A GEOSPATIAL APPROACH TO MAPPING 

JACARANDA TREE DISTRIBUTION IN 

JOHANNESBURG, SOUTH AFRICA.  

 

 

Rohini Chelsea Reddy (1836015)  

November 2023 

Supervisor: Prof. Jennifer Fitchett  



ii 
 

Declaration 

 

I declare that this dissertation is my own, original work, except where otherwise acknowledged. 

It is being submitted for the Degree of Master of Science at the University of the Witwatersrand, 

Johannesburg. It has not been submitted previously for any degree or examination, at this, or 

any other university.   

 

 

 

 

________________________ ______     24 November 2023_ 

Rohini Chelsea Reddy       Date 

 

 

 

 

 



iii 
 

Abstract 

Accurate mapping of the spatial distribution of invasive species is vital for the implementation of 

effective monitoring and management strategies. In countries where resources are scarce and 

costly, citizen science provides a cost-effective and accurate alternative for large-scale data 

collection. Citizen’s familiarity with their environment contributes to aspects such as accurate 

identification of features on the landscape. Advances in a geographic information system (GIS) 

together with open-sourced photography from Google Street View, provide accurate methods 

for in-field and remote validation of citizen science data for invasive mapping and assists with the 

creation and compilation of maps to visualize the spatial distribution of invasive plants upon the 

landscape. In this study, the first spatial distribution maps for invasive tree species, Jacaranda 

mimosofolia (common name: Jacaranda), are created for the City of Johannesburg (CoJ). 

Jacaranda trees are well-known by citizens in the CoJ for their district purple flowers which 

blanket the landscape during springtime. A combination approach using citizen science, GIS, and 

Google Street View for data collection, analysis, and creation of the first spatial distribution map 

of exact location and prevalence of Jacaranda trees within certain suburbs of the CoJ, is 

produced. A total of 8,931 ground-truthing geopoints together with extensive Google Street View 

validation for Jacaranda tree presence, formed the basis of accurate spatial distribution maps. 

The first research question of this study focused on the spatial distribution of Jacaranda trees in 

the CoJ and was answered as a total of 54 suburbs were confirmed as having a large presence of 

Jacaranda trees in the CoJ. Citizen science data collected a total of 488 geotags for possible 

Jacaranda tree presence in the CoJ, over a 75-day online survey collection period. Although 

citizen science data provided a lower spatial resolution compared to successful fieldwork and 



iv 
 

Google Street View approaches, citizen science data provided very high accuracy for the 

identification and geolocation of Jacaranda tree presence in the CoJ which answers the second 

research question based on the effectiveness of the geospatial approach towards citizen science, 

ground-truthing and Google Street View as data collection methods.  

Since the accuracy of citizen science resulted in 66% of collected geotags within the categories of 

‘very high’, ‘high’ and ‘moderate’ accuracy ranges of between <7-24m from a confirmed 

Jacaranda tree, together with the accuracy of 8,931 in-field collected geolocation of Jacaranda 

trees and Google Street View’s accuracy and capability of collecting street view imagery – it is 

concluded that the combined approach of ground-truthing, citizen science and Google Street 

View contribute not only to effective data collection, but also towards the successful mapping of 

Jacaranda tree presence in the CoJ. 

 

 

 

 

 

 

 

 



v 
 

Acknowledgements 

First and foremost, I would like to thank my supervisor, Prof. Jennifer Fitchett. Prof, thank you 

for seeing the potential in me from my Honours year and for providing the best teaching support 

I have received throughout my entire academic career. You have always been patient, 

understanding and kind from the time I changed my entire focus four months into the year, right 

the way through to my final days of submission whilst I juggled moving to a new city. Your passion 

is admirable. 

I would like to thank the South African National Space Agency for awarding me a bursary that 

provided the funding for my Masters studies. I would also like to acknowledge the support and 

assistance from the fantastic team at Esri South Africa. To Lizette Rust for offering me an 

abundance of resources and support. To Cameron Muller for being the best Mentor and now one 

of my role models in the GIS field. To Deon Lengton for helping with Esri devices and providing 

support during fieldwork. To the Esri Interns who kept me smiling (and dancing). Specifically, I 

would also like to acknowledge Lindokuhle Nhlabathi, Shelton Ndlovu, Basil Rabophala, Tears 

Rankapole, and Linda Mabeso for accompanying me on data collection during my fieldwork 

campaigns, you each made the fieldwork component one of the best experiences of my academic 

career. To everyone from the Esri Durban office for sharing this academic journey with me.  

I would like to acknowledge all anonymous respondents involved in the survey aspect of my 

research project. I would also like to thank Dean Peters and Dale Carter from the City of Cape 

Town for the ongoing support and motivation towards the completion of my studies.  



vi 
 

Last, but certainly not least, I would like to thank my entire family for their immeasurable support. 

To my grandparents – Ma and Pa, thank you both for ensuring I get away from my desk with 

weekly brunch and coffee outings which were always filled with laughter and wise words of 

encouragement. To my little brother, Kieron, for always offering warm hugs, funny faces, and 

constant reminders to make time for fun.  

To my parents, Sangita and Noel, this research project is dedicated to the both of you. I would 

never have made it this far if it wasn’t for the fighting spirit you have instilled in me. Thank you 

for the endless amount of love, guidance, understanding, support, and patience you both 

continue to offer. 

 

 

 

 

 

  

 

 

 

 



vii 
 

Table of Contents 

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

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

Acknowledgements ....................................................................................................................................... v 

Table of Contents ........................................................................................................................................ vii 

List of Figures ................................................................................................................................................ x 

List of Tables ...............................................................................................................................................xiii 

List of Acronyms and Abbreviations ........................................................................................................... xiv 

Glossary ....................................................................................................................................................... xv 

CHAPTER 1: INTRODUCTION ......................................................................................................................... 1 

1.1 Background ......................................................................................................................................... 2 

1.2 Rationale ............................................................................................................................................. 4 

1.3 Aims and Objectives ............................................................................................................................ 6 

1.4 Research Questions ............................................................................................................................ 6 

1.5 Structure of the Dissertation .............................................................................................................. 7 

CHAPTER 2: LITERATURE REVIEW ................................................................................................................. 8 

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

2.1.1 Background and Previous Research ............................................................................................. 9 

2.2 Tools for Recording Vegetation ........................................................................................................ 11 

2.2.1 Citizen Science as a Tool ............................................................................................................ 11 

2.2.2 Google Earth and Google Street View as a Tool ........................................................................ 16 

2.2.3 Social Media as a Tool ................................................................................................................ 18 

2.3. Combined Citizen Science and Geospatial Approaches ................................................................... 19 

2.3.1 Spatial Distribution Mapping ..................................................................................................... 20 

2.4 Synthesis ........................................................................................................................................... 24 

CHAPTER 3: STUDY SITE .............................................................................................................................. 25 

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

3.2 History ............................................................................................................................................... 26 

3.3 General Geography ........................................................................................................................... 27 

3.4 Climate .............................................................................................................................................. 29 

3.5 Population ......................................................................................................................................... 30 



viii 
 

CHAPTER 4: METHODOLOGY ...................................................................................................................... 32 

4.1 Introduction ...................................................................................................................................... 33 

4.2 Data Collection .................................................................................................................................. 33 

4.2.1. Ground-truth Data .................................................................................................................... 35 

4.2.2 Virtual Campaign Data ............................................................................................................... 41 

4.2.3 Citizen Science Data ................................................................................................................... 42 

4.3 Data Analysis ..................................................................................................................................... 45 

4.3.1 Geoprocessing ............................................................................................................................ 45 

4.4 Synthesis ........................................................................................................................................... 52 

CHAPTER 5: RESULTS ................................................................................................................................... 53 

5.1 Introduction ...................................................................................................................................... 54 

5.2 Ground-truth Validation of Jacaranda Tree Presence ...................................................................... 54 

5.3 Google Street View Validation of Jacaranda Tree Presence ............................................................. 57 

5.4 Citizen Science Survey Responses ..................................................................................................... 61 

5.4.1 Social Media Posting Dates for Survey Responses..................................................................... 63 

5.5 Citizen Demographic Information ..................................................................................................... 65 

5.5.1 Citizen Familiarity with the CoJ .................................................................................................. 66 

5.5.2 Citizen Suburb of Residence ...................................................................................................... 67 

5.6 Citizen Science Certainty Ratings ...................................................................................................... 69 

5.6.1 Certainty Ratings Linked To Citizen Length of Residency in the CoJ .......................................... 71 

5.7 Citizen Science Geotagging ............................................................................................................... 73 

5.8 Citizen Science Geotag Accuracy ...................................................................................................... 76 

5.8.1 Inaccurate, Unconfirmed, or Invalid Geotags from Citizen Science .......................................... 78 

5.8.2 Accuracy of Geotagging Certainty from Citizen Science ............................................................ 87 

5.9 Spatial Distribution of Jacaranda Trees ............................................................................................ 89 

5.10 Synthesis ......................................................................................................................................... 96 

CHAPTER 6: DISCUSSION ............................................................................................................................. 97 

6.1. Introduction ..................................................................................................................................... 98 

6.2 Mapping Jacaranda Tree Distribution ............................................................................................... 99 

6.2.1 Invasive Species Monitoring and Management ....................................................................... 105 

6.2.2 Citizen Science Engagement .................................................................................................... 107 

6.2.3 Tourism Opportunities ............................................................................................................. 109 

6.3 Future Applications ......................................................................................................................... 111 



ix 
 

6.4 Limitations ....................................................................................................................................... 113 

6.5 Synthesis ......................................................................................................................................... 116 

CHAPTER 7:  CONCLUSIONS ...................................................................................................................... 117 

7.1 Introduction .................................................................................................................................... 118 

7.2 Accomplishment of Study Aim, Objectives, and Research Questions ............................................ 119 

7.3 Implications of Jacaranda Tree Distribution in the CoJ ................................................................... 122 

7.4 Future Research Avenues ............................................................................................................... 123 

References ................................................................................................................................................ 126 

Appendix A: Ethics Clearance Certificate .................................................................................................. 145 

Appendix B: Survey Questionnaire ........................................................................................................... 146 

Appendix C: Examples of Social Media Posts ............................................................................................ 163 

Appendix D: Example of Poster with QR Code to Survey ......................................................................... 169 

 

 

  

 

 

 

 



x 
 

List of Figures 

Figure 3.1: Study site map of the CoJ showing the main regions of Diepsloot, Midrand, Sandton, 

Alexandra, Inner City, Johannesburg South, Ennerdale, Soweto, Diepkloof, Rosebank, 

Roodepoort and adjacent municipalities.…………………………………………………………………… 29 

Figure 4.1: Methodology diagram…………………………………………………………………………… 34 

Figure 4.2: Own photograph during fieldwork showing a) winter-time green tree, b) winter-time 

green leaf c) spring-time purple tree, and d) spring-time purple flower of Jacaranda trees. 

…………………………………………………………………………………………………………………………….. …… 36 

Figure 4.3: Own photograph of CT8 Tablet and Garmin device provided by Esri South 

Africa…………………………………………………………………………………………………………………………… 37 

Figure 4.4: Own photograph during fieldwork in Soweto, showing three-button layout of 

QuickCapture application……………………………………………………………………………………………. 40 

Figure 4.5: Model builder workflow showing custom built SQL queries for data filtering and 

analysis..…………………………………………………………………………………………………….………………. 50 

Figure 5.1:  Map showing ground-based validation of a total of 8,931 geopoints where a single 

geopoint represents Jacaranda tree presence in the CoJ.  

………………………………………………………………………………………………………………………………      56 

Figure 5.2: Map showing Google Street View captures of Jacaranda tree presence in the CoJ, 

where a single white geopoint represents between 1 - 6 Jacaranda trees………………… 58 

Figure 5.3: Google Street View capture showing a) two green and b) four flowering Jacaranda 

trees in a single capture.…………………………………………………………………………………………… 59 



xi 
 

Figure 5.4: Count of online survey responses collected per month from citizens of the CoJ, over 

the 75-day survey collection period…………………………………………………………………………. 62 

Figure 5.5: Response count per day as influenced by posting days on five social media platforms 

of Instagram, Facebook, WhatsApp, LinkedIn, and Twitter………………………………………. 64 

Figure 5.6: Response count of each citizen’s timeframe for which they have lived in the 

CoJ…………………………………………………………………………………………………………………………… 66 

Figure 5.7: Citizen science online survey responses per Suburb in the CoJ, ordered 

alphabetically…………………………………………………………………………………………………………... 68 

Figure 5.8: Certainty rating of total citizen science responses per suburb, ordered from ‘Very 

certain’, ‘Certain’, ‘Neither certain nor uncertain’, ‘Uncertain’, and ‘Very uncertain’…….. 70 

Figure 5.9: Certainty ratings of Jacaranda tree presence in citizen’s suburbs, as influenced by 

citizen’s residence in the CoJ………………………………………………………………………………………... 72 

Figure 5.10: Map showing all captured citizen science geotags for Jacaranda tree presence in the 

CoJ………………………………………………………………………………………………………………………………… 74 

Figure 5.11: Map showing total citizen science geotags categorized by ‘my street’, ‘other streets’ 

and ‘exact location’ of Jacaranda tree presence in CoJ…………………………………………………. 75 

Figure 5.12: Screen capture from Google Street View of misidentification of Jacaranda tree to the 

common London Plane tree in the CoJ……………………………………………………………………….. 84 

Figure 5.13: Screen capture from Google Street View of misidentification of Jacaranda tree to 

similar pigmented flowering trees of differing tree species………………………………………... 85 



xii 
 

Figure 5.14: Screen capture from Google Street View of streets with available Google Street View 

imagery shown in blue. Streets without a blue highlight, do not have Google Street View imagery 

available……………………………………………………………………………………………………………………... 86 

Figure 5.15: Screen capture from Google Street View of blurred Google Street View imagery due 

to strong sunlight limiting visibility of tree species in this capture………………………………. 87 

Figure 5.16: Accuracy of citizen science geotagging compared to certainty ratings for Jacaranda 

tree presence on citizen’s streets..……………………………………………………………………………. 88 

Figure 5.17: Accuracy of citizen science geotagging compared to certainty ratings for Jacaranda 

tree presence on other streets..………………………………………………………………………………… 89 

Figure 5.18: Map showing combined geopoints of ground-based validation, Google Street View 

validation and citizen science geotags for Jacaranda tree presence in the CoJ………....  90 

Figure 5.19: Zoomed-in map showing select suburbs with large numbers of Jacaranda trees where 

ground-based geopoints, Google Street View geotagging, and citizen science geotagging data is 

abundant.………………………………………………………………………………………………………………….. 92  

Figure 5.20: Density map showing regions with large numbers of confirmed Jacaranda trees, as 

supported by ground-based and Google Street View validation...………………………………  93 

Figure 6.1: Google Street View of a blurred capture due to camera malfunction or privacy 

concerns…………………………………………………………………………………………………………….……… 102 

Figure 6.2: Google Street View capture of Kruger Street intersection in the CoJ where Jacaranda 

trees are captured in a) flowering and b) bare morphology………………………………………. 103 

 

 

 



xiii 
 

 List of Tables 

Table 5.1: Number of validated Jacaranda tree geopoints collected from each both field 

campaigns carried out in winter and spring seasons…………………………………………………….. 55  

Table 5.2: List of hashtags, tags and mentions which were used in Jacaranda posts on social 

media..…………………………………………………..……………………………………………………………………. 65 

Table 5.3: Certainty ratings of Jacaranda tree presence from citizen’s street and other 

streets…………………………………………………………………………………………………………………………. 73 

Table 5.4:   Six categories for level of accuracy of citizen science geotags from confirmed 

Jacaranda tree presence.…………………………………………………..……………………………………….. 76 

Table 5.5: Citizen science geotags categorized by six levels of accuracy. Accuracy levels were 

analyzed for the total geotag count, geotags on ‘my street’, geotags on ‘other streets’ and ‘exact 

location’.…………………………………………………..………………………………………………………………... 77 

Table 5.6: Description of 29 citizen science collected geotags which lie within the 'inaccurate’, 

‘unconfirmed' or 'invalid' accuracy range for >200m of confirmed Jacaranda tree 

presence...…………………………………………………..……………………………………………………………... 80 

Table 5.7:  Suburbs with high concentrations of Jacaranda tree presence, as confirmed from field 

campaigns (highlighted in pink) and Google Street View (highlighted in green), citizen science 

(indicated with an *). Suburbs where both field campaigns and Google Street View were used are 

highlighted in purple..…………………………………………………..……………………………………………. 94 

 

 



xiv 
 

List of Acronyms and Abbreviations 

CoJ: City of Johannesburg 

EVI: Enhanced Differential Vegetation Index 

GIS: Geographic information system 

GLONASS: Global Navigation Satellite System 

GPS: Global Positioning System 

GUIDs: Globally Unique Identifiers  

KML: Keyhole Markup Language  

LAI: Leaf Area Index  

LHWP: Lesotho Highlands Water Project 

NDVI: Normalized Differential Vegetation Index 

PPGIS: Public Participation GIS  

PSHB: Polyphagous Shot Hole Borer 

QR: Quick Response  

SABAP: The South African Bird Atlas Project  

SQL: Structured Query Language  

TOD: Transit-oriented Development 

VGI: Volunteered Geographic Information  

 

 

 



xv 
 

Glossary 

Capture: This is a single frame which is captured from a desired perspective using Google Street 

View. This single frame, or capture, has embedded geographic co-ordinates from which the street 

view was captured. 

Citizen Science: “The collection and analysis of data relating to the natural world by members of 

the general public, typically as part of a collaborative project with professional scientists” (Oxford 

2016). In this study, the general public were not actively engaged in the analysis of data, but 

rather with data collection only. 

Crowdsource:  Is to “obtain (information or input into a particular task or project) by enlisting the 

services of a large number of people, either paid or unpaid, typically via the Internet” (Oxford 

2016). In this study, the term ‘crowdsourced’ must be read with an understanding of the citizens 

of the City of Johannesburg (CoJ) as the unpaid volunteers who obtained Jacaranda tree 

information for the purpose of this study.   

Esri: Esri is a private company, started in 1969 where the first commercial geographic information 

system was developed. Today, Esri is a world leader in GIS software. 

Geographic information system (GIS):  A geographic information system produces maps and 

manages and analyzes a wide variety of types of data. In this study, a GIS assisted with the 

successful mapping of Jacaranda tree presence and distribution in the CoJ, whilst simultaneously 

providing the mapping platform for citizen science data collection. 

 



xvi 
 

Geopoint: These points represent the exact geographic co-ordinates of a Jacaranda tree collected 

through two field-campaigns in the months of August and October, using Esri’s Quickcapture 

application. 

Geotag:  This tag represents a single citizen science capture of a Jacaranda tree using Esri’s 

Survey123 Connect, which is a sophisticated geocoded survey platform. 

Ground-truth(ing): This process involved ground-based validation of Jacaranda tree presence 

whereby physical visits to Jacaranda tree locations were done.



1 
 

CHAPTER 1: INTRODUCTION 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Own photograph taken in the suburb of Northcliff, on 12 October 2022. 



2 
 

1.1 Background 

The City of Johannesburg (CoJ) agglomerates more than one third of the Gauteng City-region 

population (Schäffler and Swilling, 2013). The CoJ is one of the fastest growing cities in South 

Africa with the highest numbers of migrants (Rasmeni and Mudia, 2019). South Africa’s history 

of apartheid posed urban pressures which are still seen today (Risimati and Gumbo, 2018). 

Segregation of ethnic races lead to the CoJ having division of northern suburbs of the CoJ as 

affluent and southern suburbs as poor (Rasmeni and Mudia, 2019). The division of suburbs 

influenced a tree planting boom where many invasive tree species were planted predominantly 

in these affluent areas in an effort to reduce dust, air, heat, and noise pollution in the CoJ 

(Schäffler and Swilling, 2013). One of these many tree species was the Jacaranda mimosifolia 

(common name: Jacaranda).  

 Jacaranda is a deciduous tree belonging to the Bignoniaceae family (Xie et al. 2021). Jacaranda 

trees were brought to South Africa from Brazil in the 1800s not only to address pollution from 

increased mining in the CoJ, but also to beautify the landscape (Gachet and Schuhly, 2009). 

Jacaranda trees are identifiable by their bright purple flowers which blossom in spring and form 

a large part of the urban identity and urban forest of the CoJ (Schäffler and Swilling, 2013).  

Apart from the excitement which springtime Jacaranda flowering offers to citizens, these trees 

are also notorious for their conservation status. Jacaranda trees hold category 3 alien-invasive 

status and hold a unique position as an “invasive flagship species” (Fitchett and Fani, 2018, p. 

470). Owing to this status, Jacaranda trees are prohibited from being replanted, transplanted, or 

sold – as a result, the aging population of Jacaranda trees will not always be part of the urban 

fabric of the CoJ (Schäffler and Swilling, 2013). Spatial information on the distribution of 



3 
 

Jacaranda trees around the CoJ is scarce, with only one remote sensing-based study to have 

identified the location of Jacaranda trees in the CoJ, with limited accuracy (Newete et al. 2022). 

Owing to this, it is imperative to know the spatial distribution of invasive Jacaranda trees, as 

stated in the primary and secondary aims of this study. The aim to produce the first Jacaranda 

distribution map in the CoJ using a three-way combined data collection approach ensures 

accurate and efficient data collection points and resulting maps, as seen later in this dissertation.   

Citizen science, in this context, has been defined as “the collection and analysis of data relating 

to the natural world by members of the general public, typically as part of a collaborative project 

with professional scientists” (Oxford 2016). In this study, the general public were not actively 

engaged in the analysis of data, but rather with data collection only. Citizen science has been 

used in vegetation distribution mapping as data is collected at a fine spatial resolution over a 

large geographic area, at a very low cost (Chandler et al. 2017). There is an increased use of citizen 

science for invasive tree mapping (although often for their removal), providing data outputs 

which are both reliable and accurate in complex urban environments (eg. Hawthorne et al. 2015; 

Seiferling et al. 2017). This provides a framework for mapping Jacaranda trees which both 

contributes to the growing literature on mapping invasive species and assists those in monitoring 

the current distribution of this more unique flagship invasive. Moreover, as stated in the 

secondary aims, a citizen science approach for invasive species mapping in the CoJ is useful for 

both data collection and for quantifying the accuracy of citizen science data (eg. Hawthorne et 

al. 2015). 

Citizen science data collection is limited by issues of accuracy in the identification of specific tree 

species by non-specialists; these are addressed by ground-truthing through physical fieldwork 



4 
 

and the use of digital imagery such as Google Street View (Visser et al. 2014). Cross referencing 

of citizen science logged geopoints with trees identified on Google Street View, together with 

ground-based fieldwork – permits both an assessment of accuracy of citizen science as an 

approach, and a more comprehensive and accurate Jacaranda distribution map as explored in 

the research questions of this study - as linked to the spatial distribution of Jacaranda trees in the 

CoJ as well as the effectiveness of the geospatial approach towards citizen science, ground-

truthing and Google Street View as data collection methods. 

 

1.2 Rationale 

One of the many purposes of invasive mapping is to represent their distributions across both 

local and global landscapes and environments (Pedrotti 2012; Seiferling et al. 2017). Knowledge 

about the main features of a landscape, such as invasive presence, is imperative for efficient and 

effective maintenance and monitoring (Qiu et al. 2019; Choi et al. 2022). Repeated mapping of 

vegetation or individual species over time permits a collection of data which may offer 

information such as rate of growth, phenological changes, direction of growth and population 

presence – to name a few (Pedrotti 2012; Dujardin et al. 2022). Furthermore, it provides a visual 

interpretation of any trends which occur long-term and assists with possible future trend analysis 

(Masocha and Skidmore, 2011; Buldrini et al. 2015).  

There is an urgent need for accurate mapping of the spatial distribution of invasive species as 

their presence leads to a loss of biodiversity and ecosystem services (Visser et al. 2014). 

Jacaranda trees as an invasive species share similar needs for their spatial distribution to be 

mapped for the CoJ (Newete et al. 2022). Additionally, given the importance of Jacaranda trees 



5 
 

to the identity of the CoJ, and the use of Jacaranda lined streets in photography and videography 

during the spring season, a map of their distribution has social and cultural value. These maps 

provide vital location information for citizens to experience well-known Jacaranda blooms, whilst 

simultaneously allowing for improved monitoring of these invasive trees for both ageing analysis 

and the presence of Polyphagous Shot Hole Borer (PSHB) – an invasive insect which burrows 

through tree bark, weakening the tree’s ability to transport moisture – presenting great risk to 

many urban trees (Paap et al. 2018; Newete et al. 2022).  

Remote sensing has been widely used for invasive species mapping; however, challenges arise 

when mapping heterogeneous landscapes which are vegetation dense (Fonji et al. 2014). These 

challenges include cost – as ground-truthing is expensive and time-consuming, spectral mixing – 

as similar spectral reflectance from other land cover types skew the dataset, and misclassification 

– as coarse spatial resolution results in several land cover types as classified into a single type 

(Huete et al. 2002). Citizen science presents the opportunity for large-scale ground-validation of 

invasive Jacaranda trees, at a high spatial resolution and low cost (Tang and Liu, 2016). Ground-

truthing by field-visits are reduced in time as geolocation of Jacaranda trees are collected through 

citizen science, providing random sampling sites for ground-validation (Hawthorne et al. 2015). 

Moreover, Google Street View is a free platform compiling street-level views of roads and 

pavements across the world, which is useful for the identification of streets where these urban 

trees are located, without being physically present in-field (Visser et al. 2014).  

 



6 
 

1.3 Aims and Objectives  

The primary aim of this study is to produce the first map of the distribution of Jacaranda trees in 

the CoJ using a geospatial approach through a combination of citizen science, ground-truthing 

and Google Street View. The secondary aim is to explore the accuracy of citizen science in 

mapping distinct tree species.  This will be achieved through the following objectives: 

1) To use a citizen science approach to crowd-source data on the location of Jacaranda trees 

in the CoJ and to compare this data to ground-truthing and Google Street View data on 

confirmed Jacaranda tree presence in the CoJ, to then map all three data collection methods 

through an integrated geocoded survey and GIS platform. 

2) To use both a GIS and Google Street View to validate ground-truthing data of both reported 

and un-reported Jacaranda tree locations in the CoJ, as reported from citizen science. 

3) To perform an accuracy assessment of the citizen science derived data and map the level 

of accuracy on the final product. 

 

1.4 Research Questions 

1) What is the spatial distribution of Jacaranda trees in the CoJ? 

2) How effective is the geospatial approach in the combined citizen science, Google Street 

View, and ground-truthing methods of data collection? 

 



7 
 

1.5 Structure of the Dissertation 

This dissertation comprises seven chapters. Chapter two explores studies that have engaged a 

GIS, citizen science and Google Street View as tools for successful vegetation mapping, together 

with the pertinent literature on the CoJ’s urban forest. Chapter three provides contextualization 

of the study site with reference to the climate, geography, urban forest, and population 

dynamics. Chapter four provides a detailed breakdown of the methods used in this research 

project, with reference to studies which have either engaged in similar or alternate 

methodological approaches. Chapter five provides a critical analysis of all core data components 

and presents these results along with their limitations. Chapter six provides a detailed discussion 

of each core component of this study where individual data sources are analyzed for trends and 

patterns on significant discoveries and their implications for this study and for future 

applications. The final chapter concludes with a synthesis of the study and achievement of the 

aim and objectives, followed by future research avenues and implications fostered from this 

research study. Due to the hierarchy of sections, sub-sections, and sub-sub-sections, to allow for 

ease of identification of figures and tables, figures and tables are numbered according to the 

subsection in which they fall. 

 

 

 

 

 

 

 

 



8 
 

CHAPTER 2: LITERATURE REVIEW 

 

 

Own photograph in the suburb of Waverley, taken on 17 October 2022. 



9 
 

2.1 Introduction 

Vegetation mapping is of importance as it provides spatial information on the location, type, 

direction, and rate of spread of plant species (Bois et al. 2011). In the CoJ, where the urban forest 

is one of the world’s most extensive, a spatial understanding of various urban forest species is 

imperative for effective monitoring and management of invasive species (Fitchett and Raik, 2021; 

Newete et al. 2022). This is especially significant for effective conservation efforts to manage and 

plan for future trends (Jordan et al. 2012; Thompson 2016). The literature review provides an 

overview of the published research reflecting tools that are used in recording vegetation, with a 

focus on the effectiveness of the geospatial approach towards the combined use of citizen 

science, Google Street View, and ground-truthing as data collection sources, as stated in the 

research question. In addition to reviewing existing studies that map vegetation, this literature 

review also critically reflects on methodologies employed to assess the accuracy of each of the 

data collection sources used, along with capabilities for spatial distribution mapping of invasive 

species in both the CoJ and other cities with heterogeneous urban forests. 

 

2.1.1 Background and Previous Research 

Today, alien invasive Jacaranda trees are mostly found in the northern suburbs of the CoJ, as 

these were typically the wealthy suburbs during post-colonial times (Newete et al. 2022). 

Although Jacaranda trees are abundant in the CoJ’s landscape, studies linked to spatial 

distribution are few, with many existing studies focusing on remote sensing mapping approaches 

(Jombo and Adam 2018; Abutaleb et al. 2021; Newete et al. 2022). Newete et al. (2022) present 

spatial distribution mapping of Jacaranda tree presence in the CoJ using remote sensing efforts 



10 
 

revealed four main suburbs of Saxonwold, Houghton, Parkview, and Parkwood as high 

prevalence Jacaranda tree suburbs. Though remote sensing is a commonly used tool, this study 

highlighted two research gaps linked to spectral mixing and misclassification of tree canopies 

when mapping a heterogeneous landscape such as the CoJ. 

 

The first research gap presented is towards the exclusion of smaller suburbs within the CoJ with 

potential Jacaranda tree presence as remote sensing efforts were unable to detect the small-

scale tree canopy cover of these suburbs (Newete et al. 2022). Since Jacaranda trees are a 

category three invasive species, it is important to identify all suburbs for their presence and to 

utilize both small and large-scale mapping techniques. A secondary data collection technique 

through fieldwork was aimed at collecting Jacaranda tree data using a hand-held GPS device, 

however, areas visited were few in numbers, chosen at random, and included major parks and 

suburbs only (Newete et al. 2022). To achieve a spatial distribution map which is accurate, all 

areas of the CoJ must be visited or considered irrespective of suburb size and independent of 

known areas for Jacaranda tree presence.  

 

By this, citizen science as a tool is effective to crowd-source data on both Jacaranda tree location 

and known presence as citizen science data is sourced at a fine spatial resolution, over the 

geographic area in focus (Chandler et al. 2017).  Citizen science data can form the foundation for 

other spatial distribution mapping efforts to confirm or validate Jacaranda tree presence. This 

coincides with the use of ground-truthing and Google Street View as two capable tools for the 

validation, confirmation, and mapping of the spatial distribution of Jacaranda trees in the CoJ 



11 
 

(Ives et al. 2017; Seiferling et al. 2017). Ground-truthing and Google Street View data collection 

methods can be compared to citizen science data collection for a measure of accuracy to be 

produced, together with providing an assessment of the effectiveness of using a geospatial 

approach in this regard (Bois et al. 2011). These three data sources combined will address any 

gaps which may present themselves in the data collection period. 

Therefore, it is important to highlight areas where data has not been collected, as well as to 

consider additional data collection methodologies to curb these research gaps for more accurate 

spatial distribution mapping. 

 

2.2 Tools for Recording Vegetation 

Vegetation has been observed and recorded by both scientists and members of the general public 

for hundreds of years (Marsham 1789; Sparks and Carey 1995). This has a range of applications 

from tracking phenological shifts to monitoring invasive species and identifying biomes. The 

current toolbox of methods used in recording vegetation includes the use of citizen science, a 

combination of Google Earth with Google Street View, as well as social media. Each tool provides 

fundamental opportunity for high-resolution data collection, at varied geographical scales, which 

is explored in this section. 

 

2.2.1 Citizen Science as a Tool 

Citizens as volunteers provide support for sampling activities which are essential for both 

biodiversity monitoring and public awareness of ecological changes (Miller-Rushing et al. 2012; 

van der Wal et al. 2015). Citizen science, in this context, has been defined as “the collection and 



12 
 

analysis of data relating to the natural world by members of the general public, typically as part 

of a collaborative project with professional scientists” (Oxford 2016). These citizens were not 

actively engaged in the analysis of data, but rather with data collection only. Volunteers for 

citizen science projects are made up of the public with a focus on the connection of people to 

science (Wright et al. 2015; Vahidi et al. 2021). When citizens participate, it empowers the public 

to make direct contributions to scientific research (Dickinson et al. 2012; Miller-Rushing et al. 

2012).  

Citizen science can encompass ‘crowdsourcing’ and ‘public participation’ (Brown and Kytta 2014; 

Fonji et al. 2014; Vahidi et al. 2021). However, there are distinct differences between these 

terms. Public participation is the participation by the public-at-large to scientific research, as well 

as political decision making, urban planning and other applications (Cooper et al. 2018). Here, 

scientific research is argued to have never been the exclusive domain of professional scientists 

as many ‘amateur’ scientists have been successful in this regard (Cooper et al. 2018). Scientific 

research is mostly involved in routine with documenting, monitoring, analysis and writing as 

mundane tasks which almost anyone can make useful contributions towards – if they are careful 

and stringent to protocol (Cooper et al. 2018). In citizen science, a citizen scientist is concerned 

only with science and not scientific research. In this study, citizens did not have to be literate to 

the scientific process, but merely capable of using simple data tools, such as user-friendly 

geocoded surveys. An example of citizen science in this regard is seen in studies from the data 

logging tool named CyberTracker. Cybertracker was tested in the Karoo National Park, South 

Africa, where two citizens could not read or write but were successful in the thorough data 



13 
 

collection and monitoring of endangered Desert Black Rhino through the CyberTracker field 

computer (Liebenberg et al. 1999).  

Crowdsourcing is commonly used to “obtain information or input (into a particular task or 

project) by enlisting the services of a large number of people, either paid or unpaid, typically via 

the Internet” (Oxford 2016). Therefore, the complexity and ambiguity between the terms of 

citizen science, crowdsourcing, and public participation as three separate concepts, must be 

acknowledged (Cooper et al. 2018). Though citizen science, crowdsourcing and public 

participation are similar by the involvement of the public for data collection, each term clearly 

differs by the level of which participants are expected to engage on (Harvey 2012). Therefore, 

citizen science is the most correct term for this study as the participation of the public-at-large 

for the purpose of science was evident. Moreover, the public in this study must be understood 

as citizens of the CoJ who each participated on a voluntary basis without the enlistment of their 

services or benefit of an income nor an incentive as crowdsourcing approaches would offer.  

In vegetation recording, a citizen science technique assists with the cross-validation of 

professional data collection as a larger database consisting of both professional ground-control 

data and citizen observation data is created (Roman et al. 2017; Roy-Dufresne et al. 2019). These 

multiple stakeholder collaborations present new opportunities not only for database 

management, but also for vegetation distribution modelling (Roy-Dufresne et al. 2019; Azzurro 

and Cerri, 2021). Information extracted from citizens who live near the environmental region of 

interest often complements or even replaces professional methods of ecological sampling (Ives 

et al. 2017; Azzurro and Cerri, 2021). These diverse uses of citizen science have each assisted in 

increasing public awareness of science by incorporating local knowledge of citizens into scientific 



14 
 

research (Wright et al. 2015; Roman et al. 2017). This promotes advocacy among citizens as more 

learning opportunities are presented and diverse views are considered (Cooper et al. 2008; Vahidi 

et al. 2021). However, most of these citizen science programs are carried out in North America 

or Europe, with few programs found in Africa, Asia, and Central and South America (Chandler et 

al. 2017). 

 

2.2.1.1 Citizen Science Projects 

Globally, Citizen science projects in biodiversity and conservation management have attracted 

increased attention due to awareness of citizen science programs (Newman et al. 2012; Nugent 

2018). Two global programs, among many others, are those of iNaturalist and the Global Coral 

Reef Monitoring Network (Chandler et al. 2017). iNaturalist is a mobile application which serves 

the global community for nature observation and identification by means of photography with 

embedded information on the date and location of the photograph for scientists and society to 

learn from (Nugent 2018). The Global Coral Reef Monitoring Network is an initiative which 

provides updated scientific information on the status and trends of coral reef ecosystems 

throughout the world (GCRMN 2020). Locally, a range of citizen science programs have been 

implemented by the Animal Demography Unit (ADU), based at the University of Cape Town. The 

South African Bird Atlas Project (SABAP) project is another well-known local citizen science 

project whereby bird species distribution and abundance are mapped across several countries in 

southern Africa through digital mapping (SABAP2 2023). Cape Citizen Science is a project based 

in the City of Cape Town, supported by the Forestry and Agricultural Biotechnology Institute, with 



15 
 

a focus on utilizing citizen science to collect information on plant diseases and insect pests, to 

protect the flora of the greater Cape floristic region (Cape Citizen Science 2023).  

These are but a few of the many other well-known projects which use citizen science for data 

collection. Though the above-mentioned projects are successful on both global and local scales, 

certain data collection projects may require location specific information. In the projects listed 

above, a mobile application is commonly used as the platform for users to collect data, however, 

these applications lack a geospatial component whereby a device’s precise location can be 

accurately and immediately provided therefore shadowing the location data from where the user 

is situated. This automated accuracy-controlled approach has proven to drastically increase data 

accuracy – as opposed to a user manually inputting their location – where a spatial component 

is necessary for reporting (Atzmanstorfer et al. 2014). Hence, a margin for location error, based 

on citizen inaccuracy, must be considered. 

Moreover, many of the above-mentioned projects depend on the aggregated answers from 

citizen science reports as a marker for validity and accuracy of true recordings (Swanson 2016). 

This too provides margin for error as additional validation tools are not utilized to leverage 

possibilities of inaccurate identification by citizens (Hara et al. 2013). Ground-truthing by field 

work campaigns provide the most accurate form of data validity since citizen science reports can 

be visited physically to confirm the presence or abundance of the research area in focus 

(Hawthorne et al. 2015). Further, the validation of citizen science points in-field contributes to 

accuracy measurements where citizen science collected data may be directly compared and 

assessed to control data in-field (Thompson 2016; Wakie et al. 2016). Similarly, the combined use 

of aerial imagery and digital street views – as accessed through open-source platforms such as 



16 
 

Google Earth and Google Street View – promotes validation opportunities to remotely confirm 

the presence or abundance of the research area in focus (Ives et al. 2017; Seiferling et al. 2017). 

Hence, any inaccuracies found in citizen science data can be validated or confirmed to ensure 

data accuracy within these large citizen science projects.  

 

2.2.2 Google Earth and Google Street View as a Tool 

Google Earth’s aerial imagery together with Google Street View imagery, are two views which 

complement each other when tasked with recording both vegetation and infrastructure 

components (Berland and Lange, 2017). Google Earth and Google Street View are publicly 

available and easily accessible geospatial technologies which are cost-effective and offer precise 

alternative methods for identifying trees and buildings at local and global scale (Li et al. 2015; 

Choi et al. 2022).  

Google Earth constructs pictures of the surface of the earth by downloading satellite imagery in 

a user-friendly interface to allow both the public and researchers to ‘zoom-in’ to locations of 

interest whilst still viewing the satellite image (Lisle 2006). Google Street View is an interface 

within Google Earth and Google Maps which provides a streetscape view using individual 

photographs which are ‘stitched’ together to form a continuous view (Li et al. 2015). This 

continuous view is achieved through vehicles with mounted cameras, which physically drive 

through the streets of cities, towns, and villages across the world to supply these photographs (Li 

et al. 2015). Google Street View is a type of platform which provides on-the-ground street 

imagery at a high spatial resolution, for visual validation of sites close to roads (Visser et al. 2014; 



17 
 

Qiu et al. 2019). Google Street View offers an interactive approach for users to manually ‘drag 

and drop’ a virtual pin from the software directly onto virtual streets which have been previously 

mapped using vehicle platforms with photography instrumentation (Anguelov et al. 2010; Li et 

al. 2015). Users can then maneuver in all directions across a virtual 2D, and 3D plane to identify 

features along streets such as trees and buildings (Berland and Lange, 2017; Ives et al. 2017). 

In scientific research, Google Earth can be used to capture geographic information such as the 

exact coordinates of individual land features to entire polygon areas which define perimeter 

(Fonji et al. 2014; Ferreira-Rodriguez et al. 2021). Google Street View enables researchers to 

capture geographic coordinates of individual street-level features, validated through 2D and 3D 

imagery viewing (Qiu et al. 2019; Choi et al. 2022). Google Street View is especially useful for the 

virtual identification of urban street trees as the presence and species of specific trees at multiple 

locations can be validated against citizen collected data (Ives et al. 2017; Seiferling et al. 2017). 

Google Street View in tree identification, assists with identifying numerous factors such as 

location of trees, number of trees, size of trees and species – to name a few (Visser et al. 2014; 

Berland and Lange, 2017). 

However, studies which use Google Street View as a remote validation tool of street-level 

features also utilize additional data collection tools to ensure data collection is accurate. Citizen 

science is a common tool used in conjunction with Google Street View as citizens act as ‘sensors’ 

to both collect large amounts of data from the geographic area in focus whilst also assuming the 

role of fieldworkers since citizens are often required to walk various streets to validate Google 

Street View collected data (Hara et al. 2013). Hence, citizens are not only engaging in digital data 



18 
 

collection through online surveys, but also serving as fieldworkers since certain streets are 

physically visited and features are validated or confirmed by citizens often through means of 

photographic or digital survey verification (Clark 2014; Wegner et al. 2016). Projects such as 

Opentreemap aim to create a publicly available tree inventory for each city in the world by 

utilizing a combined approach of open-source street view data together with citizen science to 

catalogue street trees (Wegner et al. 2016; Hamilton et al. 2018). The addition of citizen science 

to Google Street View has been proven to enhance data quality, specifically for cases where tree 

species identification or validation is imperative (Clark 2014; Crall et al. 2015; Roman et al. 2017). 

 

2.2.3 Social Media as a Tool 

The widespread use of social media as a valuable tool in vegetation recording has recently been 

discovered. Social media platforms which lead in photography and visual representation are the 

forefront of capturing and sharing visible vegetation with the public (Silva et al. 2018). 

Common social media sites such as Facebook, Instagram and Twitter can, when granted 

permission, capture user’s GPS location from their mobile devices, based on either the user’s 

physical location or the location at which the photograph was taken (Brown and Kytta, 2014; 

Sullivan et al. 2014). This function is particularly useful as distinct phenological stages, such as 

peak flowering, may be identified through photography ensuring correct species identification 

(Silva et al. 2018). 

There are many geotagged photographs on social media sites which circulate across the world 

through the flexibility of the internet (Wegner et al. 2016; Yang et al. 2019). The inter-site 



19 
 

variability encourages the public to participate and therefore supports the citizen science 

approach since these social platforms make engagement and accessibility seamless for data 

collection (Clark 2014; Fonji et al. 2014). This interactivity encourages users to support the 

creativity, diversity, and openness of various fields – including those focused on biodiversity and 

ecosystem monitoring as associated with plant and tree mapping (Hawthorne et al. 2015; Wakie 

et al. 2016; Li et al. 2015).  

 

2.3. Combined Citizen Science and Geospatial Approaches  

In urban environments, citizens as sensors or ground-collectors of vegetation data are a great 

help towards successful identification of various species across the landscape (Qiu et al. 2019; 

Dujardin et al. 2022). The fast-growing data source of citizen science provides reliable and 

accurate invasive species plant mapping data outputs, from both simple and complex urban 

environments (Hawthorne et al. 2015; Roman et al. 2017; Seiferling et al. 2017). Combined citizen 

science and geospatial approaches have the potential to accumulate large amounts of data, over 

a wide geographic region all within a short timeframe, which may otherwise be difficult to collect 

(Fonji et al. 2014; Ferreira-Rodriguez et al. 2021; Dujardin et al. 2022). The accurate identification 

provided by citizen science assists with the creation and compilation of maps to visualize the 

spatial distribution of invasive plants upon the landscape (Hawthorne et al. 2015; Thompson 

2016). This is highly advantageous in regions or countries which are underdeveloped and limited 

in resources (Thompson 2016; Ferreira-Rodriguez et al. 2021).  



20 
 

2.3.1 Spatial Distribution Mapping 

Combined citizen science and geospatial approaches in invasive species mapping grants 

opportunity not only for effective biodiversity and ecosystem management, but also towards 

community awareness and education (Theobald et al. 2015; Maynard-Bean et al. 2020). This 

awareness bridges the gap between citizens, researchers, and policymakers, as each person 

shares responsibility in the project as it provides an opportunity for stakeholders to engage with 

and educate the public on scientific research in the community, whilst simultaneously 

encouraging citizens to express their knowledge on the local environment (Schlossberg and 

Shuford, 2005; Dickinson et al. 2012; Brown and Kytta, 2014; Fonji et al. 2014). Scientists and 

researchers support the aid of combined citizen science and geospatial approaches towards 

invasive species mapping as data accuracy is resoundingly higher when collected by citizens who 

are familiar with their region as a higher citizen participation is observed since citizens show 

increased concern towards projects which may directly impact their community (Bois et al. 2011; 

Jordan et al. 2012; Fonji et al. 2014; Theobald et al. 2015). This data accuracy reflects on the 

secondary aims of this study which explores citizen’s accuracy in data collection of invasive 

Jacaranda tree species in the CoJ.  

Scientists and researchers further support the use of internet-based platforms in combined 

citizen science and geospatial approaches since digital surveys, websites, and mobile applications 

permit large-scale instant collection, storage and sharing of both geolocation and attribute data 

(Bois et al. 2011; Hawthorne et al. 2015; Tang and Liu, 2016; Wakie et al. 2016). Developments 

in geospatial technology allow digital surveys to access Global Positioning System (GPS) location 

information from the device used to take the survey, to produce an exact geographic coordinate 



21 
 

location in real-time (Buldrini et al. 2015; Wright et al. 2015; Berland and Lange, 2017; Joy et al. 

2019). In biodiversity and ecosystem monitoring and management, this GPS-approach with 

geospatial technology is highly beneficial for locating and mapping out high-risk areas and assists 

with the identification of any patterns or future trends (Salem 2003; O’Donoghue et al. 2010).  

This GPS information can be stored on a geospatial database where several analysis tools may be 

used simultaneously to output information on spatial distribution, rate of spread, affected areas 

or resources, and possible monitoring solutions – to name a few (Le Maitre et al. 2002; 

Hawthorne et al. 2015). This reflects on the research question of this study whereby geospatial 

approaches assist with the analysis and visualization of data collected from citizen science, 

Google Street View and ground-truthing. Studies which use these real-time combined citizen 

science and geospatial approaches for data collection stem from various fields, which include but 

are not limited to; disaster management mapping, trend prediction mapping, biodiversity 

monitoring and ecosystem monitoring (Salem 2003; O’Donoghue et al. 2010; Joy et al. 2019). The 

combined citizen science and geospatial approaches in these fields assist with real-time 

monitoring, mapping, and management networks for the identification of affected areas and 

potentially affected areas (Tran et al. 2009; O’Donoghue et al. 2010; Joy et al. 2019; Yang et al. 

2019). 

Invasive species mapping projects are varied and may include careful selection and training of 

volunteers to correctly identify specific invasive species (Jordan et al. 2012; Wakie et al. 2016).  

Selection and training of volunteers increase the accuracy of positively identified invasive plant 

species, whilst also producing much larger sample sizes (Jordan et al. 2012; Thompson 2016). 

Training may also involve volunteers’ engagement with GPS devices when capturing ground 



22 
 

control points (Thompson 2016; Wakie et al. 2016). Though training increases the accuracy of 

data collected from citizens, the uncertainty of training time needed often compromises planned 

timeframes for project start and completion (Fuccillo et al. 2015). This is a major disadvantage 

for projects which depend on seasonality for the successful identification of invasive plant species 

(Bois et al. 2011; Clark 2014).  

Well-known projects which use a combined citizen science and geospatial approach for 

distribution mapping and monitoring include eBird, SABAP, OpenStreetMap, OakMapper, 

Abandoned Developments and CoCoRaHS (Sullivan et al. 2014; Tang and Liu, 2016).  All six 

projects use volunteer collected data over online platforms to collect user inputs over a large 

geographic and temporal scale (Haklay and Weber, 2008; Tang and Liu, 2016). eBird is a web-

interface that records recreational and professional bird observations from the Western 

Hemisphere and beyond (Sullivan et al. 2014). This is similar to our local SABAP Bird mapping 

program in South Africa. OpenStreetMap is a web-based interface which allows citizens to report 

on the built environment and street conditions on a global scale (Sullivan et al. 2014). OakMapper 

uses an interactive map to record sudden oak tree death in California (Connors et al. 2012).  

Abandoned Developments records abandoned residential construction sites throughout the 

south-eastern United States using an interactive map (Tang and Liu, 2016). CocoRaHS is an 

interactive website used to quickly record precipitation observations across North America (Tang 

and Liu, 2016; Qiu et al. 2019).  

Two major kinds of interactive platforms were evident from these six combined citizen science 

and geospatial approach projects: direct and indirect (Tang and Liu, 2016). The indirect approach 

as used by eBird, SABAP and CocoRaHS involved indirect submission of data through a checklist, 



23 
 

whilst the direct approach as used in OpenStreetMap, OakMapper, and Abandoned 

Developments, involved a direct submission of data on an interactive map (Tang and Liu, 2016; 

Qiu et al. 2019). eBird, SABAP and OpenStreetMap resulted in the highest interaction rates 

because citizens were attracted to the project since they belonged to specific interest groups 

(Dickinson et al. 2012; Tang and Liu, 2016). eBird and SABAP attracted ‘Birders’ who are typically 

people with an interest in birdwatching, whilst OpenStreetMap attracted people who were 

interested in mapping the built environment (Sullivan et al. 2014; Tang and Liu, 2016). The 

Abandoned Developments and OakMapper projects did not have specific interest groups – which 

resulted in low levels of participation towards these projects from citizens (Haklay and Weber, 

2008; Tang and Liu, 2016).  

There are often concerns around the quality of citizen science data, however, studies show that 

data quality can be maintained when monitoring programs are well developed and carefully 

implemented (Atzmanstorfer et al. 2014; Crall et al. 2015). Volunteers may be specified or 

selected based on factors around the project scope to ensure data integrity (Joy et al. 2019; 

Dujardin et al. 2022). The specification of volunteers within citizen science projects is imperative 

to reduce issues around distribution of observers versus distribution of species as high 

concentrations of data points may represent higher population responses of observers rather 

than higher research sample data from species (Crall et al. 2015; Van de Wal et al. 2015).  

Additionally, the layout of surveys, websites, and applications contribute to a higher data quality 

from combined citizen science and geospatial approaches in invasive species mapping (Fonji et 

al. 2014). Many online surveys follow a common approach which either requires the citizen to 

identify an invasive plant species from a list of pre-configured invasive species options, or to 



24 
 

identify a single invasive plant species only (Fonji et al. 2014, Ferreira-Rodriguez et al. 2021). 

Questions within surveys, websites and applications often require details on the count of invasive 

species seen and a geo-tagged point of location from where the sighting was made (Tang and Liu, 

2016; Wakie et al. 2016). Geo-tagged points collected from these online surveys of specific 

invasive species can be represented as spatial distribution maps using a GIS as a tool (Hawthorne 

et al. 2015; Ferreira-Rodriguez et al. 2021). 

 

2.4 Synthesis 

Many conservation decisions made by cities and countries are based on imperfect or varied data 

(Bird et al. 2014; Barnard et al. 2017). Reliant policymaking is dependent on updated and 

accurate data on biodiversity, to predict species’ responses to local and global change (Bled et al. 

2013; Altwegg et al. 2014). Although phenological and remote sensing mapping efforts exist 

within the CoJ, spatial distribution mapping of Jacaranda presence has yet to be accomplished. 

Spatial presence of invasive trees is pivotal not only towards effective management, planning 

and monitoring, but is also key towards the identification of suburbs and regions which may be 

at risk for high invasive tree presence. This offers an additional opportunity to re-model the 

existing conservation and land management plans of the CoJ, to consider key areas which may 

pose future risk.  

 

 

 



25 
 

 

CHAPTER 3: STUDY SITE 

 

 

 

 

 

 

 

 

 

 

 
 

Own photograph taken in suburb of Saxonwold, on 12 October 2022. 



26 
 

3.1 Introduction 

The mapping of Jacaranda tree distribution using a combined citizen science and geospatial 

approach in this study, is focused on the CoJ, within the Gauteng province of South Africa. In late 

spring, the CoJ is a well-known location to experience the distinct purple flowering of invasive 

Jacaranda street trees (Fitchett and Raik, 2021). The CoJ is rapidly developing with high-rise 

buildings and industrial zones taking over vegetated areas, leaving the urban forest at risk due to 

poor monitoring and management practices (Newete et al. 2022). It is important to identify high 

risk areas where old Jacaranda trees may die from invasive pest species attack or may fall onto 

private property. This chapter describes the CoJ’s history, general geography, climate and 

population. 

 

3.2 History 

The CoJ is one of the largest and fasting growing cities in Africa, with one of the world’s largest 

urban forests assets which require attention (Schäffler and Swilling, 2013). The CoJ experienced 

an apartheid regime past whereby segregation and land separation of different races, each 

contributed to shaping to landscape of the CoJ today (Schäffler and Swilling, 2013). This 

segregation led to those of the white race residing in northern suburbs of the CoJ with the black 

races occupying the southern regions (Turton et al. 2006).  

The CoJ started major mining interventions from the 1800s with the discovery of gold in the Rand 

area (Turton et al. 2006). An influx in mining activities led to a tree-planting boom in the late 19th 

century to address the increase in dust, air, noise, and heat pollution in the CoJ (Schäffler and 



27 
 

Swilling, 2013). Quick-growing species were selected such as Eucalyptus, Black Wattle, London 

Planes, Oaks, and Jacaranda – to name but a few of the Colonial familiar species which were 

planted (Turton et al. 2006). Resultantly, major greening efforts were abundant in northern 

affluent suburbs of the CoJ, forming greenbelts and public parks (Schäffler and Swilling, 2013). 

Today, these trees form large parts of the CoJ’s urban forest with well over 10 million trees which 

occupy approximately 16.1% of the CoJ’s area as covered by both native and alien species 

(Schäffler and Swilling, 2013; Hardy and Nel, 2015). 

 

3.3 General Geography 

The CoJ is classified as part of the South African highveld, located in the Gauteng province of 

South Africa (Crétat et al. 2012). The CoJ is the largest City in South Africa with an area of 1,645 

km2 and is commonly referred to as ‘’The City of Gold’’ after gold was discovered in the Rand 

region in 1886 (Vearey 2010, p.37; Harrison and Zack, 2012). The CoJ is the economic and 

transport hub of South Africa and contributes to approximately 10% of Africa’s economic activity 

(Risimati and Gumbo, 2018).   

The main regions of the CoJ are those of Diepsloot, Midrand, Sandton, Alexandra, Inner City, 

Johannesburg South, Ennerdale, Soweto, Diepkloof, Rosebank, and Roodepoort (Risimati and 

Gumbo, 2018). The CoJ is administratively divided into seven regions, namely A, B, C, D, E, F, and 

G, which each consists of several suburbs (Risimati and Gumbo, 2018).  The exact boundaries of 

these regions are regularly changed and therefore difficult to map. The CoJ promotes transit-

oriented developments (TODs) in previously marginalized areas by focusing on development of 

economic and business sectors within Regions F, B and E. Regions G and D are defined as medium 



28 
 

to low-income residential spaces whilst regions A and C are defined as medium to high income 

residential spaces with some commercial activities (Risimati and Gumbo, 2018). The CoJ is a 

metropolitan City, with two neighboring metropolitan cities, namely City of Tshwane (Commonly 

known as Pretoria) and City of Ekurhuleni (Fitchett and Raik, 2021).  

The CoJ is situated along the watershed of the Witwatersrand and depends on the Lesotho 

Highlands Water Project (LHWP) transboundary transfer scheme as the main water supply source 

(Mguni et al. 2022).  The Witwatersrand is at the headwaters of two major river basins, namely 

the Lesotho and the Orange River basins, where any pollution occurring in these basins impact 

activities downstream (Turton et al. 2006). Hence, capping of the LHWP transboundary transfer 

scheme, by means of a maximum water capacity limit, until the second phase of the LWHP is 

completed – adds to water availability pressure in the CoJ (Mguni et al. 2022).  The CoJ has a 

highly variable topology with the central regions of the CoJ found at an altitude of 1,763 m above 

sea-level and northern regions found at lower altitudes, resulting in different air temperatures 

between these regions (Newete et al. 2022; Souverijns et al. 2022).  

   

 



29 
 

 

Figure 3.1: Study site map of the CoJ showing the main regions of Diepsloot, Midrand, Sandton, 

Alexandra, Inner City, Johannesburg South, Ennerdale, Soweto, Diepkloof, Rosebank, Roodepoort  

and adjacent municipalities. 

 

3.4 Climate 

The CoJ has a temperate climate with mean annual temperatures of 22°C (Dyson et al. 2015). 

Solar radiation and air temperatures are the largest influencers to heat stress in the CoJ, 

compared to humidity (Souverijns et al. 2022). The variable topology of the central suburbs found 

at higher altitudes than northern suburbs contribute to different air temperatures experienced 

across these regions (Hardy and Nel, 2015). The central regions of the CoJ emit the highest 

concentrations of heat due to high building density, increased anthropogenic activity and low 



30 
 

vegetation presence (Souverijns et al. 2022). Urban heat island stress is counteracted by the 

presence of vegetation which assists with detoxifying the air and providing shaded cover which 

lowers temperatures in the vicinity (Hardy and Nel, 2015).  

The CoJ experiences a mean annual rainfall of approximately 600 mm annually, with most of the 

rainfall during the summer months of October and March (Crétat et al. 2012).  Rainfall decreases 

from east to west of the country as influenced by the warm Agulhus current and presence of the 

Indian ocean on the east and cooler Benguela current from the Atlantic Ocean on the west (Crétat 

et al. 2012). Summer rainfall may lead to intense Highveld thunderstorms which may cause 

hazardous flash flooding due to poor infiltration rates of the urban environments artificial 

surfaces (Schäffler and Swilling, 2013). The low mean annual rainfall of the CoJ, combined with 

high evaporative potential and the variability of rainfall, results in South African rivers having 

among the lowest conversion of mean annual rainfall to mean annual runoff (Turton et al. 2006).  

 

3.5 Population 

The CoJ has a current population of approximately 6 million people with an approximate annual 

growth rate of 2% (United Nations 2022). Population growth and urbanization in the CoJ 

contributes to changes in the quality and quantity of urban green spaces (Abutaleb et al. 2021). 

The result of a rapidly growing population leads to decreased availability of land space as ‘green 

spaces’ in the City (Rasmeni and Madyira, 2019). An increase in population is translated to a 

mounting demand for service infrastructure to meet growing needs of the City, which is 

contributed by both domestic and international immigration (Abutaleb et al. 2021). The spatial 

distribution of highly populated suburbs in the CoJ is influenced by apartheid, where urban 



31 
 

growth was controlled, and the City was segregated by racial lines (Todes 2012). Spatial change 

in the CoJ has been rapid after 1994. In the CoJ, northern and western edges of the City house 

higher population density with the development of popular residential complexes (Todes 2012). 

Resultantly, suburbs which are highly populated, usually offer reduced vegetation cover. Rapidly 

developing suburbs within the CoJ such as Sandton, Melrose Arch and Midrand are just a few 

examples of areas where increased urbanization leads to reduced presence of vegetation 

(Mudede et al. 2020). 

 

 

 

 

 

 

 

 

 
 



32 
 

CHAPTER 4: METHODOLOGY 

 

 

 

 

  

 

 

 

 

 

 

 

 

 
Own photograph in the suburb of Northcliff, on 12 October 2022. 

 



33 
 

4.1 Introduction 

This dissertation is the first to map ground-level Jacaranda tree spatial distribution in the CoJ. 

The primary aim of this study is to map the spatial distribution of Jacaranda trees in the CoJ, 

through a combination of citizen science, ground-truthing using a GIS, and Google Street View 

techniques. The secondary aim is to explore the accuracy of citizen science in mapping distinct 

tree species.  This is followed by two research questions: First, “what is the spatial distribution of 

Jacaranda trees in the CoJ?”. Second, “how effective is the geospatial approach in the combined 

citizen science, Google Street View, and ground-truthing methods of data collection?” The 

combined approach requires the integration of processes to ensure harmony between the three 

data sources.  

This chapter outlines the data acquisition process from two field campaigns, Google Street View 

efforts and survey responses collected from citizen science, as supported by Esri South Africa. 

Thereafter, methodological processes for each dataset will be discussed to determine any 

relationships between data acquisition procedures and resulting patterns or trends. Additionally, 

the calculation process for citizen science data accuracy will be discussed.  

 

4.2 Data Collection 

As this research study aims to produce the first spatial distribution map of Jacaranda tree 

presence in the CoJ, initial ground-based data acquisition was imperative to the success of the 

study since this would form the foundation to identify any data gaps or inaccuracies from citizen 

science collected data (Section 2.2.1). Two field campaigns for ground-data collection were 

necessary for the collection of control points for validated Jacaranda tree presence in the CoJ. In 



34 
 

the absence of research studies which use geospatial technology as a tool for invasive tree 

mapping in the CoJ, a secondary data collection technique to address any missing data from 

fieldwork campaigns was important. Google Street View for virtual data collection of confirmed 

Jacaranda trees in the CoJ assisted here (Section 2.2.2). These two data collection techniques 

ensured digital standardization of data, so geopoints could be processed and integrated into 

digital maps (Crall et al. 2015). Citizen science data collection was carried out through the use of 

digital surveys (Vahidi et al. 2021). This digital component granted seamless transition of both 

attribute and spatial data, which were collected from individual users and stored on online 

geospatial databases (Figure 4.1).  

 

 

 

Figure 4.1: Methodology diagram 



35 
 

4.2.1. Ground-truth Data 

Two field campaigns were carried out from 5-11 July 2022 and 11-17 October 2022 to collect 

validated Jacaranda tree location. Based on extensive research prior to field campaigning, a 

preselection of suburbs with a large number of Jacaranda trees guided the first field campaign in 

July 2022 (Section 2.1.1; Newete et al. 2022). Jacaranda tree prevalence varies per suburb with 

some suburbs having a high prevalence, with Jacaranda trees lining either side of multiple streets, 

whilst other suburbs have a very low prevalence with Jacaranda trees located sporadically across 

the suburb. Some suburbs, especially the more recently developed suburbs, do not have any 

Jacaranda trees at all. Since the two field campaigns were conducted during the seasons of winter 

and spring, Jacaranda trees vary is physical appearance. In winter, Jacaranda trees are identified 

by lush green bipinnately compound leaves which made these trees stand-out against the dry 

winter landscape and were identified by their distinct appearance (Figure 4.2). Similarly, field 

campaigns during spring blossoming of Jacaranda trees facilitated easy identification of the trees 

as Jacaranda trees blossom with bright purple flowers. 

 



36 
 

a b  

c  d  

Figure 4.2: Own photograph during fieldwork showing a) winter-time green tree, b) winter-time 

green leaf c) spring-time purple tree, and d) spring-time purple flower of Jacaranda trees. 

 

 

Data from field campaigns were captured as geopoints using software and hardware provided by 

Esri South Africa. Observations for Jacaranda tree presence was made through physically driving 



37 
 

to and walking along individual streets per suburb and capturing exact coordinates of Jacaranda 

tree presence using both the Cedar CT8 Rugged Tablet (CT8 Tablet) and Garmin Glo (Garmin) GPS 

device, together with Esri ArcGIS QuickCapture software (Figure 4.3).  

 

 

Figure 4.3: Own photograph of CT8 Tablet and Garmin device provided by Esri South Africa. 

 

A manual log was updated of individual street names which were not completed during the 

allocated day of fieldwork. These incomplete streets would either be visited the next day or 

validated remotely using Google Street View. In-field navigation to different streets was assisted 

by navigation services provided by Google Maps (Fonji et al. 2014). Up to six individual street 

names were added as ‘stops’ within a single Google Maps journey for routes of ‘best travel’ to be 

mapped. The first and last ‘stop’ or street name within Google Maps assisted as anchor points to 

ensure fieldwork was conducted within the boundary of the suburb in focus, as navigation 



38 
 

without Google Maps became disorientating in larger suburbs or within streets with similar 

names to others. The extensive field campaigning for control points of confirmed Jacaranda trees 

formed part of this study’s main objective to have enough control points of known Jacaranda 

tree presence to measure citizen science data against, and therefore test for accuracy of citizen 

science data. 

 

4.2.1.1 Field Devices and Software 

The CT8 Tablet, which is commonly used with Esri ArcGIS products, is a durable field device 

powered by Android 8.1 with Google mobile services included (Juniper systems 2019; Esri 2022a). 

The CT8 Tablet has renowned GPS accuracy between 5-8m under a dense tree canopy, due to an 

external GNSS Antenna attachment (Juniper systems 2019). The handheld Garmin device boosts 

GPS accuracy of the CT8 Tablet with space-based global navigation satellite system (GLONASS) 

receivers (Garmin 2023). The Garmin device has a GPS accuracy of between 1-3m under a dense 

tree canopy (Garmin 2023). The Garmin device connects wirelessly through Bluetooth to the CT8 

Tablet, and these devices together form the hardware to collect accurate ground-truthing 

averaging between 1-3m under a dense tree canopy.  

 Esri’s QuickCapture application was used for the rapid collection of individual geopoints. 

QuickCapture is an application which can be easily configured by the user for a tailored fieldwork 

experience (Esri 2023a). For this study, a template for reporting was used and configured as an 

application where one-touch button activations were utilized. This configuration process 

involved the deconstruction of a reporting template, and reconstruction which was fine-tuned 

and tailored for optimal user functionality for this research study. The first button activates a 



39 
 

drop-down list of pre-filled regions within the CoJ, so a specific region could be selected during 

fieldwork. The second button would capture the exact geographic co-ordinates from where the 

button was pressed in-field. Thereafter, a secondary screen would open to display a high-

resolution aerial imagery map, sourced from Esri South Africa, within the application. The 

LANDSAT8 aerial image map would display every recorded point for Jacaranda tree presence in 

the form of geopoints. This application layout was chosen for its high efficiency and simplicity to 

easily capture a large number of geopoints in-field with very high accuracy. 

A third button provided a live-tracking route option. This tracking option was a valuable guide at 

the end of each day of fieldwork, to keep a record of streets and suburbs visited. All geopoints 

with live tracking routes were then configured to automatically save every 30 seconds. Automatic 

uploads at 5-minute intervals were configured for all geopoints to upload onto Esri’s ArcGIS 

Online for storage, management, and analysis. ArcGIS Online is a cloud-based mapping and 

analysis solution used for map creation, data analysis, and sharing of data whether in-field on 

behind a machine (Esri 2023b; Figure 4.4). 

 



40 
 

 

 

 

 

 

 

 

 

 

Figure 4.4: Own photograph during fieldwork in Soweto, showing three-button layout of 

QuickCapture application. 

 

All collected geopoints were accessed through ArcGIS Online at the end of each day of fieldwork. 

In cases where streets were covered more than once (Section 6.6), these duplicated geopoints 

were filtered and removed using Esri’s Map Viewer. Map Viewer is an interactive web mapping 

and data visualization application for creating, exploring, and saving web maps which is available 

through ArcGIS Online (Esri 2023c). Live route tracking data was removed from the geopoint 

dataset to appear as its own separate dataset. Geopoints were processed for geographic 

projection to comply with decimal degree standardization throughout the project (Fonji et al. 

2014).  



41 
 

4.2.2 Virtual Campaign Data 

Virtual campaigns were carried out intermittently from August 2022 to November 2022 using 

Google Street View was the chosen tool for virtual validation of Jacaranda trees along streets in 

the CoJ (Section 2.2.2).  

In this study, a single Google Street View capture contained at least one Jacaranda tree (Section 

5.3). Once a capture was taken, it was named chronologically on the software, and the spacebar 

button on a keyboard was pressed once, to maneuver in equal intervals down a virtual street. If 

a view did not contain Jacaranda trees, this capture was still taken and instead named “None” to 

ensure a log existed for streets visited without Jacaranda tree presence. This process was 

repeated until the entire street was covered, and ended when an entire suburb was complete. 

Google Street View captures were manually categorized into virtual folders which were named 

after the suburb in which they were captured. Entire virtual folders per suburb were saved 

automatically on Google Earth Pro software.  

The digital file format of completed suburbs was that of Keyhole Markup Language (KML) files 

within Google Earth Pro (Ives et al. 2017). KML files are a universal format used for storing 

geographic data, however these file types are not easily read through ArcGIS products (Esri 

2023d). Instead, KML files were uploaded to Esri’s ArcGIS Earth where they could be saved as 

compressed files known as KMZ files. ArcGIS Earth is an interactive 3D experience similar to 

Google Earth Pro, except it does not yet have functionality for street-level analysis (Esri 2023d). 

In this study, ArcGIS Earth was useful as a conversion and sharing tool since converted KML files 

could be uploaded to ArcGIS Online for storage. This ensured data integrity and standardization 



42 
 

across Esri products since field campaign and virtual campaign data were uploaded to an online 

geodatabase and could be accessed through ArcGIS Online (Ferreira-Rodriguez et al. 2021).  

This systematic approach to the rapid collection and efficient storing of Google Street View 

captures assists with accuracy assessment efforts since these Google Street View captures form 

part of the control points from which citizen science data will be compared to. Moreover, the 

segregation of Google Street View captures as stored per suburb name, assists with detailed 

accuracy comparisons in cases where citizens have captured Jacaranda tree locations in suburbs 

which were not covered by ground-truthing efforts. Hence, careful attention is paid towards 

mapping spatial distribution of possible Jacaranda tree presence in these suburbs since ground 

validation points may not exist. 

 

4.2.3 Citizen Science Data 

Citizen science data was obtained through digital surveys which were purpose-built, refined, and 

tailored for this research project. These digital surveys were produced using Esri’s Survey123 

Connect. Survey123 Connect is a sophisticated desktop application which permits the user to 

author surveys using Microsoft Excel, which is alternatively coded with questions, instructions, 

restrictions, geotagging, and prompts each enabled through the advanced technological 

capability of Survey123 Connect (Esri 2023e). In this study, an advanced survey format was 

required whereby each question required tailored styling and functional formatting to guide the 

user through a series of geospatial instructions. This formatting included ‘drop-down’ lists, ‘select 

one’ or ‘select many’ options, ‘required questions’ with prompt or instruction messages for the 



43 
 

user, accuracy warnings for geotag questions, among others. Therefore, template survey layouts 

which exist on ArcGIS Online could not be used.   

Survey123 Connect automatically creates the digital survey with styling and layout as prompted 

from the coded spreadsheet program which is linked to the ArcGIS Online portal for all responses 

from users to be stored and managed digitally (Esri 2023e). Furthermore, Survey123 Connect can 

be restricted with GPS warnings to ensure a user can only capture a geotag once their device is 

in a specific accuracy zone. To ensure geolocation accuracy in this study, surveys were configured 

with a 7m accuracy and would not allow the user to capture a geopoint if their accuracy was out 

of this range.  

Survey123 Connect automatically generates both a survey link which can be accessed using any 

web browser, together with a Quick Response (QR) code which smart devices may scan from 

either a paper or digital based source to automatically redirect the user to a web browser to view 

the survey (Esri 2023e). Both digital links and QR codes assisted with the dissemination of the 

survey to a wide and varied sample of respondents through both snowball-sampling and 

convenience-sampling approaches and is in-line with this study’s objective to make this survey 

accessible through social media platforms for citizens of the CoJ to participate in Jacaranda tree 

data collection (Section 2.2.3). Snowball-sampling was useful in cases where a single user passed 

the survey link to others, while convenience-sampling was useful since social media formed a 

major part of attracting a wider audience and increased attention for survey completion.  The 

five social media platforms of WhatsApp, Instagram, Facebook, LinkedIn, and Twitter would each 

result in different trends of responses from citizens of the CoJ since each of the above-mentioned 

social media platforms distribute website links uniquely. Instagram was the preferred platform 



44 
 

to share digital survey links as accompanying Jacaranda photography could be posted along with 

main headings or attractive text directed at the CoJ audience, in an effort to easily attract a CoJ 

citizen’s attention. Additionally, physical posters were created for residents of the CoJ to scan 

the QR code and access the survey. Since surveys were coded and created for this research 

project, and as geotagging was a key component of the survey, a decision was made to target 

smart devices for survey collection instead of paper-based approaches. The use of social media 

platforms and QR codes to disseminate the survey ensured most users who received the survey 

had access to smart devices that could enable their participation.  

 

4.2.3.1 Social Media Data  

Social media assisted with identifying the best days and times of day to share survey links which 

correlated with increased user activity of Jacaranda photography. These data trends were 

collected through manual logging in a Microsoft Excel spreadsheet. Select Jacaranda season 

inspired events such as the “Jacaranda In Your Pocket” photo competition, “Johannesburg Free 

Walking Trails” for Jacaranda season walks in the City and “I Heart Jhb” for information on 

Jacaranda markets – among others – were logged in Microsoft Excel, capturing key dates for 

upcoming public events, competition open and close dates, and any information on the location 

of events. These dates formed the framework for the social media drive ensuring that survey 

links were shared in-line with upcoming dates of interest.  

The trend analysis of dates which corresponded to key interest from users was an iterative 

process which required constant monitoring not only for key dates and events, but also for the 

on-time creation of posts. These posts assisted with sharing the survey link to users whilst 



45 
 

simultaneously assisting with increased attention to the ‘Jacaranda Mapper’ Instagram account 

which had unique and attractive Jacaranda content linked to the survey for this research project. 

Response count from users on days where survey link was shared was monitored on an online 

dashboard, accessible through ArcGIS Online. Survey123 Connect intertwines with ArcGIS Online 

not only with automatic saving of user’s submitted responses, but also provides a virtual 

dashboard where survey response statistics are updated in real-time (Esri 2023e). The strategic 

monitoring of social media trends resulted in a total of 203 responses and 488 geotags for 

Jacaranda tree location collected during the 75-day survey collection period. 

 

4.3 Data Analysis 

Data standardization was insured since all data from field campaigns, Google Street View, and 

citizen science surveys were stored and managed entirely on ArcGIS Online. Although data was 

manipulated from Google Earth Pro to ArcGIS Earth, the strict decision to stay within Esri’s 

product suites not only ensured data integrity, but also ease of access across the several Esri 

products and applications used in this study (Ferrara 2020). This was further driven by use of 

Esri’s well-known ArcGIS Pro software which was pivotal for data analysis and visualization within 

this study as it served as an integration and analysis platform of all collected geopoints from 

citizen science, Google Street View and ground-truthing efforts.  

 

4.3.1 Geoprocessing 

ArcGIS Pro is a powerful single desktop GIS application which includes world-leading geospatial 

tools for data visualization, advanced analysis, and map creation (Esri 2023f). ArcGIS Pro hosts a 



46 
 

diversity of built-in spatial analysis tools which each assisted in the analysis of data collected from 

citizen science, Google Street View and ground-truthing efforts for the successful creation of 

several high-resolution spatial distribution maps for Jacaranda presence in the CoJ. 

 

4.3.1.1 Structured Query Language 

Structured Query Language (SQL) is a tool widely used for spatial modelling and database 

management in a GIS (Lei 2021). SQL expressions in Esri’s ArcGIS Pro adhere to standard SQL 

syntax with common use of Boolean statements (Esri 2022b). SQL is particularly useful for 

relational databases, which was the format used to relate attribute citizen science data to spatial 

citizen science data collected from geotagging efforts (Aufaure-Portier 1995). Here, several 

‘select layer by attribute’ SQL expressions were used to isolate data in unique pairs which were 

specific to this study. The use of SQL was essential in this study since attribute questions and 

respective unique answers per user (as adopted from citizen science survey responses), could be 

selected in isolation or cumulatively with associated spatial information. This resulted in 

unlimited data combinations for deeper regression analysis (Aufaure-Portier 1995). 

 

4.3.1.2 Joins and Relates 

Data collected from citizen science surveys were exported from ArcGIS Online in Microsoft Excel 

format. Hereafter, the Excel spreadsheet was cleaned and standardized to meet formatting 

requirements for use in ArcGIS Pro. Three Excel spreadsheets were compiled from citizen science 

responses. The first spreadsheet with attribute information, the second with spatial information 

of collected geotags from the citizen’s street and other streets, and the third with spatial 



47 
 

information from geotags linked to exact location tags. The separation of these datasets was 

imperative for analysis of citizen science accuracy where non-spatial data such as citizen’s ‘length 

of stay’ or ‘suburb of residency’ in the CoJ could be isolated and directed compared to the same 

user’s geotagging accuracy to find any trends which support the study’s secondary aim. 

A series of spatial joins were made between the three spreadsheets based on both Global IDs 

and Globally Unique Identifiers (GUIDs) which each represent an individual user through a unique 

identity code (Esri 2023e). These identity codes were extracted and linked across all three 

spreadsheets ensuring spatial analysis ability for single or multiple users simultaneously (Godfrey 

and Stoddart, 2018). These advanced spatial joins provided an opportunity to cross-compare 

answers from various users who contributed to the citizen science dataset, for new trend 

possibilities to be identified.  

  

4.3.1.3 Spatial Analysis 

Once all spreadsheets were linked based on a Global IDs or GUIDs, geometric analysis to correct 

spatial co-ordinates from the three datasets were done (Suzuki 2005). Since ground-truth 

geopoints were collected in-field and configured through QuickCapture, and citizen science data 

were collected through Survey123 Connect; geometric analysis ensures spatial standardization 

between the datasets before spatial analysis can be carried out (Breytenbach 2016). In this study, 

the Lo29 reference system based on the WGS84 reference ellipsoid was the geo-reference 

framework of choice since Esri software re-projects Lo29 co-ordinates with accuracy, whilst 

enforcing strict pixel-to-pixel registration with the gridded reference provided from ground-

truthing data (Breytenbach 2016).   



48 
 

Several spatial analysis tools such as proximity, intersect, distance, X-Y to line, buffer, and near-

distance, among others, were used.  Esri offers extensive and advanced geoprocessing tools built 

on various statistical approaches such as Pearsons’s correlation coefficient, Cohens kappa 

coefficient, Multiple regression, and Thiessen Polygon, among others (Esri 2022b). 

Geoprocessing tools are selected and applied on a case-by-case basis. In this study, several 

geoprocessing toolboxes were utilized such as the ‘Spatial Statistics Toolbox’, ‘Geostatistical 

Analyst Toolbox’ and ‘Spatial Analyst Toolbox’, to name a few. Each toolbox contains an array of 

individual tools which each perform advanced statistical analysis once the dataset has been 

formatted correctly from the initial collection, cleaning, filtering, and storage of data (Esri 2022b). 

ArcGIS Pro automatically processes individual or multiple spatial analyses at any given time, as a 

supervised classification (Esri 2023f).  

Proximity analysis from citizen science geotags to nearest ground-truth geopoints from both 

ground-truthing and Google Street View datasets were completed. For this, a new dataset was 

created and comprised of combined ground-truth geopoints and Google Street View geotags, as 

control points for validated Jacaran