Drought history and vegetation response in the Angolan Highlands…..65

6.1 Brief synopsis..............................................................................................65

Chapter 7: Fire regime of peatlands in the Angolan Highlands...............................84

7.1 Brief synopsis……………….........................................................................84

Chapter 8: General Discussion….........................................................................103

8.1 Introduction................................................................................................103

8.2 Peat Preservation in the Angolan Highlands…...........................................103

8.3 Environmental Change in the Angolan Highlands......................................108

8.4 Concerns for the Angolan Highlands region…………….............................111

8.5 Research Limitations…………..……..…....................................................114

8.5.1 Data and analytical limitations for each journal paper chapter……..…115

Chapter 9: Conclusion…......................................................................................118

9.1 Synthesis……............................................................................................118

9.2 Achievement of study aim and objectives…...............................................119

9.3 Future research directions………………………...…..................................122

Comprehensive Reference List................................................................................126

Appendix 1. Chapter 4 Supplementary Material.........................................…………171

xii

Appendix 2. Chapter 5 Supplementary Material....……………………………………182

Appendix 3. Chapter 6 Supplementary Material..……………………...………………192

Appendix 4. Chapter 7 Supplementary Material…..…………………………………...198

xiii

LIST OF FIGURES

Figure 2.1. (a) Angola and its 18 provinces, and (b) the WWF ecoregions occurring in

Angola (Olson et al., 2001).........................................................................................10

Figure 2.2. (a) Elevation of Angola depicting the extent of the study site, and (b)

elevation of the study site including the WWF HydroSHEDS Basins Level 04 (Lehner

and Grill, 2013) and major rivers…………………………………………………………..11

Figure 2.3. Soil map of Angola (from Jones et al. 2013)…..…………………………….14

Figure 2.4. (a) The Cuito Source Lake, Moxico Province. Moist miombo woodland

grows on the hillsides adjacent to the lake. Peatland surrounds the source lake, and a

narrow band of grassland grows between the peatland and the miombo (From Goyder

et al., 2018). Photograph D. Goyder. (b) Lungui Bungu River Source, Moxico Province.

Small pool of acidic water filtering out of the surrounding peatland, miombo woodland

surrounds the bowl-shaped peatland, and evidence of a recent fire in the background.

Photograph R. von Brandis. (c) Lungui Bungu River, Moxico Province. The seep line

indicated by a narrow band of white sand between the miombo and the floodplain

environment, evidence of small-scale farming within the floodplain environment

towards the south of the drone photograph. Photograph J. Guyten............................16

Figure 2.5. (a) Cut section of a peat soil sample extracted by M. Lourenco with a

Russian corer. Photograph J. Guyton. (b) Burned (surface), cut, and drained peatland

patch, cassava and lavender growing in the background near a small village along the

Lungui Bungu River. Photograph M. Lourenco. (c) Drone photograph of a fire event

along the Lungui Bungu River floodplain. Photograph J. Guyton................................17

Figure 2.6. Mean annual rainfall in Angola (from Cain, 2017)…..................................21

xiv

Figure 2.7. Köppen-Geiger climate classification map for Angola (1980-2016: from

Beck et al., 2018)........................................................................................................22

Figure 2.8. (a) Average precipitation (1981-2020) and (b) Average day time land

surface temperature (2000-2020) for the delineated study area.................................24

Figure 4.1. (Appearing in the paper as Figure 1): Development of peat definitions

through time…………………………………………………………………………………35

Figure 5.1. (Appearing in the paper as Fig. 1): Study site map, (a) Map extent within

Angola, (b) Hillshade view of the four riparian Lungui Bungu River cores; samples 1,

2 and 3 lie within the current river floodplain, terrace 1, and sample 4 lies on the relict

floodplain, terrace 2, and (c) Map extent for this study, showing the three remaining

peat core locations and the hillshade view extent……………………………………….52

Figure 5.2. (Appearing in the paper as Fig. 2): Copernicus Land cover class area and

coverage of the Angolan Highlands………………………………………………………53

Figure 5.3. (Appearing in the paper as Fig.3): (a) Landsat 8 and (b) Sentinel 2 RF

classifications including coverage and area of each class……………………………...56

Figure 5.4. (Appearing in the paper as Fig. 4): (a) Overlap and non-overlap map

showing the extent of panel b and c, (b) and (c) are zoomed in sections of the mapped

area………………………………………………………………………………………….57

Figure 5.5. (Appearing in the paper as Fig. 5): Optical, vegetation, standing water

occurrence and topographic data of the mapped area from Landsat 8 (a–c), NASA

SRTM (d–f) and Sentinel-2 (g–i) sensors……………………………………………...…58

Figure 5.6. (Appearing in the paper as Fig. 6): Distribution plots for L8 and S2 peatland

with respect to NDVI, NDWI and SRTM data (elevation and slope). The peaks of each

distribution plot show the mode of values for individual peatland pixels. The NDVI and

NDWI plots relate to each respective peatland class from each RF classification, the

SRTM topographical data relates only to L8 peatland.………………………………….59

Figure 5.7. (Appearing in the paper as Fig. 7): Bacon Age-depth profiles for Angolan

Highlands peat cores, panels (a–c) represent the age models for CNV: Cuanavale

source lake peat, CS: Cuito source lake peat and CU: Cuando source lake cores,

respectively. Panels (d–g) represent the age models for Lungui Bungu River cores 1

to 4, respectively……………………………………………………………………………60

Figure 5.8. (Appearing in the paper as Fig. 8): Cross-section of the peat core sampling

site at the Lingui Bungu River……………………………………………………………..61

Figure 6.1. (Appearing in the paper as Fig. 1): Study area map showing the study site

extent in Angola and a Landsat 8 (USGS, 2021) red, green blue (RGB) optical image

of the study site……………………………………………………………………………..69

Figure 6.2. (Appearing in the paper as Fig. 2): (a) The mean annual precipitation

(mm/year) over the study site, (b) digital elevation model at 30 m resolution, and (c)

the mean cumulative precipitation (mm/month) over the period 1981–01-01 to 2020–

12-31………………………………………………………………………………………...71

Figure 6.3. (Appearing in the paper as Fig. 3): (a–d) The 3-, 6-, 12-, and 24-month SPI

for the period 1981–01-01 to 2020–12-31………………………………………………..72

Figure 6.4. (Appearing in the paper as Fig. 4): Mean (a) EVI and (b) NDVI over the

study area for the period 2000–02-18 to 2020–12- 31…………………………………..76

Figure 6.5. (Appearing in the paper as Fig. 5): The CHIRPS mean daily precipitation

for each month across the Angolan Highlands study area from 1981–01-01 to 2020–

xvi

12-31 and the mean NDVI and EVI for each month for the period 2000–02-18 to 2020–

12-31 across the three vegetation regions………………………………………………76

Figure 6.6. (Appearing in the paper as Fig. 6): (a) The mean NDVI and (b) EVI for each

vegetation region per month, the total precipitation per month, and the common

drought periods over the period (2000–02-18 to 2020–12-31)…………………………77

Figure 7.1. (Appearing in the paper as Fig. 1): (a) Study area map showing the study

site extent in Angola and (b) the LULC of the study area………………………………..88

Figure 7.2. (Appearing in the paper as Fig. 2): Peatland sites 1–5 along the Lungu

Bungu River…………………………………………………………………………………91

Figure 7.3. (Appearing in the paper as Fig. 3): Fire frequency per pixel over the period

01/01/2001 to 31/12/2020 at 250 m resolution…………………………………………..92

Figure 7.4. (Appearing in the paper as Fig. 4): Burn area per year for the study site

over the period 01/01/2001 to 31/12/2020…………………………………………….....92

Figure 7.5. (Appearing in the paper as Fig. 5): (a) Burn proportion and (b) burn area

per year for each LULC class over the period 01/01/2001 to 31/12/2020……...……...93

Figure 7.6. (Appearing in the paper as Fig. 6): Proportion of area having specific fire

frequencies for each LULC class………………………………………………………….94

Figure 7.7. (Appearing in the paper as Fig. 7): Box-and-whisker plots of the maximum,

minimum, average, mode, and inter-quartile range of the annual burned area for each

LULC class and all classes per month from 2001 to 2020………………………………96

Figure 7.8. (Appearing in the paper as Fig. 8): Peatland vegetation (average NDVI per

month) and burn history (2001–2020) for sites 1–5 (a–e). The black line indicates the

time series of average NDVI for each month, a burn value of 1 (red vertical line)

xvii

corresponds to the month in which the peatland burnt. In some instances, fires

occurred over multiple months in the same year, indicating that the peatland site

burned in unique locations at slightly different times in the year. (f) Average NDVI per

month for all sites during non-burn years (black line) and burn years (red line)……....97

xviii

LIST OF TABLES

Table 4.1. (Appearing in the paper as Table 1): Peat nomenclature used in

definitions…………………………………………………………………………………...37

Table 4.2. (Appearing in the paper as Table 2): Basis for peatland classification

systems……………………………………………………………………………………..38

Table 4.3. (Appearing in the paper as Table 3): Peat defined according to depth from

various sources……………………………………………………………………………..39

Table 4.4. (Appearing in the paper as Table 4): Threshold proportion of organic

carbon, organic matter and ash content that soils must contain to be considered

peat………………………………………………………………………………………….40

Table 5.1. (Appearing in the paper as Table 1): Land area coverage of overlap and

nonoverlap between the two RF classifications…………………………………………58

Table 6.1. (Appearing in the paper as Table 1): SPI scores and classification (Svoboda

et al., 2012)………………………………………………………………………………….70

Table 6.2. (Appearing in the paper as Table 2): Number of months classified as either

moderately, severely, or extremely dry per year according to the respective SPI

calculations…………………………………………………………………………………74

Table 6.3. (Appearing in the paper as Table 3): Common drought periods according

to each SPI calculation and ENSO years………………………………………………...76

Table 6.4. (Appearing in the paper as Table 4): Correlation matrix between vegetation

indices and precipitation…………………………………………………………………...77

Table 7.1. (Appearing in the paper as Table 1): Average fire frequency for each LULC

class…………………………………………………………………………………………93

xix

Table 7.2. (Appearing in the paper as Table 2): Percentage burn proportion per month

for each LULC class over the period 01/01/2001 to 31/12/2020. A colour pallet is used

to indicate high or low burn proportion……………………………………………………94

LIST OF ACRONYMS AND SCIENTIFIC TERMS

~ cal. yr BP: interpolated, calibrated AMS dates calculated using the BACON model;

years before present

AMS: Accelerator Mass Spectrometry

API: Application Programming Interface

ASTM: American Society for Testing and Materials

CHIRPS: Climate Hazards Group InfraRed Precipitation with Station data

CMI: Crop Moisture Index

CNV: Cuanavale Source Lake

CO2: Carbon dioxide

COP: Conference of Parties

CS: Cuito Source Lake

CU: Cuando Source Lake

DEM: Digital Elevation Model

DRC: Democratic Republic of the Congo

ENSO: El Niño Southern Oscillation

ESA: European Space Agency

EVI: Enhanced Vegetation Index

FAO: Food and Agriculture Organization

xxi

FAO UNESCO: Food and Agriculture Organization United Nations Educational

Scientific and Cultural Organization

GEE: Google Earth Engine

GIS: Geographical Information System

GHG: Greenhouse gas

IPCC: The Intergovernmental Panel on Climate Change

IPS: International Peatland society

ITA: International Trade Administration

ITCZ: The Inter-tropical Convergence Zone

IUCN: International Union for Conservation of Nature

L8: Landsat 8

LB: Lungui Bungu River

LULC: Land use land cover

m.asl: meters above sea level

MODIS: Moderate Resolution Imaging Spectroradiometer

NASA: National Aeronautics and Space Administration

NDVI: Normalised Difference Vegetation Index (NDVI)

NDWI: Normalised Difference Water Index

NGOWP: National Geographic Okavango Wilderness Project

NIR: Near Infrared

xxii

NOAA: National Oceanic and Atmospheric Administration

PDSI: Palmer Drought Severity Index

RCM: Regional Climate Model

RDI: Reconnaissance Drought Index

RF: Random Forest

RGB: Red, green, blue

RS: Remote Sensing/ Remotely Sensed

RVAA: Regional Vulnerability Assessment and Analysis Programme

S2: Sentinel 2

SASSCAL: Southern African Science Service Centre for Climate Change and

Adaptive Land Management

SF: Surface reflectance

SPI: Standardized Precipitation Index

SPEI: Standardized Precipitation Evaporation Index

SVM: Support Vector Machine

SWIR: Shortwave Infrared

SRTM: Shuttle Radar Topography Mission

TTT: Temperate Tropical Trough

UNESCO: United Nations Educational, Scientific and Cultural Organization

UNFCCC: United Nations Framework Convention on Climate Change

xxiii

USGS: United States Geological survey

WHO: World Health Organization

WMO: World Meteorological Organisation

WWF: World Wildlife Fund

CHAPTER 1: INTRODUCTION

1.1 Background

1.1.1 The Angolan Highlands

Angola has unique habitats and species and is one of the least recognised biodiversity

hotspots in the world (Myers et al., 2000; Huntley et al., 2019). Angola is source of

many major rivers in southern Africa and is referred to as the “water tower” of the

region (Huntley, 2019). These river systems originate from the interior Bié Plateau of

the Angola Highlands and flow into large river catchments such as the Cuanza, Cassai

(Congo), Lungui-Bungu (Zambezi), Cuito and Cubango (Okavango: Huntley, 2019).

The lack of ecological and environmental data and reporting for the upper catchments

of the Angolan Highlands is due to historic conflicts in the country (Huntley et al.,

2019). The Angolan War of Independence (1961-1974), the Angolan Civil War (1975-

2002) and resultant extensive minefields have prevented access and any scientific

exploration or study in the region for over 50 years (Conradie et al., 2016; Midgley and

Engelbrecht, 2019).

The Okavango Delta, a United Nations Educational, Scientific and Cultural

Organization (UNESCO) World Heritage site (International Union for Conservation of

Nature: IUCN, 2022), is dependent on precipitation occurring in the highlands of

central Angola, where water flows into the Okavango River from two tributaries: the

Cuito River and Cubango River (McCarthy et al., 2000; Gumbricht et al., 2004).

Concern of the threats to the upper catchments of the Angolan Highlands, and the

potential downstream consequences to the Okavango Delta, resulted in the

establishment of the National Geographic Okavango Wilderness Project in 2015

(NGOWP; Quammen, 2017). The aim of the NGOWP is to create a network of newly

protected areas to conserve the length of the Okavango Catchment (Quammen,

2017). The NGOWP has funded numerous surveys and led a series of expeditions to

document the biodiversity and ecology of the Angolan Highlands (Taylor et al., 2018;

Goyder et al., 2018; Skelton, 2019; Barber-James and Ferreira, 2019; Van Wilgen et

al., 2022). The findings of these expeditions, surveys and studies are intended to

provide a scientific foundation to inform the establishment of a system of protected

areas within the Angolan Highlands and for the entire Okavango Catchment (Van

Wilgen et al., 2022).

1.1.2 Peat

Exploratory surveys conducted by the NGOWP between 2015 and 2018 identified peat

deposits in the eastern Angolan Highlands (Conradie et al., 2016; Goyder et al., 2018).

Peat is a type of organic-rich soil that consists of partially decomposed organic matter

derived from plant material (International Peatland society: IPS, 2021). Peat forms

when the ground surface is waterlogged due to the interaction between landform,

climate, and vegetation (Lindsay, 2016). The term peatland refers to both the peat soil

and the terrestrial wetland ecosystem growing on its surface (Dargie et al., 2017).

Peatlands, representing at least one third of the global wetlands (Parish et al., 2008),

are important ecosystems for biodiversity conservation, carbon storage, climate

regulation, biomass production and human welfare (Erwin, 2009; Minasny et al.,

2019). Peatlands provide multiple ecosystem services that support the Sustainable

Development Goals (Food and Agriculture Organization: FAO, 2020) including

purifying water (Frolking et al., 2011; Evers et al., 2016), reduction of flooding and soil

erosion (Harenda et al., 2018), aiding in agricultural production and food security

(Page et al., 2011), and supporting biodiversity (Minayeva et al., 2017; Xu et al., 2018).

Peatlands are the largest natural terrestrial carbon store (Rieley and Page, 2016).

Climate change, land use land cover (LULC) change, peatland drainage and fire

because of human activities reduce peatland resilience to drought and fire and are

linked to peatland degradation, which releases the stored carbon into the atmosphere

(Harenda et al., 2018; Minasny et al., 2019). Efforts to reduce the anthropogenic

impact on climate has placed great importance on preserving natural carbon sinks

such as peatlands (Friedlingstein et al., 2019).

Peatlands are estimated to cover 3% of the global land surface (Page et al., 2011; Xu

et al., 2018). Although this is a relatively small area, the role of peatlands in the global

carbon cycle is significant (Harenda et al., 2018; Loisel et al., 2021). Globally, peat

and peatlands have been recognised as essential to multiple international conventions

that protect habitats, biodiversity, carbon sinks and reduce greenhouse gas (GHG)

emissions (FAO, 2020). The first Peatland Pavilion took place at the 2021 United

Nations Framework Convention on Climate Change (UNFCCC) Conference of Parties

(COP) 26 (IUCN, 2021). The UNFCCC includes peatlands (organic soils) in its Kyoto

Protocol and the Paris Climate Agreement (IUCN, 2021). The Intergovernmental Panel

on Climate Change (IPCC) has produced recommendations on national GHG

reporting and accounting from drained, rewetted and burning organic soils (IPCC,

2014). Despite its importance, there is no uniform definition for peat or peatland (IPS,

2021). Peatland conservation is dependent on accurate quantification of peatland

extent, status, and carbon stock (Minasny et al., 2019; FAO, 2020). Current estimates

of global and regional peatland carbon stock and extent are highly varied (Rieley and

Page, 2016; Xu et al., 2018). Digital mapping using field observations combined with

remotely sensed (RS) imagery and statistical models have been demonstrated to map

peatlands more accurately and decrease this uncertainty (Minasny et al., 2019).

1.1.3 Remote Sensing

RS products and Geographical Information Systems (GIS) tools and techniques have

been successfully utilised for multiple applications including LULC mapping (Mutanga

and Kumar, 2019), peatland mapping (Draper et al., 2014; Dargie et al., 2017;

DeLancey et al., 2019), vegetation (Huang et al., 2020), drought (AghaKouchak et al.,

2021) and fire-related studies (Parks et al., 2018; Bar et al., 2020). The information

obtained is used to aid scientists and policy-, and decision-makers (Opolot, 2013), and

provides data from inaccessible and extensive areas (Mutanga and Kumar, 2019). The

Geo Big Data problem is a view from geospatial and data scientists that require

technologies and resources capable of handling large volumes of satellite imagery

data (Shelestov et al., 2017). For example, the United States Geological Survey

(USGS) has been collecting global earth observation data with frequent intervals since

1972 through the Landsat program (Masek et al., 2020). The data record will continue

in future through the joint National Aeronautics and Space Administration (NASA)-

USGS Landsat 9 mission which was launched in September 2021 (Gross et al., 2022).

Open access to the entire Landsat archive was made available in 2008, however, this

archive has been underutilized due to the challenges in collecting, storing, processing,

and manipulating this multi-temporal and multi-spectral RS data (Shelestov et al.,

2017; Teluguntla et al., 2018). In addition, it is impractical to use common image

processing software on desktop PC-based systems on data that span large

geographic areas over five decades (Teluguntla et al., 2018; DeLancey et al., 2019).

Most recently, users have benefitted from the growing availability of high volume,

freely accessible RS data and the development of cutting-edge, easy to use machine

learning tools that have strong computing capacity (Shelestov et al., 2017;

Mahdianpari et al., 2017, 2019; Amani et al., 2020). These advances offer new

opportunities for applications at broader spatial and temporal scales, overcoming the

limitations of existing methods and products in geospatial sciences (Mutanga and

Kumar, 2019; Amani et al., 2020). The establishment of powerful cloud computing

infrastructure has been made available through multiple platforms such as NASA

Earth Exchange, Amazon’s Web Services, Microsoft’s Azure, and Google cloud

platform to address the Geo Big Data problem (Mahdianpari et al., 2019). For instance,

Google Earth Engine (GEE), a freely accessible, web-based cloud computing platform

that contains numerous geospatial datasets and satellite imagery, allows for algorithm

development and processes petabyte-scale data in good time (Hird et al., 2017).

1.2. Rationale

The Angolan Highlands are ecologically and hydrologically important (Van Wilgen et

al., 2022), supporting peatland deposits (Goyder et al. 2018), and are an essential

source water region for the Okavango Delta (Gumbricht et al., 2004) and southern

Africa more broadly. Access to the region has been hampered by historical conflicts

and persistent minefields, and as a result, it is under-studied (Taylor et al., 2018;

Goyder et al., 2018). Peatlands regulate global climate as undisturbed peatlands

prevent further climate change (Rieley and Page, 2016). Mapping peatland extent has

been documented to facilitate its preservation (FAO, 2020). Estimates of global

peatland extent and carbon stock are inconsistent and have high variability (Minasny

et al., 2019). This is evident in tropical zones where peatland deposits are

comparatively less documented outside of those in Europe and North America (Page

et al., 2007; 2011; Rieley and Page, 2016). To address this, the peatland extent in the

Angolan Highlands warrants quantification. During periods of drought, peatlands have

the potential to become net sources of carbon dioxide as aerobic respiration increases

when soil moisture and the water table decrease (Fenner and Freeman, 2011; Lund

et al., 2012; Jassey et al., 2018).

Having been well documented in the tropics, smouldering combustion of deep peat

soil leads to peatland degradation (Page et al., 2002; Page and Hooijer, 2016; Rieley

and Page, 2016), causing damage to the environment and human health (Marlier et

al., 2013). Undisturbed peatlands are resistant to smouldering combustion due to

waterlogging, burning of the surface vegetation does not result in peatland degradation

(Vetrita and Cochrane, 2019). Peatland functioning is directly influenced by vegetation

and hydrological conditions and is sensitive to drought and fire (Belyea and Malmer,

2004; Page et al., 2011). To address this, an assessment of both drought and the

vegetation response to drought, peatland fire regimes and peatland response to fire

necessitates investigation. The dynamics and of these peatlands have direct

implications on the environment and communities of the Angolan Highlands and

surrounding areas, and global climate more broadly.

1.3. Study Aim and Objectives

The primary aim of this research is to investigate dynamics relating to peatland extent,

age and growth, and peatland response to drought and fire in the Angolan Highlands.

To address the primary aim of this research, specific objectives of this research are:

1. To determine and map the extent of peatlands in the Angolan Highlands through

machine learning.

2. To determine the drought history and the vegetation response to drought in the

Angolan Highlands region through assessment of historical satellite data products.

3. To determine the contemporary fire regime of peatlands and peatland response to

fire in the Angolan Highlands using fire and vegetation satellite data.

1.4. Structure of the Thesis

This thesis is divided into 9 chapters. Chapter 1: Introduction, provides the

background, the rationale, and the study aim and objectives. Chapter 2: Study Region,

provides a broad synopsis of the Angolan Highlands study region, including

descriptions, maps and photographs of the topography, drainage characteristics, soils,

vegetation, and contemporary climate. The NGOWP Lungui Bungu River transect took

place in June 2022, none of the data collected are reported here, however,

photographs obtained during the expedition are presented in Chapter 2. Chapter 3:

Methodology, includes details regarding the use of GEE for RS data acquisition, and

provides an overview of the NGOWP Lungui Bungu River transect. Chapters 4-7

represent each of the accepted journal papers. Chapter 8: General Discussion,

explores three separate themes that cover the four research papers with reference to

academic literature, and includes a discussion of the main limitations of the research.

Chapter 9: Conclusion, reflects on the extent to which the research aim and objectives

have been achieved, the key findings and their significance, and possible avenues for

future work. A comprehensive reference list follows the concluding chapter and

contains all references from the journal papers and from the thesis chapters. In the

case of chapters 4-7, a reference list is included according to the reference style of the

respective journal the manuscript was submitted to. The supplementary files from each

journal paper appear in the appendices at the end of the thesis rather than at the end

of each journal paper.

CHAPTER 2: STUDY REGION

2.1 Introduction

In this chapter, a description of the World Wildlife Fund (WWF) ecoregions in Angola

is presented, followed by the geographic location, topographical and drainage

characteristics, soils and vegetation, and the contemporary climate of the study region.

Angola is located on the west-coast of south-central Africa, and the country spans

4°22’-18°02’ S, 11°41’-24° 05’ E (Huntley et al., 2019). Angola borders the Democratic

Republic of the Congo (DRC) to the north, Zambia to the east, Namibia to the south,

and the Atlantic Ocean to the west (Figure 2.1a). Angola covers a total surface area

of 1,246,700 km2 (Huntley et al., 2019), and has a population of 33,933,000 (2021),

with a population density of 27.22 people per km2 (World Bank, 2022). The WWF

terrestrial ecoregions is a global data product that reflects the zonation of the Earth’s

terrestrial biodiversity, defining ecological units that share similar environmental

conditions, species, and dynamics (Olson et al., 2001). The WWF terrestrial

ecoregions showcase the diverse biogeographic and ecological conditions across the

country (Figure 2.1b). Dominant ecoregions include the moist forest-savanna mosaics

in the north, in the west a 1,600 km coastline contains scarp savanna and woodlands,

and desert environments (Olson et al., 2001; Catarino et al., 2020). The miombo

woodland ecoregion dominates the central and eastern parts of the country and the

southern regions contain arid grasslands, savannas and woodlands along its 1,200

km border with Namibia (Olson et al., 2001; Catarino et al., 2020).

Figure 2.1. (a) Angola and its 18 provinces, and (b) the WWF ecoregions occurring in Angola (data from

Olson et al., 2001).

2.2. Topography and drainage

Angola has a heterogeneous topography (Figure 2.2a). In the west, coastal lowlands

below 200 m.asl occupy a band between 10–150 km in width and covers 5% of the

country (Huntley, 2019). Adjacent to these lowlands, a mountainous escarpment rises

to 1000 m.asl and covers 23% of Angola. The plateau is 1000-1500 m.asl and covers

65%, and areas above 1500 m.asl cover 7% of the country (Huntley, 2019). The study

area is in the southeast of Angola (Figure 2.2b). This area covers an estimated 61,590

km2 and spans 11°54′-13°54′ S, 18°05′-20°34′ E. The area has an elevation range of

1,117 m.asl in the southeast to 1,678 m.asl in the west.

Figure 2.2. (a) Elevation of Angola depicting the extent of the study site, and (b) elevation of the study

site including the WWF HydroSHEDS Basins Level 04 (data from Lehner and Grill, 2013) and major

rivers.

The study area is situated within the Zambezi-Cubango Peneplain which is a vast

generally flat area with rivers that meander down a gently dipping plateau (Huntley,

2019). The rivers originate from the northwest watershed with the Cuanza and Congo

Basins and flow towards the southeast, eventually reaching the Namibian and

Zambian borders (Figure 2.2). The study area forms part of the NGOWP core study

region which includes the upper sections of both the Okavango and Zambezi

watershed that originate in the Angolan Highlands. The NGOWP extracted peat cores

at the Cuito, Cuanavale and Cuando source lakes and near the Lungui Bungu River

source. The delineated study area is constrained to the mapped region of the journal

paper presented in Chapter 5. This journal paper is a classification map of peatland

extent based on the site locations of the peatland cores obtained during the NGOWP

expeditions. For consistency, the same study area delineation was used in the journal

papers presented in Chapter 6 and 7.

The drainage network forms part of the Angolan ‘water tower’, located in the central

part of the country (Huntley, 2019). This water tower has nine river basins, of which

seven are transnational (Huntley, 2019). The delineated study site includes six major

river catchment boundaries: the Lungui Bungu which is a tributary of the Zambezi

River and flows east into Zambia; the Cuando which flows south into Namibia,

Botswana, Zambia, and Zimbabwe; the Cuito and Cubango which are part of the

greater Okavango Catchment flowing southeast into Namibia and Botswana; the

Cuanza which flows northwest covering large parts of Angola and the Congo which

flows north into DRC. These rivers drain extensive and deep Kalahari sands, due to

the filtration caused by the action of the sand layer, the rivers have high clarity and low

nutrient levels (Huntley, 2019).

2.3 Soils and Vegetation

Over 75% of Angola is covered by two main soil groups, the sandy arenosols in the

east and the ferralsols in the west and central plateaus of the country (Jones et al.

2013: Figure 2.3). The arenosols are the dominant soil group, covering more than half

of Angola (Jones et al. 2013) and most of the study area. The arenosols form part of

the Kalahari Basin, the largest body of aeolian sand on earth, extending for 2,500 km

from the Cape to the Congo with a maximum width of 1,500 km (Garzanti et al., 2022).

These sands cover almost half of Angola, hiding the underlying geology (Huntley,

2019). The quartz grains have no mineral nutrients, little accumulated organic matter,

are infertile and have low water-holding capacity (Vainer et al., 2022). The Miombo

woodlands ecoregion is particularly well adapted to the arenosols and ferralsols (Olson

et al., 2001), and cover most of Angola (Huntley, 2019). The study area also contains

alluvial fluvisols within drainage lines, these have high water-retaining capacity and

organic content and are suitable for cultivation when not waterlogged (Jones et al.

2013; Huntley, 2019). In addition, gleysol clays are present and typically acidic,

waterlogged and occasionally extensive within floodplains (Jones et al. 2013; Huntley,

2019).

Figure 2.3. Soil map of Angola (from Jones et al. 2013).

Peat soil and peatland deposits were identified during exploratory surveys conducted

by the NGOWP (Conradie et al., 2016; Goyder et al. 2018). Tropical peatland deposits

have been defined as peatlands which are located between the Tropics of Cancer and

Capricorn, including lowland and upland peatlands (Page et al., 2007, 2011). Tropical

peatlands occur in Southeast Asia, the Caribbean and Central America, South

America, and Africa (Rieley and Page, 2016). Africa has a diversity of peatland

depositional environments that vary between sites, the majority of which are

minerotrophic (groundwater fed), reflecting the dry climate of the continent (Grundling

and Grootjans, 2016). Large ombrotrophic (rainwater fed) bogs exist in wet equatorial

regions such as the Congo Basin (Dargie et al., 2017; Davenport et al., 2020). The

Angolan Highlands peatlands have formed in upland valleys. Similar peatland

formation in upland valleys has also occurred in Rwanda (2,100 m.asl; Hategekimana

and Twarabamenya, 2007), Lesotho (2400 m.asl; Trettin et al., 2008), and Burundi

(1500 m.asl; Pajunen, 1996). In these high-altitude regions, the conditions for peat

formation are comparable to temperate regions (Andriesse, 1988; Page et al., 2011).

The Angolan Highlands peatlands have diverse depositional environments, forming in

lake margins (Figure 2.4a: Goyder et al., 2018), river floodplains (Figure 2.4b) and

river terraces. These peatlands are minerotrophic; with evidence from field

observations, drone (Figure 2.4c) and optical imagery, a distinct seep-line indicated

by a narrow band of white sand exists parallel to higher ground immediately adjacent

to the peatlands. This seep line is the inflow point of groundwater, the valleys are a

gentle V-shape, and the peatlands all contour downwards towards the river.

Figure 2.4. (a) The Cuito Source Lake, Moxico Province. Moist miombo woodland grows on the hillsides

adjacent to the lake. Peatland surrounds the source lake, and a narrow band of grassland grows

between the peatland and the miombo (from Goyder et al., 2018). Photograph D. Goyder. (b) Lungui

Bungu River Source, Moxico Province. Small pool of acidic water filtering out of the surrounding

peatland, miombo woodland surrounds the bowl-shaped peatland, and evidence of a recent fire in the

background. Photograph R. von Brandis. (c) Lungui Bungu River, Moxico Province. The seep line

indicated by a narrow band of white sand between the miombo and the floodplain environment,

evidence of small-scale farming within the floodplain environment towards the south of the drone

photograph. Photograph J. Guyten.

These peatlands are mostly waterlogged and dark in colour, forming in wet grasslands

(Figure 2.5a). The rural communities practice small-scale subsistence agriculture

which is reliant on rainfall and environmental conditions (Luetkemeier and Liehr,

2019). Communities often clear the miombo woodlands for cultivation, here the soils

(a) (b)

(c)

are nutrient poor and dry due to the high sand content and high infiltration rates (Hunter

and Crespo, 2019; Afonso et al., 2020). In addition, communities cultivate directly on

the peatlands (Figure 2.5b), where it is common practice to cut drainage lines and

burn the surface vegetation. These subsistence farmers produce low yields of corn,

wheat, rice, potatoes, beans, cassava, sugarcane, peanuts, sunflower, sesame, and

tobacco (Reyes et al., 2012). The woodlands and grasslands burn frequently (Figure

2.5c). Communities generally burn grasslands in the early dry season (June to July)

to aid in hunting practices and clear village surroundings, and woodlands are burnt

late in the dry season (August to September) to prepare fields for the growing season

(Meller et al., 2022; Teutloff et al., 2022).

Figure 2.5. (a) Cut section of a peat soil sample extracted by M. Lourenco with a Russian corer.

Photograph J. Guyton. (b) Burned (surface), cut, and drained peatland patch, cassava and lavender

growing in the background near a small village along the Lungui Bungu River. Photograph M. Lourenco.

The Angolan Highlands lie in the Angolan Miombo Woodland ecoregion which

contains grasslands, woodlands, savannas and shrublands (Huntley et al., 2019).

Barbosa (1970) described the vegetation as dense Zambesian and Congolian miombo

woodland with “chanas” or geoxylic-rich grasslands. Within southern Africa, Angola is

the least inventoried country for plants, and the east of the country has little to no geo-

(a) (b) (c)

referenced specimen data (Marshall et al., 2016). Goyder et al. (2018) provide a

vegetation checklist and a baseline of plant diversity for the Angolan headwaters, a

study that was part of the NGOWP series of expeditions. The species checklist (total

of 417) is categorised into four main vegetation types: moist miombo woodlands,

swamp forest, seasonally burned savannas, and wetlands (Goyder et al., 2018).

Miombo woodlands are tall, closed canopy woodlands dominated by trees in the

genera Brachystegia, Julbernardia, and Cryptosepalum (Goyder et al., 2018; Van

Wilgen et al., 2022). The miombo is extensive on the hillslopes and dominated by

Julbernardia. Miombo growing on the plateau is dominated by Cryptosepalum which

can form dense, closed canopy miombo forest rather than miombo woodland. In

general, the miombo forests do not have a flammable grass undergrowth that is

present in miombo woodland (Goyder et al., 2018). Large scale tree felling of the

miombo vegetation has occurred within the Cubango catchment, by contrast, the

vegetation in the Cuito catchment remains intact and homogeneous (Mendelsohn,

2019).

According to Goyder et al. (2018), swamp forest is rare and only occurs within Moxico

Province, unlike the miombo woodlands and the seasonally burned savannas which

are extensive. The seasonally burned savannas grow on highly leached Kalahari sand

and are described as high-rainfall grasslands (Goyder et al., 2018). The habitat is fire-

adapted and is dominated by grasses or geoxylic suffrutices, plants with large

underground woody biomass and seasonal above-ground shoots. The factors

controlling whether grasses or geoxylic suffrutices dominate are unclear. Maurin et al.

(2014) argue fire is an important evolutionary driver, whereas Finckh et al. (2016)

provide evidence that frost plays a principal role, with cold air pooling in valley bottoms

in the winter dry season and “burning” new shoots. Proximity to the water table limits

growth of trees within these savannas (Goyder et al., 2018). Collins et al. (2019)

classified these areas as open to sparse woodland and grasslands.

Historically, wetlands are under-sampled in Angola (Conradie et al., 2016). According

to Goyder et al. (2018), wetlands are typically not botanically diverse, and do not host

local endemics. The wide river valleys are characterised by extensive wet grasslands,

peatlands, and ox-box lakes (Conradie et al., 2016). These seasonal floodplains along

drainage lines are known locally as “chanas” (Van Wilgen et al., 2022) and more

commonly referred to as “dambos” across tropical zones of Africa (Midgley and

Engelbrecht, 2019; Skelton, 2019; Moore et al., 2022). The impeded drainage and

high precipitation in the rainy season cause temporarily waterlogged soils in the

valleys that support humid grassland borders with humic topsoil and dwarf shrubs and

prevent the development of miombo woodland (Conradie et al., 2016). The river

source lakes have peat accumulations at their margins and the river valleys also

contain peat deposits (Goyder et al., 2018). Although described as not botanically

diverse, the species checklist provided by Goyder et al., (2018) contains a total of 115

plant species identified in the wetland habitat of the Cuito catchment alone, 94 of which

are associated with peat soils.

2.4 Contemporary Climate and Weather

The climate and weather of Angola is diverse, owing to topographic, atmospheric, and

oceanic factors. There is a mean annual temperature difference of 4 °C (24.7 °C near

the Equator versus 20.7 °C near the Tropic of Cancer), decreasing from north to south

(Huntley, 2019). Temperatures generally decrease with altitude, driven by the

adiabatic lapse rate. There are major atmospheric systems over central and southern

Africa that influence rainfall patterns across the country (Nicholson, 2018). The Inter-

tropical Convergence Zone (ITCZ) is a belt of low pressure which circles the globe

near the equator (Schneider et al., 2014). This is where the trade winds from the

Southern and Northern Hemispheres converge, generating convective activity which

drives moist conditions that are typical of the tropics (Nicholson, 2018).

Seasonal shifts in the location of the ITCZ affects the precipitation of many equatorial

countries, resulting in distinct wet and dry seasons common in the tropics, rather than

cold and warm seasons that occur in higher latitudes (Schneider et al., 2014;

Nicholson, 2018). The ITCZ is on the equator during mid-autumn and mid-spring and

migrates towards the Tropic of Cancer in the austral winter. The wet season starts

when the ITCZ and the Congo Air Boundary shift over the north of Angola in early

summer and reach the south of the country in late summer (Huntley, 2019). Mean

annual rainfall is generally highest in the north and northeast (1680 mm/ year) and

decreases towards the southwest (100 mm/year), high mean annual rainfall (1500

mm/ year) occurs in the high-altitude regions near Huambo Province (Figure 2.6; Cain,

2017).

Figure 2.6. Mean annual rainfall in Angola (from Cain, 2017).

The climate is seasonal; hot and wet summers occur from October to May and mild to

cool and dry winters occur from June to September (Huntley, 2019). The Köppen-

Geiger climate classification (1980-2016) illustrates distinct climate zones in Angola

(Figure 2.7; Beck et al., 2018). Tropical, savanna (Aw) covers large expanses of the

central and northern parts of the country. The coastline is mostly arid, containing hot

deserts (BWh) in the southwest (Beck et al., 2018). The Angola and Benguela

Currents have a stabilising effect on the lower atmosphere and prevent cloud

formation, resulting in the evolution of the Namib Desert (Lima et al., 2019). There are

two high-pressure systems that shift along with the ITCZ over the Atlantic Ocean and

southern Africa: The Botswana Anticyclone and the South Atlantic Anticyclone. In

winter, the anticyclones block the southward movement of moist air and prevent cloud

formation. In summer, the anticyclones shift south and rainfall returns (Huntley, 2019).

During the months spanning winter through to early summer, strong east-west winds

are induced by the Botswana Anticyclone, picking up dust from the arid steppe and

contributing to the Namib dunes (Huntley, 2019). The study area is characterized by

Tropical, savannah (Aw), Temperate, dry winter, hot summer (Cwa) and Temperate,

dry winter, warm summer (Cwb) climates which span the plateau (Beck et al., 2018).

Figure 2.7. Köppen-Geiger climate classification map for Angola (1980-2016: adapted from Beck et al.,

2018).

Studies that focus on Angola’s climate have been hampered due to the collapse of an

extensive network of metrological stations (Huntley, 2019). Between 1974 and 2010,

Angola’s climate network was reduced from 225 climatological posts (Silveira, 1967)

to zero, and 29 synoptic stations to 23 stations (Government of Angola, 2013). This is

demonstrated in the study by Silveira (1967), a critical assessment of Angola’s climate,

that contained data from 184 stations across all 18 provinces. In addition, the study by

Azevedo (1972), to map and classify the climatic regions, was based on this

considerable national dataset, and still provides an accurate representation of Angolan

bio-climatic systems (Huntley, 2019). In the case of Peel et al. (2007), a more recent

study which provides an updated world map of the Köppen-Geiger climate

classification, only five temperature and sixteen precipitation stations were used for

Angola. Furthermore, records of extreme minimum temperatures and frost in Angola

are absent (Zigelski et al. 2019). These factors, in combination with fires and herbivory,

are important to the floristic diversity and composition of the country (Zigelski et al.

2019).

Due to the sporadic national meteorological station coverage and lack of reliable

climatic data, this study makes use of climatic data obtained from RS products (Figure

2.8). Historical climate data from weather stations, specific to the study area, are

unavailable. Distinct differences in precipitation and land surface temperatures occur

within the study area. Historical precipitation and land surface temperature data for the

region were collated from the GEE platform. Precipitation data were obtained from the

Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) 30+ year

quasi-global rainfall dataset (Funk et al., 2015). CHIRPS synthesises 0.05° resolution

satellite imagery with in-situ station data to create a gridded rainfall time series (Funk

et al., 2015). Land surface temperature data were obtained from the Terra Moderate

Resolution Imaging Spectroradiometer (MODIS) Land Surface Temperature/

Emissivity 8-Day (MOD11A2) Version 6.1 product (Wan et al., 2021). This product

provides an average 8-day per-pixel land surface temperature and emissivity with a 1

km spatial resolution in a 1,200 by 1,200 km grid (Wan et al., 2021).

Figure 2.8. (a) Average precipitation (1981-2020) and (b) Average day time land surface temperature

(2000-2020) for the delineated study area.

The mean annual precipitation over the study area is 1113 mm for the period 1981-

01-01 to 2020-12-31. The highest annual precipitation (1295 mm) is recorded in the

northwest and northeast regions of the study area (Figure 2.8a). Precipitation in the

region is closely related to latitude and altitude. Both precipitation and altitude

decrease towards the southeast where the mean annual precipitation is lowest (904

mm). The landscape has distinct topographical features dominated by river processes;

high elevations exist adjacent to low lying river valleys. Land surface temperatures

reflect these topographical features and are closely related to vegetation (Figure 2.8b).

On average, miombo woodlands which grow on high elevations have cooler (24.5 °C)

land surface temperatures in comparison to the warmer valleys at lower elevations

(33.6 °C).

2.5 Conclusion

Angola is a large country on the western coast of southern Africa. The important

components of Angola’s diverse ecoregions, topographical and drainage

characteristics, soils, vegetation, and contemporary climate are detailed. These range

from the Namib desert in the southwest, to the arid savannas of the coastal and

southern areas, to the steep Angolan escarpment. Above the escarpment, high

mountains contain Afromontane grasslands and forests. These mountainous areas

are the source of several large rivers which drain into multiple neighbouring countries.

Extensive miombo woodlands dominate the plateau and peneplains. Mean annual

rainfall varies from below 100 mm in the southwest to over 1600 mm in the north of

Angola. The precipitation and temperature across the country are closely linked to

global and regional synoptic systems creating distinct wet and dry seasons. In

addition, the topography and ocean currents have an influence on climate. Within the

study area, the landscape contains distinct vegetation types, dominated by miombo

woodlands. The peatlands which grow in the river valleys are controlled by multiple

factors including groundwater, climate, topography, and vegetation. These peatlands

provide extremely important ecosystem services as they purify water, limit river flow,

provide fertile soil for agriculture, are a fuel source, and store carbon. Although the

study area is rural, the strong anthropogenic influence on fire, wetland drainage and

tree clearing are a threat to this region. This underscores the importance of

investigating the peat dynamics within this area of Angola.

CHAPTER 3: METHODOLOGY

3.1 Introduction

This chapter provides a brief methodological background to the use of Google Earth

Engine in remote sensing studies. In addition, an overview of the NGOWP Lungui

Bungu River transect conducted in June 2022 is included. The specific analytical

methods used in the analysis of peat extent, of drought and fire dynamics are detailed

in each of the respective journal papers.

3.2 Remote Sensing

The Angolan Highlands are an extensive, understudied area that is largely

inaccessible because of landmine presence following historic conflicts (Taylor et al.,

2018; Goyder et al., 2018). Since the inception of the NGOWP in 2015, although

conducting the most extensive scientific exploration across the region for 50 years

(Conradie et al., 2016; Midgley and Engelbrecht, 2019), access to the region remains

difficult and costly. RS has several advantages over traditional mapping and data

collection approaches, field measurement techniques are limited by accessibility,

scale, and cost, whereas remote sensing is quicker, more cost effective, and provides

detailed information at a regional scale (Lees et al., 2018; Wu et al., 2014). Satellite

data cover areas that are inaccessible and the repeat data collection enables effective

environmental monitoring of dynamic phenomena such as precipitation, vegetation

growth and fire (Mutanga and Kumar, 2019; Shiklomanov et al., 2019).

There are a wide range of RS datasets available that are often used in combination:

multi-sensor approaches are highly useful for a variety of environmental monitoring

research topics and have been used to identify and discriminate peatland from other

wetland features (Hird et al., 2017; Mahdianpari et al., 2019; DeLancey et al., 2019).

The availability of big data from earth observation products and the advances in

machine learning, has provided further opportunities for new methods to aid in earth

environmental monitoring (Gorelick et al., 2017; Amani et al., 2020; Yuan et al., 2020).

Over the last decade, substantial progress in earth sciences has been observed, these

new advances have outperformed traditional RS models with considerable

improvement in performance (Tamiminia et al., 2020; Yuan et al., 2020).

3.3 Google Earth Engine

In recent years, following the significant advancement in sensors and increase in the

number of RS datasets, multiple challenges to users working with big data have been

created (Chi et al., 2016; Tamiminia et al., 2020). The challenges include big data

computing, collaboration, and methodologies, as well as the appropriate data

identification, deployment, representation, fusion, visualisation, and interpretation (Chi

et al., 2016; Gorelick et al., 2017; Amani et al., 2020). The development of a safe,

efficient, and advanced cloud computing platform was one of the most important

requirements to provide a comprehensive solution to these challenges (Gorelick et al.,

2017; Amani et al., 2020). Cloud computing platforms provide infrastructure, storage

services, datasets, and software packages that enable the performance of a

supercomputer on a standard device (Yuan et al., 2020; Amani et al., 2020).

GEE is a cloud computing platform launched by Google in 2010 (Tamiminia et al.,

2020). GEE uses Google’s computational infrastructure and has a large collection of

open access satellite imagery and RS datasets that are constantly updated (Gorelick

et al., 2017; DeLancey et al., 2019; Tamiminia et al., 2020). GEE is freely accessible

and the most popular big data processing platform (Amani et al., 2020). Users can

access GEE through an internet-based Application Programming Interface (API) and

a web-based interactive development environment (Mahdianpari et al., 2019).

Furthermore, users do not need to download the large datasets and GEE has an

automatic parallel processing and fast computational platform that effectively deals

with big data at a petabyte scale (Gorelick et al., 2017; Amani et al., 2020). In addition

to its computing and storage capacity, several well-known machine learning algorithms

have been implemented, and users are able to develop and share their own algorithms

easily (Chi et al., 2016; Tamiminia et al., 2020). The data outputs and map exports are

complimentary to existing RS and GIS software packages (Chi et al., 2016; DeLancey

et al., 2019).

Considering the trends of GEE use in research over the recent past, it is projected that

users will more frequently use GEE as the main cloud computing service in future

(Tamiminia et al., 2020). In this study, GEE was used as the primary platform for RS

data extraction and storage and used in some analytical approaches. Map

visualisations and calculations were performed using ArcMap software. Sample GEE

scripts were obtained from the dedicated GEE developers guide and API Javascript

tutorials. These were revised for the study site and specific date ranges for each

journal paper chapter. During this study, an ethics clearance was not needed as the

data collected was solely obtained from freely accessible RS datasets on GEE.

3.4 NGOWP Lungui Bungu River expedition

To ensure long-term, sustainable protection for the Okavango watershed, the NGOWP

has been conducting surveys, gathering scientific data concerning the river systems

and collaborating with local communities, non-government organisations and the

governments of Angola, Namibia, and Botswana (National Geographic Society, 2022).

Since 2015, the NGOWP have completed multiple expeditions, performing river

transects across the length of the catchment from the source waters in the Angolan

Highlands to the Makgadikgadi Pan in Botswana (National Geographic Society, 2022).

The NGOWP core study region has since expanded to other major river basins

originating from the Angolan Highlands including the Zambezi catchment. The first

NGOWP expedition conducted in Angola following the COVID-19 global pandemic

took place in June 2022. The transect for this field work as part of the PhD, started at

the Lungui Bungu River source, and continued east, along the river, until the Angola-

Zambia border. The Lungui Bungu River is a major tributary to the Zambezi River.

Prior to this expedition, observations of the landscape were limited to RS satellite

datasets as this was the candidates first visit to the study area. In previous NGOWP

expeditions, peat cores were collected on the river floodplain near the Lungui Bungu

River source. Following Accelerator Mass Spectrometry (AMS) radiocarbon dating,

important data regarding peatland formation was obtained. With this background

knowledge, a peatland classification map for the region was generated prior to this

expedition. The Lungui Bungu River transect provided an opportunity to validate this

peatland classification. In addition, observations of the peatland landscape revealed

further insight of the threats to these deposits and provided additional verified peatland

extent data that was used in Chapter 7. The specific details regarding the data and

methodologies used are mentioned in each journal paper chapter.

The principal aim of this expedition was to conduct the first known scientific exploration

of the length of the Lungui-Bungu River. Selected outcomes and achievements of the

expedition are described in this thesis. However, the primary results of this expedition

fell outside of the scope of this work and therefore do not appear in this thesis.

Importantly, this provided the opportunity to visit the study site and to understand the

context in which the RS work was done. Outcomes specific to the peatlands include

an additional 17 core samples of peat soil that were collected along the length of the

river. The cores were collected using a Russian peat corer and have been radiocarbon

dated at the iThemba LABS. In addition, water quality data using a multiparameter

probe and water discharge using an acoustic doppler current profiler were collected

every 10 km along the river length. Biodiversity counts, evidence of human interaction

with the river (for example, fishing nets) and fire events near the river were recorded

along the transect. River invertebrate biodiversity assessments were performed every

30 km along the river.

CHAPTER 4: PEAT DEFINITIONS: A CRITICAL REVIEW

4.1 Brief synopsis

This critical review presents a brief history on peatland definitions through time. The

review highlights the current state and discrepancies between definitions concerning

peatland nomenclature and classifications. Multiple disparate definitions are

presented with specific focus on the criteria concerning minimum depth, organic

carbon, organic matter, and ash content. This study presents a new definition for peat,

the motivation behind this definition is for conservation purposes and from the

perspective of climate science, preservation, and carbon accounting. This is an

important contribution as the literature concerning peat and peatland definitions is

inconsistent, causing multiple implications, including the quantification of global

peatland carbon stock.

Author contributions:

Contributor role Author contribution

Conceptualisation Lourenco M, Fitchett JM, and Woodborne S

Methodology Lourenco M, Fitchett JM, and Woodborne S

Validation Lourenco M

Formal analysis Lourenco M

Investigation Lourenco M

Resources Lourenco M

Data curation Lourenco M

Writing – the original draft preparation Lourenco M

Writing – review and editing Lourenco M, Fitchett JM, and Woodborne S

Visualisation Lourenco M

Supervision Fitchett JM and Woodborne S

Project leaders Fitchett JM and Woodborne S

According to the above-mentioned CRediT system the author contribution was

calculated as follows:

Lourenco M – 70%

Fitchett JM and Woodborne S – 30%

Submitted to: Progress in Physical Geography

Impact factor for 2021-2022: 4.283.

Submission date: 29 March 2022

Revision date: 15 July 2022

Accepted: 21 July 2022

Published online: 4 October 2022

Available from: https://journals.sagepub.com/doi/full/10.1177/03091333221118353

The version to follow is exactly as published in the journal on 4 October 2022, with

the addition of contiguous page numbers for this thesis.

https://journals.sagepub.com/doi/full/10.1177/03091333221118353

CHAPTER 5: ANGOLAN HIGHLANDS PEATLANDS: EXTENT, AGE

AND GROWTH DYNAMICS

5.1 Brief synopsis

This study presents the first classification of the Angolan Highlands peatlands,

including a map of the peat extent. In addition, analysis of the age and growth

dynamics of these deposits are presented. This is a multi-disciplinary study that

incorporates RS datasets, machine learning and GEE with AMS radiocarbon dates of

peat cores. Tropical peatland deposits are poorly mapped and documented within the

literature. This is an important contribution as the study provides a first conservative

estimate of the extent of these peatlands, contains details of peatland characteristics

with respect to topographical data and RS indices concerning vegetation and standing

water and suggests possible dates for peatland growth initiation, peatland growth

dynamics, potential carbon storage, and threats to this deposit.

Author contributions:

Contributor role Author contribution

Conceptualisation Lourenco M, Woodborne S, and Fitchett JM

Methodology Lourenco M, Fitchett JM, and Woodborne S

Validation Lourenco M

Formal analysis Lourenco M, Fitchett JM, and Woodborne S

Investigation Lourenco M

Resources Lourenco M

Data curation Lourenco M

Writing – the original draft preparation Lourenco M

Writing – review and editing Lourenco M, Fitchett JM, and Woodborne S

Visualisation Lourenco M

Supervision Fitchett JM and Woodborne S

Project leaders Woodborne S and Fitchett JM

According to the above-mentioned CRediT system the author contribution was

calculated as follows:

Lourenco M – 65%

Fitchett JM and Woodborne S – 35%

Submitted to: Science of the Total Environment

Impact factor for 2021: 10.753

Submission date: 14 July 2021

Revision dates: 20 October 2021 and 1 December 2021

Accepted: 7 December 2021

Published: 13 December 2021

Available from:

https://www.sciencedirect.com/science/article/pii/S0048969721073915

The version to follow is exactly as published in the journal on 13 December 2021, with

the addition of contiguous page numbers for this thesis.

https://www.sciencedirect.com/science/article/pii/S0048969721073915

CHAPTER 6: DROUGHT HISTORY AND VEGETATION RESPONSE

IN THE ANGOLAN HIGHLANDS

6.1 Brief synopsis

In this study, an investigation of the drought dynamics for the Angolan Highlands are

presented from the analysis of a 40-year CHIRPS precipitation and a 20-year MODIS

vegetation growth record. The study reveals the contemporary seasonality of

precipitation and vegetation growth for the region. It presents the 40-year drought

history and the describes the relationship between drought and ENSO and regional

synoptic systems. The implications of drought are discussed in this study, with specific

reference to peatland vegetation growth under drought conditions and future climate

scenarios. This is an important contribution as the study provides the first historical

assessment of drought in the Angolan Highlands, a region which contains vulnerable

rural populations that are dependent on rain-fed subsistence agriculture.

Author contributions:

Contributor role Author contribution

Conceptualisation Lourenco M, Fitchett JM and Woodborne S

Methodology Lourenco M, Fitchett JM, and Woodborne S

Validation Lourenco M

Formal analysis Lourenco M

Investigation Lourenco M

Resources Lourenco M

Data curation Lourenco M

Writing – the original draft preparation Lourenco M

Writing – review and editing Lourenco M, Fitchett JM, and Woodborne S

Visualisation Lourenco M

Supervision Fitchett JM and Woodborne S

Project leaders Fitchett JM and Woodborne S

According to the above-mentioned CRediT system the author contribution was

calculated as follows:

Lourenco M – 70%

Fitchett JM and Woodborne S – 30%

Submitted to: Theoretical and Applied Climatology

Impact Factor for 2021: 3.409

Submission date: 28 February 2022

Revision dates: 31 May 2022, 3 September 2022, and 31 October 2022

Accepted: 31 October 2022

Published online: 10 November 2022

Available from: https://link.springer.com/article/10.1007/s00704-022-04281-4

The version to follow is exactly as published in the journal on 10 November 2022,

with the addition of contiguous page numbers for this thesis.

https://link.springer.com/article/10.1007/s00704-022-04281-4

CHAPTER 7: FIRE REGIME OF PEATLANDS IN THE ANGOLAN

HIGHLANDS

7.1 Brief synopsis

Using MODIS fire and vegetation data in combination with the LULC map presented

in Chapter 5, this study presents the first assessment of the contemporary fire regime

of peatlands in the Angolan Highlands. Fire and associated degradation have been

extensively documented in tropical peatland deposits in southeast Asia, with little to

no reporting for African tropical peatlands. This is an important contribution as the

study provides a first assessment of the fire regime of peatlands, including the

implications of fire for peatlands, and discusses fire management, future research, and

area conservation for the Angolan Highlands.

Author contributions:

Contributor role Author contribution

Conceptualisation Lourenco M, Fitchett JM, and Woodborne S

Methodology Lourenco M, Fitchett JM, and Woodborne S

Validation Lourenco M

Formal analysis Lourenco M

Investigation Lourenco M

Resources Lourenco M

Data curation Lourenco M

Writing – the original draft preparation Lourenco M

Writing – review and editing Lourenco M, Fitchett JM, and Woodborne S

Visualisation Lourenco M

Supervision Fitchett JM and Woodborne S

Project leaders Fitchett JM and Woodborne S

According to the above-mentioned CRediT system the author contribution was

calculated as follows:

Lourenco M – 75%

Fitchett JM and Woodborne S – 25%

Submitted to: Environmental Monitoring and Assessment

Impact factor for 2021: 3.307

Submission date: 22 August 2022

Revision date: 19 October 2022

Accepted: 25 October 2022

Published: 7 November 2022

Available from: https://link.springer.com/article/10.1007/s10661-022-10704-6

The version to follow is exactly as published in the journal on 13 December 2021,

with the addition of contiguous page numbers for this thesis.

https://link.springer.com/article/10.1007/s10661-022-10704-6

100

101

102

103

CHAPTER 8: GENERAL DISCUSSION

8.1 Introduction

This chapter presents the general discussion, divided into three themes cross cutting

the four research papers (Chapters 4-7), namely peat preservation in the Angolan

Highlands, environmental change in the Angolan Highlands, and concerns for the

Angolan Highlands region. In addition, this chapter integrates and expands on the

findings and connections between each research paper with reference to literature.

During the PhD, Chapter 5 was the first to be published, followed by Chapter 4.

Chapters 6 and 7 were in review at the same time and published shortly after one

another. Although having similar vegetation, precipitation, and fire time-series records,

the journal articles that appear in Chapters 6 and 7 were submitted as standalone

research papers. A discussion of the limitations of this study is included in this chapter.

8.2 Peat preservation in the Angolan Highlands

Peatlands have been identified in at least 175 countries, occurring in polar and tropical

regions and every climatic zone (Lindsay, 2010; Bain et al., 2011; IPS, 2020). Due to

available resources and widespread peatland coverage, peatlands have been mapped

extensively in the northern hemisphere; however, shortfalls exist in adequate peatland

distribution maps both for the southern hemisphere and the tropics (Page et al., 2011;

Evans et al, 2014). As a result, the dissemination of global peatland distribution maps

is biased towards the northern hemisphere (Wu et al., 2017; Xu et al., 2018). Global

peatland maps and estimates of peatland extent are also inconsistent as they are

produced from several differing data sources at global, regional, and national levels

104

(Wu et al., 2017; Xu et al., 2018). These collection approaches are both dependent

and constrained to high quality peatland extent data that are only available for a small

selection of countries and regions, including Canada, Sweden, and West Siberia (Wu

et al., 2017).

The first peatland map of the Angolan Highlands (Chapter 5) provides a conservative

estimate (1,634 km2) of peatland extent for the region. The best estimate (2,640 km2)

of peatland extent for Angola reported by Page et al. (2011) makes no mention of

these deposits, and it is likely that the estimate is an addition to peatland extent

reported for Angola. This estimate is a crucial first step in providing the peat carbon

inventory and facilitates conservation efforts for the region and surrounding basins. In

addition, it provides updated peatland data for Angola that may be used in updated

estimates of global peatland extent.

Peatlands store more carbon than the above-ground carbon stored in all the world’s

forests (Page et al., 2011; Bain et al., 2011). Despite this, peatlands have received

less attention under the UNFCCC (Ramsar, 2002; Evans et al, 2014), and have only

recently received substantial recognition at the 2021 UNFCCC COP26 (IUCN, 2021).

The focus of the literature over the last decade has steered toward peatland

preservation (Finlayson and Milton, 2018; Rieley and Page, 2016). This is facilitated

by accurate classification of peatland extent (Minasny et al., 2019; FAO, 2020) and

quantification of peatland carbon (Law et al., 2015; Xu et al., 2018). In Chapter 4, the

study extends this focus, stating that discrepancies in peat and peatland definitions

are negatively influencing efforts in quantifying global peat carbon stock. In this study,

a new definition for peatland is proposed, motivated by climate science, preservation,

105

and carbon accounting. The implication is that a common definition for peat and

peatland will also contribute towards its preservation. Chapter 5 does not include the

conservative definition proposed in Chapter 4 as it was written and accepted before

the comprehensive review of peat and peatland definitions.

There are several anthropogenic practices that threaten the preservation of the

Angolan Highlands peatland deposits, including extensive fires, slash and burn

agriculture (also known as shifting agriculture), peat fuel extraction, wetland drainage,

and overgrazing (Conradie et al., 2016; Taylor et al., 2018). As calculated in Chapter

7, the peatlands burn more frequently, have the smallest proportion of area which does

not burn, and have the largest average proportion of burnt area per year among all the

LULC classes. Peatland functioning is dependent on continual recycling of organic

matter under waterlogged, anaerobic conditions (Lawson et al., 2015). Conditions

where either vegetation is removed, or the water table is lowered are major threats to

peatland growth and functioning (Rieley and Page 2016).

Human interference on peatlands can be highly detrimental (Brevik and Homberg,

2004; Hooijer et al., 2010) as drainage, peat extraction, inappropriate burning, and

conversion to agriculture all lower the water table (Bain et al., 2011; Page et al., 2011).

The more intensively the peatland is disturbed, the quicker the peat degrades, oxidises

and releases GHGs (Bain et al., 2011; Evans et al., 2014; Lawson et al., 2015).

Compromised peatlands become a major source of GHG emissions which may

continue for many years until all the peat is lost (Joosten 2011; Bain et al., 2011). The

combination of extensive area covered by peatlands and potential long-term emissions

makes the climate effect of degraded peatlands fundamentally distinct from other

106

ecosystems (Joosten 2011; Bain et al., 2011), requiring consistent monitoring and

unique preservation strategies (FAO, 2020).

Peatlands are sensitive to temperature, precipitation, and prolonged periods of

drought (Belyea and Malmer, 2004). In Chapter 6, drought events in the region are

shown to increase through time and negative SPI trends indicate that drought

occurrence and frequency are likely to increase in future. The vegetation response to

drought was calculated for the study area, and more specifically for the vegetation

within the valley environment which contains much of the peatlands classified in

Chapter 5. Under drought conditions, both soil moisture and the water table decrease,

facilitating the increased potential for aerobic respiration, causing peatlands to become

net sources of CO2 (Fenner and Freeman, 2011; Lund et al., 2012; Jassey et al.,

2018). The relationship between drought and ecosystem response is often non-linear

(Jassey et al., 2018), however, tipping points have been identified for individual peat

forming species in response to drought, when reached these thresholds that are

detrimental to the ecological functioning of the peatland (Lund et al., 2012;

Lamentowicz et al., 2019).

Peatlands rarely burn when waterlogged (Rieley and Page, 2016), however, if

desiccated due to drought and or drainage, they are more easily combustible (Vetrita

and Cochrane, 2019). Peatland loss and degradation due to the combustion of deeper

peat has been documented in boreal and tropical peatlands across the globe (Rieley

and Page, 2016; Turetsky et al., 2015). The results presented in Chapter 7 suggest

that peatland burning within the five intact peatland sites are surface vegetation fires,

where the above-ground peat forming vegetation is burnt and recovers shortly after

107

burning. Smouldering (deep peat) fires would likely result in peatland degradation

(Vetrita and Cochrane, 2019), where the surface vegetation would not be able to

recover post burn and continue to burn year on year. It is critical that these peatlands

remain undisturbed and are protected because they have a natural resistance to

smouldering combustion and degradation, protecting the deeper peat layers that have

accumulated carbon over thousands of years. Within undisturbed peatlands

throughout the world, most of the peatland carbon stock is typically protected from

smouldering, but drying because of human activity, drought and climate change lowers

the water table, exposing the deeper peat to smouldering combustion and degradation

(Vetrita and Cochrane, 2019).

According to Regional Climate Models (RCMs) for Angola, the Angolan Highlands are

projected to become both drier (decrease in precipitation of up to 4%) and warmer

(increase in air temperature of up to 4.9 °C) by 2100 (Carvalho et al., 2017). Continued

population growth, subsistence agriculture and human pressure on the land are

increasing in unprotected rural areas of Angola such as the Angolan Highlands

(Catarino et al., 2020). Local communities are likely to target the moist and organic

rich peatlands for cultivation in a warmer and drier climate future (Humpenöder et al.,

2020), lowering the water table and exposing deeper peat to fire and degradation.

Multiple countries have developed national peatland strategies to promote

preservation to ensure continued functionality (Andersen et al., 2017). Standard

peatland management practices include the promotion of peat forming vegetation,

prevention of water loss and water pollution and reduction of peat extraction,

cultivation, and drainage (Dohong et al., 2017). Context-driven strategies should be

108

considered with the input of local ecological and indigenous knowledge, communities,

and stakeholders (Van Noordwijk et al., 2014; Fleming et al., 2021).

8.3 Environmental change in the Angolan Highlands

Over the last decade, unprecedented changes in the human and biophysical

environments have occurred (Loisel et al., 2021). These include the increase of fossil

fuel emissions and rising global temperatures, which place greater importance on

carbon sinks (Friedlingstein et al., 2019). Reducing the atmospheric abundance of

carbon dioxide (CO2) and other GHGs is required to mitigate the risks of climate

change (Houghton, 2002; Archer et al., 2009). Peatlands contribute towards this

mitigation as they are an important natural reservoir of global carbon (Ramsar, 2002).

In the 6th Assessment report of the IPCC, a headline statement (high confidence)

states that: “Under scenarios with increasing CO2 emissions, the ocean and land

carbon sinks are projected to be less effective at slowing the accumulation of CO2 in

the atmosphere” (IPCC, 2021: 36). Under the SSP1-1.9 (very low GHG emissions)

scenario, the total cumulative CO₂ emissions taken up by land and ocean is projected

to be 70% in 2100 (remaining 30% stays in the atmosphere). However, under the

SSP5-8.5 (very high GHG emissions) scenario, this uptake reduces to 38% as ocean

and land carbon sinks become less effective, meaning that there is likely to be a higher

proportion of emitted CO2 remaining in the atmosphere (IPCC, 2021).

In Chapter 5, AMS radiocarbon dates of these peatlands reveal that peatland initiation

began at least ~ 7100 cal. yr BP; the implications are that these deposits may store a

significant amount of carbon and have the potential to be an important part of the

global carbon budget (Goyder et al., 2018). If these peatlands are disturbed or

109

degraded, carbon intake will be reduced, ultimately intensifying CO2 concentrations

and further contribute to climate change, influencing the deposits in a negative, self-

perpetuating cycle (Brevik and Homberg, 2004; Joosten 2009, 2011). In addition, if

strong mitigation took place and GHG concentrations in the atmosphere were

drastically reduced below current levels, historical emissions of long-lived GHGs will

remain significant to future contributions of warming due to past accumulation and the

inertia of the climate system (Skeie et al., 2021), and continue to impact these

peatlands in future. Therefore, initial monitoring and preservation of the deposits are

necessary in an uncertain climate future that depends on intact, natural carbon sinks

(Friedlingstein et al., 2019).

Peat is a soil distinguished from other soil types owing to the build-up of organic matter

due to the combination of plant growth and waterlogging (Lindsay, 2010). Peat is

therefore a direct product of the vegetation growing on the surface, reflecting prevailing

hydrological and nutrient conditions controlled by climate and underlying landforms

(Lindsay, 2010; Bain et al., 2011). In Chapter 6 and 7, valley vegetation EVI and NDVI

and peatland NDVI indicated that vegetation growth has the same seasonality as the

precipitation in the region. The implications are that vegetation occurring in the valley

environment (which supports most of the peatland deposits classified in Chapter 5) is

more strongly correlated with seasonal precipitation, and its growth is hydrologically

dependent.

Chapter 6 demonstrates that the Angolan Highlands vegetation was buffered against

the drought periods that occurred since 2001. The NDVI and EVI timeseries indicates

that the valley vegetation, which supports the peatlands, recovered at the start of each

110

rainfall season, a pattern which was identified even after the driest rainfall season on

record during 2018/2019. In addition, although Chapter 7 did not include the drought

data obtained from Chapter 6, the NDVI timeseries of the five intact peatland sites

does overlap these drought periods. The peatland NDVI timeseries provide further

evidence that the peatland vegetation is buffered against drought conditions,

recovering at the start of each rainfall season even during drought periods. It is

important that these peatlands remain undisturbed, as human activities would likely

contribute to peatland degradation, limiting the peatland natural resistance to both

drought and fire.

In comparison to the valley vegetation region, miombo vegetation has peak greenness

three months after peak precipitation. The ecological importance of the miombo

woodlands cannot be overstated (Chiteculo and Surovy, 2018). The precipitation

which falls on the upland environment, covered by Kalahari sand, is buffered by the

miombo, and it is the interaction between the groundwater filtering through the sand

and the adjacent seep line that allows and sustains peatland growth in this region, a

mechanism for peat growth that has likely been present for millennia. In addition, the

peatlands limit river flow and are a control valve between groundwater flow and the

river as described in Chapter 5.

The use of RS datasets provides contemporary environmental change indicators in a

region which is historically war stricken and has had limited accessibility for scientific

research (Conradie et al., 2016; Midgley and Engelbrecht, 2019). Chapters 6 and 7

demonstrate both direct and indirect anthropogenically induced environmental

changes in the landscape. Over the 20-year fire record (2001-2020) presented in

111

Chapter 7, four common drought periods were identified within Chapter 6, of which

three occurred since 2014. The two highest (2017 and 2019) annual burn area totals

for the study area coincide with the 2017-2018 and 2018-2020 droughts, with 2019

having both the lowest seasonal rainfall total and highest burn area on record since

1981 and 2000, respectively.

Increased burn area is likely due to the increased volume of dry material across the

landscape, making fires more prevalent during periods of low rainfall and drought

(Laris et al., 2016). Although fire occurrence could not be directly attributed to humans,

the total burn area across the region highlights the potential influence of rural

communities on fire, with an increase in burn area during 2003 following the end of the

Angolan Civil War and a decrease during the 2020 lockdown associated with the

COVID-19 pandemic. Precipitation deficits accounted for in the last 40 years indicate

that rainfall has a negative trend, with the possibility of increased drought occurrence

in future, consistent with what has been documented in Angola (Brooks et al., 2005;

Cain, 2015; Carvalho et al., 2017) and throughout southern Africa (Abiodun et al.,

2019; Nhamo et al., 2019; Gore et al., 2020). As a result of anthropogenic climate

change, southern Africa, a region that is already characterised as dry and hot (Geppert

et al., 2022), is projected to become generally drier under low-mitigation climate

futures (Archer et al., 2018).

8.4 Concerns for the Angolan Highlands region

Temperatures over southern Africa have increased rapidly over the last five decades,

at a rate of near twice the global rate of temperature increase (Archer et al., 2018;

112

Gore et al., 2020). Under low-mitigation futures, further increases, up to 6°C by the

end of the century relative to the present-day climate, may occur over the central and

western interior regions (Engelbrecht et al., 2009; Archer et al., 2018; Abiodun et al.,

2019). In projected changes of annual rainfall over southern Africa for the period 2080-

2099 compared to present day (1971-2000), the largest rainfall decreases are

projected for Angola and the southern parts of South Africa (Archer et al., 2018). The

projected decreases in Angola may be occu