University of the Witwatersrand Faculty of Science School of Geography, Archaeology and Environmental Studies MULTITEMPORAL ANALYSIS OF LAND COVER AND EVALUATION OF LANDSCAPE INFLUENCES ON VEGETATION DYNAMICS USING REMOTE SENSING DATA AND MACHINE LEARNING IN A KARST ENVIRONMENT: A CASE STUDY OF THE CRADLE NATURE RESERVE By Charles Matyukira (2530526) A thesis submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Doctor of Philosophy Supervisor Supervisor: Professor Paida Mhangara (Wits University) September 2024 i DECLARATION I declare that this thesis is my own, unaided work. It is being submitted for the Degree of Doctor of Philosophy at the University of the Witwatersrand, Johannesburg. It has not been submitted before for any degree or examination at any other University. ______Signature of candidate) ________03___________day October 2024 at Johannesburg ii LIST OF OUTPUTS OF THE THESIS These outputs (stand-alone original peer-reviewed manuscripts or conference proceedings) bear the following status: (1) Submitted – has been submitted to a journal, and the editor is yet to send it for review; (2) In-Review – a manuscript has been submitted to a journal, and the editor has deemed it relevant and worthy of consideration to be published and is now being reviewed by experts in the field, (3) Accepted – a manuscript has gone through a rigorous journal-specific peer-review process and various iterations of revisions, but not yet published pending production by the publisher, and (4) Published - a manuscript is now published in the journal issue and has a DOI. Matyukira, C., & Mhangara, P. (2023). Advancement in the application of geospatial technology in archaeology and cultural heritage in South Africa: A scientometric review. Remote Sensing (Basel), 15(19), Article 4781. https://doi.org/10.3390/rs15194781 Matyukira, C., & Mhangara, P. (2024). Advances in vegetation mapping through remote sensing and machine learning techniques: A scientometric review. European Journal of Remote Sensing (In review). Matyukira, C., & Mhangara, P. (2023). Land cover and landscape structural changes using extreme gradient boosting random forest and fragmentation analysis. Remote Sensing (Basel), 15(23), Article 5520. https://doi.org/10.3390/rs15235520 Matyukira, C., Mhangara, P., Gidey, E., & Hussein, J. (2024). Monitoring vegetation growth dynamics using enhanced vegetation index and principal component analysis in the Krast Environment: A case of the Cradle Nature Reserve, Gauteng Province, South Africa. Annals of GIS (In review). Matyukira, C., & Mhangara, P. (2024). Utilising RGB drone imagery and vegetation indices for accurate above-ground biomass estimation: A case study of the Cradle Nature Reserve, Gauteng Province, South Africa. Geocarto International, 39(1). https://doi.org/10.1080/10106049.2024.2390512 Matyukira, C., Mhangara, P., & Gidey, E. (2024). Modelling topographic influences on vegetation vigour in the Cradle Nature Reserve, Gauteng Province, South Africa. Geocarto International, 39(1). https://doi.org/10.1080/10106049.2024.2395313 Matyukira, C., Mhangara, P., & Gidey, E. (2024). Impacts of land use change on annual evapotranspiration in the Cradle Nature Reserve (2000–2023) using MODIS data and XGBoost classification: A case of the Cradle Nature Reserve, Gauteng Province, South Africa. Annals of GIS (In review). I declare that I had a maximum contribution (i.e., >80%) in each of the abovementioned research outputs. _ I declare that I had a maximum contribution (i.e., >80%) in each of the abovementioned research outputs. ___(Signature of candidate) 03 day October 2024 at Johannesburg https://doi.org/10.3390/rs15194781 https://doi.org/10.3390/rs15235520 https://doi.org/10.1080/10106049.2024.2390512 https://doi.org/10.1080/10106049.2024.2395313 iii ABSTRACT Monitoring vegetation dynamics and ecosystem processes is crucial for understanding and managing natural environments, particularly in sensitive regions such as protected heritage sites in karst landscapes prone to land degradation. Geospatial science and technology, especially remote sensing, have revolutionised environmental monitoring by providing unparalleled insights into complex ecological processes and land cover dynamics. Despite their immense potential, these advanced technologies remain underutilised in sensitive ecological archaeological sites, particularly within South African contexts. The Cradle of Humankind, a protected World Heritage site, is confronting a myriad of environmental problems, such as land degradation, invasion by alien plants, bush thickening, loss of native vegetation cover, increased soil erosion, and habitat loss, threatening its ecological integrity, biodiversity and archaeological significance. Furthermore, there is a notable paucity of research focusing on land degradation and land cover changes in protected heritage sites in Africa and other parts of the world, even though geospatial technologies capable of accurately mapping and tracking these changes have proliferated. Addressing this research gap is critical for developing effective conservation strategies and ensuring the preservation of sensitive ecological and archaeological areas. This study aims to demonstrate the significant value of integrating geospatial technology and machine learning in environmental monitoring by identifying and addressing research gaps in the application of these technologies, particularly remote sensing, for vegetation mapping. Through a detailed case study of the Cradle Nature Reserve in South Africa, the study employs novel machine learning classification algorithms and advanced geospatial analytical methods to assess land cover change, landscape fragmentation, multitemporal analysis of vegetation trends, topographic influences on vegetation vigour, biomass estimation, and impacts of land cover change on evapotranspiration using multisource satellite, drone, and field geospatial datasets. The study utilised a systematic review and scientometric analysis using the Scopus database to investigate the application and trends of geospatial technologies in archaeology and cultural heritage in South Africa from 1990 to 2022, employing VOSviewer for visualising bibliometric data. Additionally, the study employed a systematic review and scientometric analysis using the Scopus database to examine the growth and trends in the application of remote sensing and machine learning techniques for vegetation mapping from 2000 to 2024, using VOSviewer for visualising relationships and trends in the collected data. The research also utilised Google Earth Engine (GEE) to download and preprocess satellite imagery, employing the XGBoost and Naïve Bayes classifiers for land cover classification and accuracy assessment, followed by applying landscape metrics for fragmentation analysis using QGIS. Further, Sentinel-2 multispectral imagery was processed using QGIS and ArcGIS software to compute the Enhanced Vegetation Index (EVI) and analyse vegetation dynamics through Principal Component Analysis (PCA) and multilinear regression, examining the relationships between EVI and climatic factors such as rainfall, temperature, and soil moisture. The study also utilised RGB drone imagery and various vegetation indices (BGVI, ExG, GBVI, GRBI, NGRDI, RBVI) combined with nDSM and chlorophyll concentration measurements to estimate above- ground biomass (AGB) in riparian and non-riparian zones of the Cradle Nature Reserve. High- resolution satellite imagery from Sentinel-2A and the Shuttle Radar Topography Mission iv Digital Elevation Model (SRTM DEM) was used to calculate the Enhanced Vegetation Index (EVI) and various topographic indices such as the Topographic Position Index (TPI), Topographic Ruggedness Index (TRI), and Topographic Wetness Index (TWI), with the data analysed using QGIS and ArcGIS software to model the relationship between topography and vegetation vigour in the Cradle Nature Reserve. Finally, the study employed MODIS data and XGBoost classification to analyse the impacts of land use change on annual evapotranspiration (ET) in the Cradle Nature Reserve from 2000 to 2023, incorporating statistical analysis and regression models to explore the relationship between ET and different land cover types. The research identified significant gaps in the application of advanced geospatial technologies, noting that only 12% of global research output is attributed to South African studies despite an increase in publications from 2018 to 2022. Key research gaps identified include the limited use of virtual and augmented reality applications, aerial photography, optical radar, and UAVs, which remain underexplored and present opportunities for future research. The analysis revealed a significant increase in research publications, particularly from 2018 to 2023, with the number of articles rising from 6 in 2018 to 48 in 2023, highlighting the critical role of advanced machine learning algorithms and remote sensing technologies in enhancing vegetation mapping accuracy and depth. China led in research output, followed by the United States and India. However, we identified significant research gaps, such as the need for more comprehensive studies on evapotranspiration and above-ground biomass estimation and the integration of climate change data with remote sensing technologies. The study showed that between 1990 and 2020, the Cradle Nature Reserve's landscape became much more fragmented. There was a 39% rise in bare ground or rock outcrop and a 26% and 12% decrease in native forest and natural grassland, respectively. This was mostly caused by human activities and the introduction of alien plant species. The analysis revealed that the EVI exhibited strong positive correlations with rainfall (r = 0.71), temperature (r = 0.62), and soil moisture (r = 0.76), highlighting the significant influence of these climatic factors on vegetation health. The PCA revealed that the first two principal components accounted for 90.08% of the data variability, underscoring the combined impact of these environmental factors on vegetation vigour. The riparian vegetation model had an R² value of 0.42 (adjusted R² = 0.34), which means that chlorophyll and BGVI were important factors. Conversely, the non-riparian model had a higher R-value of 0.66 (adjusted R² = 0.64), which meant that nDSM and RBVI were important predictors. This shows how important vegetation height, and stress indicators are in estimating biomass. The analysis revealed strong correlations between EVI and topographic indices, with EVI showing a significant negative correlation with TPI (R² = 0.95) and TRI (R² = 0.94) and a strong positive correlation with percentage slope gradient (R² = 0.85), highlighting how terrain features like sinkholes and depressions impact vegetation health. The research identified significant ecological changes in the Cradle Nature Reserve, with ET values showing substantial fluctuations across different land cover types. Notably, the mean ET increased from 47.115 mm in 2000 to 57.316 mm in 2020, reflecting overall improvement in vegetation health and water availability, while the Indigenous Forest area saw an increase in ET from 45.59 mm in 2000 to 59.89 mm in 2020, underscoring the impact of land use changes and climatic factors on the reserve's ecosystem. This study equips researchers and conservationists focused on the Cradle Nature Reserve with advanced tools to enhance the precision of monitoring ecosystem changes and detecting land degradation. These insights facilitate targeted conservation efforts and informed management v strategies tailored to the reserve's unique ecological needs, contributing significantly to the global understanding of sustainable land management and conservation practices. Keywords: geospatial technologies, remote sensing, machine learning, vegetation mapping, land cover change, Cradle Nature Reserve, evapotranspiration vi DEDICATION I dedicate this work to my family, mum and dad vii ACKNOWLEDGEMENT I would like to extend my deepest appreciation to Professor Paida Mhangara for the excellent assistance, mentoring, and unwavering support I received during my studies. I would like to extend my gratitude to the Lee Burger Foundation for generously financing my study, granting me entry to Cradle Nature Reserve, and offering logistical assistance. I would like to express my profound appreciation to my family and friends, whose constant encouragement and emotional support have been invaluable to me throughout my academic path. viii ABBREVIATIONS AGB: Above-Ground Biomass APCs: Article Processing Charges BGVI: Blue-Green Vegetation Index CAS: Chinese Academy of Sciences CHMs: Canopy Height Models CNN: Convolutional Neural Networks COH: Cradle of Humankind COHWHS: Cradle of Humankind World Heritage Site CSV: Comma-Separated Value DEA: Department of Environmental Affairs DEM: Digital Elevation Model DSM: Digital Surface Models DTM: Digital Terrain Models ESA: European Space Agency ET: Evapotranspiration EVI: Enhanced vegetation index ExG: Excess Green Index GBVI: Green-Blue Vegetation Index GCPs: Ground Control Points GEE: Google Earth Engine GIS: Geographic Information Systems GMAO: Global Modeling and Assimilation Office GNSS: Global Navigation Satellite Systems GRBI: Green-Red Vegetation Index IGARSS: International Geoscience and Remote Sensing Symposium ISRO: Indian Space Research Organisation LAI: Leaf Area Index LC: Land Cover LecoS: Landscape Ecology Statistics LiDAR: Light Detection and Ranging LOS: Length of the growing season LPI: Largest patch index LULCC: Land use and land cover changes M3T: Mavic 3 Thermal MDPI: Multidisciplinary Digital Publishing Institute MERRA-2: Modern-Era Retrospective Analysis for Research and Applications, Version 2 METRIC: Mapping Evapotranspiration with Internalized Calibration MODIS: Moderate Resolution Imaging Spectroradiometer MPS: Mean patch size MSI: Multispectral Instrument NASA: National Aeronautics and Space Administration nDSM: normalised Digital surface model NDVI: Normalized Difference Vegetation Index ix NGRDI: Normalised green-red difference index NIR: Near-infrared NP: Number of patches PCA: Principal Component Analysis PCs: Principal components PD: Patch density PLAND: Percentage of Landscape RBVI: Red-Blue Vegetation Index RF: Rainfall RGB: Red, Green, and Blue ROI: Region of interest RS: Remote Sensing SAGA: System for Automated Geoscientific Analyses SAR: Synthetic Aperture Radar SAVI: Soil Adjusted Vegetation Index SfM: Structure from motion SH: Shrub SHDI: Shannon's diversity index SNAP: Sentinel's Application Platform SOS: Start of the season SRTM: Shuttle Radar Topography Mission SVM: Support Vector Machines TEMP: Temperature TITLE-ABS-KEY: Title-Abstract-Keywords TPI: Topographic position index TRI: Topographic ruggedness index TWI: Topographic wetness index UAV: Unmanned aerial vehicle UNESCO: United Nations Educational, Scientific and Cultural Organisation USGS: United States Geological Survey VIF: Variance inflation factor WHS: World Heritage Site XGBClassification: Extreme Gradient Boost Classification XGBRFClassifcation: Extreme gradient boost Random Forest Classification x Table of Contents Chapter 1 INTRODUCTION .............................................................................................................. 1 1.1 Background ................................................................................................................................. 2 1.2 Problem Statement ...................................................................................................................... 4 1.3 Research questions ...................................................................................................................... 6 1.4 Aims and objectives of the study ............................................................................................... 6 1.5 Study area .................................................................................................................................... 7 1.6 Significance of the study ............................................................................................................. 8 1.7 Outline of the thesis .................................................................................................................. 10 Chapter 2 ADVANCEMENT IN THE APPLICATION OF GEOSPATIAL TECHNOLOGY IN ARCHAEOLOGY AND CULTURAL HERITAGE IN SOUTH AFRICA: A SCIENTOMETRIC REVIEW........................................................................................................... 11 2.1 Introduction ............................................................................................................................... 12 2.1.1 Geospatial Technology in Archaeology .................................................................................... 14 2.1.2 Understanding Space and Place................................................................................................ 15 2.1.3 Development of Geospatial Technologies in South Africa ..................................................... 15 2.2 Materials and Methods ............................................................................................................. 16 2.2.1 Overview ..................................................................................................................................... 16 2.2.2 Design/methodology/approach .................................................................................................. 20 2.3 Results and Discussion .............................................................................................................. 23 2.3.1 Trend in worldwide documents published per year ............................................................... 23 2.3.2 Trend in South Africa documents published per year ........................................................... 24 2.3.3 Trends in worldwide published articles per journal results and discussion ......................... 27 2.3.4 Density Visualisation and analysing bibliometric networks .................................................. 29 2.3.5 Network Visualisation and analysing bibliometric networks ................................................ 31 2.4. Findings ..................................................................................................................................... 36 2.5 Conclusions ................................................................................................................................ 37 Chapter 3 ADVANCES IN VEGETATION MAPPING THROUGH REMOTE SENSING AND MACHINE LEARNING TECHNIQUES: A SCIENTOMETRIC REVIEW............................... 38 3.1 Introduction ............................................................................................................................... 39 3.2 Approaches to Vegetation Mapping and Analysis Using Remote Sensing Technologies ... 41 xi 3.2.1 Advancements in Machine Learning Algorithms for Vegetation Classification .................. 41 3.2.2 Vegetation Growth Dynamics Using Remote Sensing Technologies ..................................... 44 3.2.3 Analysis of Land Cover and Land Use Change Utilising Remote Sensing Technologies .... 45 3.2.4 Utilising Remote Sensing Technologies to Monitor Land Degradation and Fragmentation ............................................................................................................................................................ 47 3.2.5 Assessing the Impact of Terrain on Plant Health Using Remote Sensing Technologies ..... 48 3.2.6 Remote sensing technologies for estimating biomass and evapotranspiration ..................... 50 3.3 Materials and Method .............................................................................................................. 51 3.3.1 Employing Scopus and Excel for Targeted Literature Review in Vegetation Mapping ..... 51 3.3.1 Bibliometric Visualisation and Analysing of Data in VOSviewer ......................................... 53 3.4 Results and Analysis ................................................................................................................. 54 3.4.1 Publication Trends in Remote Sensing and Vegetation Mapping ......................................... 54 3.4.2 Geographical breakdown of research on remote sensing and vegetation mapping ............. 56 3.4.3 Distribution of publications in remote sensing and vegetation mapping by source ............. 57 3.4.4 Types of Documents in Remote Sensing and Vegetation Mapping Research ...................... 58 3.4.5 Subject Area Distribution of Research in Remote Sensing and Vegetation Mapping ......... 59 3.4.6 Analysis of Research Affiliations in Remote Sensing and Vegetation Mapping .................. 60 3.4.7 Summary of the significance and need of remote sensing and vegetation mapping research using the Scopus database. .............................................................................................................. 60 3.4.8 Interconnected Themes in Remote Sensing and Vegetation Mapping Research ................. 61 3.5 Findings from the Scientometric Review ................................................................................ 63 3.6 Future Research Directions, Emerging Technologies, and Potential Challenges ............... 65 3.7 Conclusion ................................................................................................................................. 66 Chapter 4 LAND COVER AND LANDSCAPE STRUCTURAL CHANGES USING EXTREME GRADIENT BOOSTING RANDOM FOREST AND FRAGMENTATION ANALYSIS .......................................................................................................................................... 67 4.1 Introduction ............................................................................................................................... 69 4.1.1 Land Cover Change Overview.................................................................................................. 69 4.1.2 Machine Learning Land Cover Classification ........................................................................ 70 4.2 Materials and Methods ............................................................................................................. 73 xii 4.2.1 The Study Area .......................................................................................................................... 73 4.2.2 Method ........................................................................................................................................ 75 4.3 Results ........................................................................................................................................ 85 4.3.1 Comparison of Accuracy Assessment of Classification Algorithms using 2020 data set ..... 85 4.3.2 Ground Truthing Accuracy Assessment .................................................................................. 87 4.3.3 Accuracy Assessment and Land Cover Digital Classification of Study Area (1990-2020) .. 87 4.3.4 Land Use/Land Cover Spatial Temporal Change Detection .................................................. 91 4.3.5 The Landscape Metrics and Dynamics: Class Level .............................................................. 98 4.4 Discussion................................................................................................................................. 100 4.4.1 Comparison of Naïve Bayes and Gradient Boosting Random Forest Classifiers .............. 100 4.4.2 Land use and land cover changes from 1990 to 2020 ........................................................... 102 4.4.3 Landscape structural changes due to fragmentation ............................................................ 103 4.5 Conclusions .............................................................................................................................. 105 Chapter 5 MONITORING VEGETATION GROWTH DYNAMICS USING ENHANCED VEGETATION INDEX AND PRINCIPAL COMPONENT ANALYSIS IN THE KRAST ENVIRONMENT: A CASE OF THE CRADLE NATURE RESERVE, GAUTENG PROVINCE, SOUTH AFRICA. ...................................................................................................... 107 5.1 Introduction ............................................................................................................................. 109 5.2 Materials and Method ............................................................................................................ 113 5.2.1 Study area ................................................................................................................................. 113 5.2.2 Methods of data acquisition, processing, and analysis ......................................................... 114 5.3 Results ...................................................................................................................................... 116 5.3.1 A spatio-temporal analysis of EVI and other environmental factors .................................. 116 5.3.2 Statistical analysis of vegetation and climatic factors ........................................................... 120 5.3.3 Principal Component Analysis (PCA) .................................................................................... 123 5.4 Discussion................................................................................................................................. 126 5.5 Conclusion ............................................................................................................................... 127 Chapter 6 UTILISING RGB DRONE IMAGERY AND VEGETATION INDICES FOR ACCURATE ABOVE-GROUND BIOMASS ESTIMATION: A CASE STUDY OF THE CRADLE NATURE RESERVE, GAUTENG PROVINCE, SOUTH AFRICA ......................... 129 6.1 Introduction ............................................................................................................................. 131 6.2 Materials and Method ............................................................................................................ 133 6.2.1 Study Area ................................................................................................................................ 133 6.2.2 Procedure for doing random field measurements of chlorophyll using the Apogee MC-100 xiii Chlorophyll Concentration Meter and N3 IMU GNSS Receiver ............................................... 135 6.2.3 Construction of 3D models using the techniques of photogrammetry and UAV imaging. 135 6.2.4 Use of AGB 50m resolution for Southern Africa (Bouvet et al., 2018) ................................ 136 6.2.5 Analysis of vegetation using QGIS and UAV imaging ......................................................... 137 6.2.6 Performing Principal Component Analysis (PCA) and Stepwise Regression Analysis using Minitab Version 18 ......................................................................................................................... 138 6.3 Results and Analysis ............................................................................................................... 139 6.3.1. Comparison of variable statistics between non-riparian and riparian areas .................... 139 6.3.2. Comparison of principal components analysis (PCA) results ............................................. 140 6.3.3. Pairwise comparison of regression models for non-riparian and riparian plots in the Cradle Nature Reserve ................................................................................................................... 140 6.3.4. Comparative performance of regression models for non-riparian and riparian plots in the Cradle Nature Reserve ................................................................................................................... 142 6.4 Discussion................................................................................................................................. 144 6.4.1. Effectiveness of RGB drone imagery in estimating above-ground biomass ...................... 144 6.4.2 Differentiating predictive performance in riparian and non-riparian vegetation models 145 6.4.3 Limitations of RGB drone imagery, nDSM, and chlorophyll measurements in estimating above-ground biomass in riparian and non-riparian areas........................................................ 146 6.5 Conclusion ............................................................................................................................... 147 Chapter 7 MODELLING TOPOGRAPHIC INFLUENCES ON VEGETATION VIGOUR IN THE CRADLE NATURE RESERVE, GAUTENG PROVINCE, SOUTH AFRICA ................ 149 7.1 Introduction ............................................................................................................................. 151 7.2 Method and Materials ............................................................................................................ 155 7.3 Results and Discussion ............................................................................................................ 159 7.3.1 Evaluation of environmental factors and vegetation vigour ................................................ 159 7.3.2. Determination of vegetation health status along slope gradients ....................................... 161 7.3.3. Statistical relationships among the environmental factors and vegetation vigour ........... 163 7.4. Conclusion .............................................................................................................................. 173 Chapter 8 IMPACTS OF LAND USE CHANGE ON ANNUAL EVAPOTRANSPIRATION IN THE CRADLE NATURE RESERVE (2000-2023) USING MODIS DATA AND XGBOOST CLASSIFICATION: A CASE OF THE CRADLE NATURE RESERVE, GAUTENG PROVINCE, SOUTH AFRICA ....................................................................................................... 174 8.1 Introduction ............................................................................................................................. 176 8.2 Materials and Method ............................................................................................................ 178 xiv 8.2.1 Study area ................................................................................................................................. 178 8.2.2 Methods of data acquisition, processing, and analysis ......................................................... 179 8.3 Results and Analysis ................................................................................................................... 181 8.3.1 Annual Evapotranspiration Statistics in the Cradle Nature Reserve (2000-2024) ............ 181 8.3.2 Analysis of October Evapotranspiration Fluctuations in the Cradle Nature Reserve from 2000 to 2023 Using MODIS Data .................................................................................................. 183 8.3.4 Land cover classification using XGBRFClassifier in Cradle Nature Reserve ................... 185 8.3.5 Land cover classification using XGBoost Random Forest Classifier (2000-2023) ............. 186 8.3.6 Average October Evapotranspiration Across Various Land Cover Categories in the Cradle Nature Reserve (2000-2023)........................................................................................................... 189 8.3.7 Land Cover Changes in the Cradle Nature Reserve (2000-2023) ........................................ 191 8.4 Discussion................................................................................................................................. 192 8.4.1 Analysis of evapotranspiration fluctuations in the Cradle Nature Reserve from 2000 to 2023 using MODIS data ................................................................................................................. 192 8.4.2 Interpreting the Relationship Between Evapotranspiration and Elevation in the Cradle Nature Reserve ............................................................................................................................... 194 8.4.3 Impact of Landsat Sensor Improvements on Land Cover Classification Accuracy in the Cradle Nature Reserve October (2000-2023) ............................................................................... 194 8.4.4 Analysis of Spatial Variation and Assessment of the Effects of Changes in Land Cover . 195 8.4.5 Analysis of changes in land cover and their impact on the health of ecosystems ............... 196 8.4.6 Addressing Uncertainties in Data and Model Predictions ................................................... 196 8.5 Conclusion ............................................................................................................................... 197 Chapter 9 SYNTHESIS .................................................................................................................... 199 REFERENCE .................................................................................................................................... 205 Appendices ......................................................................................................................................... 261 Appendix A: Supplementary Tables ................................................................................................... 261 List of Figures Figure 1- 1: Study Area—Cradle Nature Reserve in South Africa with overview of the Google Earth Pro Imagery in UTM/WGS84 plane coordinate. ................................................. 8 xv Figure 2- 1: Schematic workflow for Scopus Document Search and the Bibliometrics Using VOS Viewer and Scopus Analysis Tool. ................................................................................. 22 Figure 2- 2: World Articles per Year from Scopus Database from 1990 to 2022 ................... 24 Figure 2- 3: South Africa Articles per Year from Scopus Database from 1990 to 2022 ......... 25 Figure 2- 4: World Articles per year per Source from Scopus Database from 1990 to 2022 .. 28 Figure 2- 5: Worldwide Density Visualisation-Scopus database 1990 to 2022 ...................... 31 Figure 2- 6: South Africa Density Visualisation-Scopus database 1990 to 2022 .................... 31 Figure 2- 7: Worldwide Network Visualisation—Scopus database 1990 to 2022, cluster 1 = red, cluster 2 = green, cluster 3 = blue ..................................................................................... 32 Figure 2- 8: Worldwide remote sensing and associated technologies 1990 to 2022, cluster 1 = red, cluster 2 = green, cluster 3 = blue ..................................................................................... 33 Figure 2- 9: Worldwide geographic information systems and associated technologies 1990 to 2022, cluster 1 = red, cluster 2 = green, cluster 3 = blue. ........................................................ 33 Figure 2- 10: Worldwide Virtual Reality and associated technologies 1990 to 2022, cluster 1 = red, cluster 2 = green, cluster 3 = blue .................................................................................. 34 Figure 3- 1: Examination of Research Publications in Remote Sensing and Vegetation Mapping ................................................................................................................................... 56 Figure 3- 2: Visualisation of Key Research Themes in Vegetation Mapping Using Remote Sensing, (a) Network Visualisation, (b) Density Visualisation................................................ 63 Figure 4- 1: Study Area – Cradle Nature Reserve in South Africa with Overview of the Google Earth Pro Imagery in UTM/WGS84 plane coordinate ................................................ 74 Figure 4- 2: Schematic of Workflow for Satellite data Downloading, Processing, XGBClassification, Accuracy Assessment, and QGIS processing .......................................... 76 Figure 4- 3: (a) XGBRFClassifcation image 1990; (b) XGBRFClassifcation mage 1998; (c) XGBRFClassifcation image 2009; (d) XGBRFClassifcation image 2015; (e) XGBRFClassifcation image 2020............................................................................................ 91 Figure 4- 4: (a) LULC change 1990–1998; (b) LULC change area 1990–1998; (c) LULC change 1998–2009; (d) LULC change area 1998–2009; (e) LULC changes 2009–2015; (f) LULC change area 2009–2015; (g) LULC change 2015–2020; (h) LULC change area 2015– 2020; (i) LULC change 1990–2020; (j) LULC change area 1990–2020. ................................ 98 Figure 5- 1: Study Area – Cradle Nature Reserve in South Africa with Overview of the Google Earth Pro Imagery in UTM/WGS84 plane coordinate .............................................. 114 Figure 5- 2: Schematic of workflow for data collection, processing, EVI computation, surface map generation, extraction of data statistics, pairwise correlation analysis, PCA, and results analysis ................................................................................................................................... 116 Figure 5- 3: A spatio-temporal analysis of EVI and other environmental factors during winter season from 2019-2023. ......................................................................................................... 118 Figure 5- 4: A spatio-temporal analysis of EVI and other environmental factors during summer winter season from 2019-2023. ................................................................................ 119 Figure 5- 5: Biplot of Variables and the most influential factors F1 and F2 ......................... 125 Figure 6- 1: Study Area-Cradle Nature Reserve in South Africa, Gauteng Province ........... 134 xvi Figure 6- 2: Scatter plot of residuals versus predicted AGB for Plot1 (non-riparian). .......... 143 Figure 6- 3: Scatter plot of residuals versus predicted AGB for Plot 2 (riparian). ................ 143 Figure 7- 1: The spatial distribution of a) enhanced vegetation index (EVI), b) topographic wetness index (TWI), c) topographic position index (TPI), d) topographic ruggedness index (TRI), e) air temperature (Temp), f) root zone soil moisture (Soil), g) total precipitation (RF) ................................................................................................................................................ 160 Figure 7- 2: Location of the study (Matyukira & Mhangara, 2023b) .................................... 161 Figure 7- 3: Relationship Between the Enhanced Vegetation Index (EVI) and Slope Gradient ................................................................................................................................................ 163 Figure 7- 4: Relationship between the Enhanced vegetation Index (EVI) and topographic wetness Index (TWI).............................................................................................................. 166 Figure 7- 5: Relationship between topographic Wetness Index and slope gradient. ............. 168 Figure 7- 6: Relationship between the topographic position Index (TPI) and EVI. .............. 170 Figure 7- 7: Relationship between the topographic ruggedness Index (TPI) and EVI. ......... 172 Figure 8- 1: Study Area-Cradle Nature Reserve in South Africa, Gauteng Province ........... 179 Figure 8- 2: Annual Evapotranspiration Variation (2000-2023) ........................................... 182 Figure 8- 3: Spatial Annual Evapotranspiration Variation (2000-2024) ............................... 184 Figure 8- 4: Temporal land cover classification using XGBoost Random Forest Classifier (2000-2023)............................................................................................................................ 187 Figure 8- 5: Evapotranspiration (ET) estimates across different land cover types derived using zonal statistics in QGIS version 3.28.3-Firenze..................................................................... 189 Figure 8- 6: Area covered by land cover types derived using zonal statistics in QGIS version 3.28.3-Firenze. ....................................................................................................................... 191 List of Tables Table 2- 1: Scopus search engine and queries used for the scope of this study. ..................... 17 Table 2- 2: Worldwide Articles per Year from Scopus Database from 1990 to 2022............. 24 Table 2- 3: World Articles per Year from Scopus Database from 1990 to 2022 .................... 25 Table 2- 4: Top 5 Worldwide articles per journal from Scopus Database from 1990 to 2022 28 Table 2- 5: Common Source Articles ...................................................................................... 29 Table 2- 6: Worldwide network visualisation map structural information .............................. 34 Table 3- 1: Scopus search engine and queries used for the scope of this study ...................... 53 Table 3- 2: Worldwide Articles per Year from Scopus Database from 2000 to 2024............. 54 Table 3- 3: Top 5 articles per journal from Scopus database from 2000 to 2024 .................... 57 Table 3- 4: Network visualisation map structural information ................................................ 61 Table 4- 1: Descriptions of the LULC types ............................................................................ 79 Table 4- 2: Hyperparameters configurations. .......................................................................... 82 Table 4- 3: Hypotheses for landscape metrics ......................................................................... 84 Table 4- 4: Accuracy Assessment for NaiveBayes and XGBRFClassifier for 2020 Confusion Matrix Derivatives ................................................................................................................... 86 xvii Table 4- 5: Accuracy Assessment for Ground Truthing Confusion Matrix Derivatives ......... 87 Table 4- 6: Accuracy Assessment for XGBRFClassifier Confusion Matrix Derivatives ....... 88 Table 4- 7: Land Use-Land Cover Spatial-Temporal Change Detection Matrices ................. 92 Table 4- 8: Landscape Metrics from 1990 to 2020 .................................................................. 99 Table 5- 1(a-b): Correlation and Regression Analysis of EVI with Climatic and Soil Variables (n=60) .................................................................................................................... 120 Table 5- 2(a-d): Statistical Analysis of Environmental Factors Influencing EVI ................. 121 Table 5- 3: Descriptive Statistics and Principal Component Analysis of Environmental Variables Affecting EVI (n= 60) ........................................................................................... 123 Table 5- 4: Factor Loadings and Variable Correlations in Environmental Data Analysis .... 124 Table 5- 5: Variable Contributions and Squared Cosines in Principal Component Analysis 126 Table 6- 1: Vegetation index equations using an RGB drone image. .................................... 137 Table 6- 2: Summary statistics plot 1 (non-riparian). ............................................................ 139 Table 6- 3: Summary statistics plot 2 (riparian). ................................................................... 139 Table 6- 4: Stepwise regression modelling Plot 1(non-riparian). .......................................... 141 Table 6- 5: Stepwise regression modelling Plot 2 (riparian). ................................................ 141 Table 7- 1: Slope gradient classes (Food and Agriculture Organization of the United Nations., 2006) ...................................................................................................................................... 158 Table 7- 2: Vegetation health and density value interpretation. ............................................ 159 Table 7- 3: Summary of topographic, vegetation, meteorological, and hydrological variables. ................................................................................................................................................ 161 Table 7- 4: Relationship between the Enhanced vegetation Index (EVI) and slope gradient 162 Table 7- 5: Correlation of EVI with climatic and topographic variables (n = 500). .............. 164 Table 7- 6: Multi-linear regression analysis using all the pixels of the remote sensing data.165 Table 7- 7: Relationship between Enhanced vegetation Index (EVI) and topographic wetness Index (TWI). .......................................................................................................................... 166 Table 7- 8: Relationship between the TWI and slope gradient .............................................. 168 Table 7- 9: Relationship between the TPI and EVI. .............................................................. 170 Table 7- 10: TRI and EVI relationship. ................................................................................. 172 Table 8- 1: Annual Evapotranspiration Statistics mm (2000-2024) ...................................... 181 Table 8- 2: Regression coefficients........................................................................................ 185 Table 8- 3: Performance Metrics of XGBRFClassifier for Land Cover Classification in the Cradle Nature Reserve (2000-2023) ...................................................................................... 188 Table 8- 4: Average October Evapotranspiration (ET) Values (mm) for Different Land Covers in the Cradle Nature Reserve (2000-2023) ................................................................ 189 Table 8- 5: Summary of Statistical Analysis ......................................................................... 190 Table 8- 6: Area Covered by Land Cover Types in the Cradle Nature Reserve (2000-2023) in Hectares.................................................................................................................................. 191 Table A 1: Multicollinearity and Principal Component Analysis Plot 1 ............................... 261 Table A 2: Multicollinearity and Principal Component Analysis Plot 2 ............................... 263 1 Chapter 1 INTRODUCTION 2 1.1 Background Geospatial technologies have become essential tools in archaeological research and environmental monitoring, particularly at sites with dual designations like game reserves. These technologies facilitate archaeological site identification, mapping, and analysis, supporting ecosystem management and vegetation monitoring (Ahmad & Umar, 2024; DiBiase et al., 2010). Over the years, the deployment of these technologies across various fields, including archaeology, has significantly increased, resulting in a rich body of literature (Ahmad & Umar, 2024; DiBiase et al., 2010; Y. Huang, 2024). Geospatial technology is variably defined as a science, a group of tools used in various professional fields, or a profession with specific competence standards (Kamraju, 2023; Malloy Nicolas R., 2024). It encompasses land surveying, remote sensing, mapping, geographic information systems (GIS), geodesy, and global navigation systems, collectively forming a geospatial industry (Aina, 2012; DiBiase et al., 2010). Historically, geospatial technologies in archaeology have evolved from producing hand-drawn maps to using advanced tools for visualising geometric structures and detecting settlement patterns. Robust geospatial technologies, capable of handling high volumes of spatial data, are responsible for this growth (Wheatley & Gillings, 2013). Modern applications of these technologies, including remote sensing, GIS, and 3D modelling, have become integral to archaeological investigations (Feder, 2016a; Hester, 2016b). Similarly, remote sensing technologies, a crucial subset of geospatial technologies, have become essential tools for environmental management and monitoring at archaeological sites and game reserves (Elfadaly et al., 2018; Lindsay & Mkrtchyan, 2023). These technologies enable the assessment of vegetation cover, condition, and temporal changes without compromising the archaeological integrity of the site (Elfadaly et al., 2018; Lindsay & Mkrtchyan, 2023). Advancements in sensor technology, satellite platforms, and UAVs have made it much easier to collect accurate environmental data at these dual-purpose sites, which is crucial for understanding how ecosystems work (Cui et al., 2023; Vidican et al., 2023). Continuous observation and detailed analysis facilitated by remote sensing are vital for efficient decision-making in conservation, land use planning, and resource management, ensuring that archaeological sites remain undisturbed while supporting the ecological health of the game reserve (Rocchini, 2014; Szpakowski & Jensen, 2019). Improvements in satellite imaging technology, particularly the enhanced spatial and spectral resolution of sensors like Landsat, Sentinel, and MODIS, have markedly increased the accuracy and reliability of vegetation maps at archaeological sites within game reserves (Hansen et al., 2013; Morell-Monzó et al., 2020; Pastick et al., 2018, 2020). Additionally, vegetation indices like the NDVI have become robust tools for assessing vegetation health and productivity without impacting the archaeological integrity of the site (Benhizia et al., 2024; Niyonsenga et al., 2024). Integrating advanced machine learning algorithms with remote sensing data has improved vegetation analysis. It is now easier to track plant growth, spot changes in land cover, measure land degradation and fragmentation, figure out how terrain affects plant health, and estimate biomass and evapotranspiration at these one-of-a-kind sites (Bauer et al., 2024; Mullissa et al., 2024; Pettorelli et al., 2005). These technological advancements are critical for managing and researching complex environments like the Cradle Nature Reserve. Understanding the complex interrelationships among vegetation dynamics, hydrology, and topography within this karst environment can significantly benefit from machine learning algorithms and advanced modelling techniques (Chang et al., 2022; Loehman et al., 2020). 3 Understanding and addressing the intricate interrelationships among vegetation dynamics, hydrology, and topography are crucial in the Cradle Nature Reserve. This ecosystem is characterised by its intricate topography, where variables such as moisture distribution, vegetation condition, and terrain ruggedness significantly impact ecological processes (Berhanu & Bisrat, 2018; Kopecký et al., 2021). Key metrics like the TWI and EVI provide essential information about vegetation distribution and moisture availability (Cantón et al., 2004; Talebi & Mostafazadeh, 2022). Slope gradients also play a crucial role in influencing habitat suitability, water runoff, and soil erosion, further complicating ecosystem management (Talebi & Mostafazadeh, 2022; M. Wu et al., 2018). Other topographic indices, such as the TPI and TRI provide critical insights into hydrological processes and landscape features (Al- Sababhah, 2023; Dilts et al., 2022; K. B. Jones et al., 2000; Rózycka et al., 2017). Understanding the synergies and trade-offs among these indicators is essential for developing effective conservation and management strategies that ensure the long-term viability and resilience of the dual-purpose site's fragile karst ecosystem (Berhanu & Bisrat, 2018; Kopecký et al., 2021). This comprehensive approach is particularly beneficial in studying ET, a critical hydrological process in such environments. ET, a key component of the hydrological cycle, encompasses the processes of water evaporation from soil and water surfaces and plants' release of water vapour (U.S. Geological Survey, 2023a). At dual-purpose sites, such as those that are both archaeological and game reserves, understanding ET trends is essential for managing water resources, supporting vegetation health, and maintaining ecosystem balance without disrupting the archaeological sites. Efficient water resource management ensures that the flora and fauna within the reserve are sustained, which is vital for the habitat's ecological health and biodiversity (Durán-Sánchez et al., 2018; Stephenson et al., 2022). Simultaneously, maintaining the water cycle supports the preservation of archaeological artefacts by preventing erosion and other water-related damages. Traditional methods for determining ET include hydrological, meteorological, and micro- meteorological techniques, such as the Penman-Monteith equation and eddy covariance systems (Food and Agriculture Organization of the United Nations, 2024; Shoko et al., 2015). However, these methods are often limited by their spatial and temporal resolution (Derardja et al., 2024; Tan et al., 2021). Remote sensing technology has revolutionised ET estimation by providing extensive, continuous, and replicable data on a large scale (Derardja et al., 2024). There are full and ongoing ET observations in the MODIS data, especially in the MODIS/061/MOD16A2GF dataset, which lets us study changes over time from 2000 to 2023 (LP DAAC 2024). Additionally, advancements in remote sensing technologies have facilitated more accurate and extensive monitoring of ET, which is essential for managing water resources and studying climate change effects (Derardja et al., 2024; L. Jin et al., 2024). Machine learning algorithms, like the XGBRFClassifier, have improved land cover classification by making it easier to analyse big, complicated datasets more accurately and quickly (Talukdar et al., 2020a). Recent developments in Landsat sensor technology, especially with Landsat 8 and 9, have further enhanced the accuracy and reliability of land cover classifications (Phiri & Morgenroth, 2017). Balancing the dual roles of archaeological preservation and ecosystem management in game reserves requires careful planning and the use of advanced geospatial technologies. These technologies protect archaeological sites from anthropogenic and natural disturbances while enabling sustainable management of the game reserve's ecosystem. These sites make a 4 significant economic contribution to the nation and surrounding communities, providing tourism revenue, job opportunities, and community development. By integrating geospatial technologies, we can preserve cultural heritage and natural resources for future generations. 1.2 Problem Statement The application of geospatial technologies in archaeology has made significant advancements, but a systematic review of the existing literature is lacking, leading to a fragmented understanding of trends and gaps in research (Klehm, 2023a). The use of geospatial technologies in archaeology has been shown to enhance the accuracy and efficiency of site identification, mapping, and monitoring (Wheatley & Gillings, 2013). However, despite the growing interest in remote sensing and machine learning technologies for vegetation mapping, there is a notable lack of comprehensive literature reviews synthesising the existing research in this field, particularly in the context of the Cradle Nature Reserve, a unique site that combines both archaeological significance and game reserve conservation efforts, presenting a complex dual configuration that warrants specialised attention (Brembs, 2018; Whig et al., 2024a; Y. Xie et al., 2008a). Vegetation mapping is a critical component of archaeological site management, as it provides valuable insights into site formation processes and human- environment interactions (Pettorelli et al., 2005). Moreover, vegetation mapping is essential for understanding ecosystem dynamics, biodiversity conservation, and land degradation assessment (Deb et al. 2022). The Cradle Nature Reserve, a UNESCO World Heritage Site, is an ideal location for applying remote sensing techniques in vegetation mapping due to its unique geological and ecological features, including its location in the Cradle of Humankind, a region of high archaeological significance (Bauer et al. 2024). The reserve's diverse vegetation cover, ranging from grasslands to forests, also presents an opportunity to test the efficacy of remote sensing techniques in different environments (Mullissa et al. 2024). Therefore, a literature review on geospatial technologies in archaeology should be followed by a focused literature review on the application of remote sensing and machine learning in vegetation mapping in the Cradle Nature Reserve to further explore the potential of these technologies in this specific context and identify areas for future research and application The Cradle Nature Reserve experiences significant challenges in maintaining ecological health due to the complex interplay between topography, vegetation dynamics, and hydrology. These challenges are exacerbated by the rugged terrain and varying moisture distribution, which affect plant health and ecological processes. The existing body of research highlights the need for comprehensive studies integrating topographic indices such as TWI, EVI, TPI, and TRI to understand these interactions better (Berhanu & Bisrat, 2018; Kopecký et al., 2021). Despite the availability of various geospatial tools and indices, there remains a gap in the application of these tools to fully capture and analyse the intricate details of the karst environment. This gap underscores the need for more detailed and scale-appropriate studies to provide actionable insights for sustainable land management and conservation efforts (Cantón et al., 2004; Kopecký et al., 2021). Additionally, the Cradle Nature Reserve faces challenges from varying climatic conditions and land management practices, affecting ET patterns and vegetation health—critical indicators of the reserve's ecological state (Mccollum et al., 2017; Pedrinho et al., 2024). Previous studies have identified fluctuations in ET due to climate variability and human activities, but a comprehensive understanding of these trends over an extended period remains incomplete (Arnell et al., 2019; Gerten et al., 2013). The relationship between ET and topographical 5 features, such as elevation, requires a thorough investigation to develop adaptive management practices that consider both climatic and biophysical factors in the game reserve (Nakileza & Nedala, 2020; Q. Yang et al., 2022). Even though remote sensing and machine learning have come a long way, the Cradle Nature Reserve still needs constant monitoring and accurate land cover classification to support good conservation strategies and ecological management (Manandhar et al., 2009; Phiri & Morgenroth, 2017). Land cover degradation is evident in the Cradle Nature Reserve, marked by a significant increase in bare ground and a reduction in indigenous forests and natural grasslands (L. Berger & Maduwa, 2023). This degradation leads to habitat loss, biodiversity decline, soil erosion, and ecosystem alterations, negatively impacting the area's aesthetic and tourism value (Adepoju & Salami, 2017; Martinez del Castillo et al., 2015). The invasion of alien plant species, particularly pompom weed, exacerbates this degradation, reducing the vegetation's productivity and the area's grazing capacity (O’Connor & van Wilgen, 2020). There have been improvements in machine learning algorithms, but there hasn't been a comparison of the parametric Naïve Bayes and non-parametric XGBRF classifiers for LULC classification in this region. This means that more research is needed to find the best classifier for tracking changes in land cover and figuring out how much landscape fragmentation is caused by humans and climate change. Advancements in remote sensing technology exist, but the variability and interactions of environmental factors, such as temperature, rainfall, and soil moisture, that collectively influence the health and productivity of vegetation in grassland ecosystems in archaeological settings such as the Cradle Nature Reserve are outstanding (Mashala et al., 2023; Ogungbuyi et al., 2023a). These factors exhibit complex and often non-linear relationships, making it challenging to develop accurate and reliable models for ecological monitoring and decision- making (Jobson, 1991). The inherent seasonality of these environmental factors can further complicate the analysis, potentially leading to misleading conclusions if not properly accounted for (Helali et al. 2022). Multilinear regression models, commonly used in these kinds of studies, do not always show all of these complex relationships because the independent variables aren't always linear, and seasonality is not considered (Jobson, 1991; Rebecca Bevans, 2023). According to (Fatima et al., 2022) and (Marukatat, 2023), Principal Component Analysis (PCA) is a better way to deal with these kinds of problems because it reduces the number of dimensions in the data and turns correlated variables into uncorrelated components. This makes it easier to see the patterns and relationships that lie beneath the surface. The Cradle Nature Reserve, a unique site combining archaeological significance and game reserve conservation efforts, requires accurate biomass estimation for effective vegetation management and conservation. However, the use of drone technology for biomass estimation in this context faces challenges, including reduced precision during early growth stages and varying vegetation types (Bareth et al., 2015; Possoch et al., 2016). While vegetation indices derived from RGB drone imagery can provide insights into plant health and characteristics, integrating additional data sources, such as field-measured chlorophyll concentrations, is essential for improving biomass estimate accuracy (Agapiou, 2020). The lack of research addressing these challenges in the Cradle Nature Reserve's dual configuration setting hinders the development of effective vegetation management strategies, emphasising the need for targeted investigations to overcome these knowledge gaps and improve biomass estimation accuracy in this unique environment. 6 1.3 Research questions The following research questions were addressed in this study: • What are the key concepts and themes shaping the use of geospatial technologies in archaeology, and which potential research gaps and underexplored areas remain in this field, both globally and in South Africa? • What are the key applications of remote sensing in monitoring vegetation, detecting land cover changes, assessing degradation, evaluating topography's impact, and estimating biomass and evapotranspiration? • Which classifier, Naïve Bayes or Extreme Gradient Boosting Random Forest, demonstrates higher accuracy for LULC classification in the Cradle Nature Reserve in 2020, and what are the land use and land cover changes from 1990 to 2020? • What are the spatio-temporal trends of EVI and environmental factors, and how do their statistical associations with climatic factors inform sustainable grassland management using remote sensing, multilinear correlation, and PCA in the Cradle Nature Reserve? • What factors influence above-ground biomass in the Cradle Nature Reserve, using RGB drone photos, nDSM vegetation heights, and field-measured chlorophyll analysed with PCA and stepwise regression? • Which topographic factors, TWI, EVI, TRI, TPI, and percentage slope gradient, collectively affect vegetation growth and changes, moisture distribution, terrain roughness, and habitat suitability in the Cradle Nature Reserve's karst environment? • What are ET's yearly patterns and spatial distribution, and how does ET correlate with DEM data and land cover changes in the Cradle Nature Reserve? 1.4 Aims and objectives of the study • To investigate the key concepts and themes shaping the use of geospatial technologies in archaeology and to identify potential research gaps and underexplored areas in this field, both globally and in South Africa. • To investigate the key applications of remote sensing in monitoring vegetation, detecting land cover changes, assessing degradation, evaluating topography's impact, and estimating biomass and evapotranspiration. • To investigate which classifier, Naïve Bayes or Extreme Gradient Boosting Random Forest, demonstrates higher accuracy for LULC classification in the Cradle Nature Reserve in 2020 and to determine the land use and land cover changes from 1990 to 2020. • To investigate the spatio-temporal trends of EVI and environmental factors and examine their statistical associations with climatic factors to inform sustainable grassland management using remote sensing, multilinear correlation, and PCA in the Cradle Nature Reserve. 7 • To investigate the factors influencing above-ground biomass in the Cradle Nature Reserve, using RGB drone photos, nDSM vegetation heights, and field-measured chlorophyll analysed with PCA and stepwise regression. • To investigate how topographic factors, TWI, EVI, TRI, TPI, and percentage slope gradient collectively affect vegetation growth and changes, moisture distribution, terrain roughness, and habitat suitability in the Cradle Nature Reserve's karst environment. • To investigate ET's yearly patterns and spatial distribution and to explore how ET correlates with DEM data and land cover changes in the Cradle Nature Reserve. 1.5 Study area The Cradle Nature Reserve, encompassing approximately 8,000 hectares within the Cradle of Humankind World Heritage Site (COHWHS), is situated in the Sterkfontein Valley, about 50 kilometres northwest of Johannesburg, South Africa, and 10 kilometres north of Krugersdorp (Lelliott, 2016; Rogerson & van der Merwe, 2016). The geographical coordinates of the reserve are between longitudes 27°42'58" and 27°52'57" and latitudes 25°51'13" and 25°51'19" (see Figure 1). This location was designated a World Heritage Site by UNESCO in 1999 due to its profound significance in the study of paleoanthropology (Rogerson & van der Merwe, 2016). The reserve is renowned for its diverse plant species, particularly those that bloom and hosts over 200 bird species alongside various other animals (Braga et al., 2016; Rogerson & van der Merwe, 2016). Vegetation in the reserve is influenced by key climatic factors such as temperature and precipitation, with annual rainfall ranging from 650 to 750 mm, summer temperatures reaching up to 39°C, and winter temperatures dropping as low as -12°C, creating a dynamic environment conducive to plant growth (FLOW Communications, 2022). The Rocky Highveld Grassland climatic area, characterised by fire, significantly impacts the plant structure. Additionally, natural springs, watercourses, streams, and tributaries to the Magalies and Crocodile Rivers provide essential moisture, supporting the area's wide range of plant species (Caruana & Stratford, 2019; Lelliott, 2016). Historically, both subsistence and commercial farming practices have significantly impacted plant distribution within the reserve; additionally, the reintroduction of wild animal species, which exhibit distinct grazing patterns compared to domesticated animals, has further influenced these alterations (Makokotlele, 2009). Unique geological characteristics, such as dolomitic sinkholes that protect tree species from forest fires and the chemical weathering of limestone and dolomitic rocks forming karst landforms, contribute to the distinct patterns of plant growth in the area (Bradley et al., 2010; Eloff, 2010). 8 Figure 1- 1: Study Area—Cradle Nature Reserve in South Africa with overview of the Google Earth Pro Imagery in UTM/WGS84 plane coordinate. 1.6 Significance of the study This study aims to illuminate trends and advancements in geospatial technologies for archaeological investigations, specifically focusing on South Africa. By identifying research gaps and showcasing recent developments in affordable geospatial tools, the study underscores their significance within the broader academic landscape. Encouraging the adoption of these methods in South African archaeology and fostering interdisciplinary collaboration will enhance spatial technology applications, ultimately advancing the field and contributing to global archaeological practices. Furthermore, the study aims to enhance vegetation mapping by integrating remote sensing technologies with machine learning algorithms. Through a comprehensive literature review using Scopus and employing VOSviewer for bibliometric analysis, the research will highlight key trends and advancements in the utilisation of remote sensing techniques such as Random Forest, Support Vector Machines (SVM), neural networks and XGBRFClassifier for improving vegetation map accuracy. These insights are crucial for identifying areas where research is lacking, facilitating more accurate environmental management, and aiding in creating successful conservation programs and sustainable development practices. Conducting ongoing studies in these specific areas of deficiency will improve our capacity to reduce the effects of climate change and other environmental pressures. Additionally, this study will provide critical insights into land use and land cover changes (LULCC) in the Cradle Nature Reserve from 1990 to 2020 using satellite imagery and the 9 XGBRFClassifier. By classifying the landscape, the research will reveal trends in environmental degradation and landscape fragmentation. These changes, driven by both natural dynamics and anthropogenic factors like invasive species, underscore the need for effective conservation strategies. The study will support the Cradle Nature Reserve's efforts to eradicate invasive species, restore degraded landscapes, and maintain ecological integrity. The study will assist in developing evidence-based policies and management plans that will contribute to the sustainable preservation of this important ecological and historic site through comprehensive analysis. Furthermore, the study will enhance ecological decision-making and sustainable land management in the Cradle Nature Reserve through PCA and correlation analysis. By understanding environmental factors, the research will support evidence-based policies to maintain grassland ecosystems, ensuring the health of vegetation and dependent wildlife. The study will guide policymakers in promoting healthy vegetation, creating seasonal plans, allocating resources to stressed areas, and formulating climate adaptation strategies. Continuous monitoring will assess policy effectiveness and adapt strategies as needed. By identifying key environmental variables, managers can prioritise climatic factors in conservation efforts and implement targeted interventions. The research aims to simplify complex ecological data into actionable insights, guiding grassland management strategies to maintain ecological health and sustainable use of these ecosystems. This study's significance also lies in its potential to support informed conservation policies, sustainable land management, and biodiversity preservation by uncovering significant variations in water availability and vegetation health due to climate variability and land management practices. By examining the relationship between evapotranspiration (ET) and elevation, the study will provide insights into the influence of biophysical and climatic factors, emphasising the need for adaptive management to maintain ecosystem health. The findings will underscore the importance of continuous monitoring and effective management strategies to sustain the reserve's health and biodiversity amidst climate change impacts. Additionally, this study's significance lies in addressing the complex interactions between vegetation dynamics, hydrology, and topography in the Cradle Nature Reserve. By integrating topographic indices such as TWI, EVI, TPI, and TRI, the study will comprehensively analyse how these factors collectively impact vegetation health, moisture distribution, terrain ruggedness, and habitat suitability, which is crucial for effective conservation and sustainable land management. The research will utilise advanced remote sensing technologies and machine learning algorithms to ensure robust and reliable findings, informing adaptive management practices amidst climate variability and human activities. This contribution to environmental science emphasises the importance of continuous monitoring and diverse geospatial tools, providing practical solutions for global biodiversity conservation and ecological balance. In summary, this study aims to provide critical data and insights to support informed decision- making in environmental management and conservation efforts, contributing significantly to the fields of environmental science and ecological management. 10 1.7 Outline of the thesis This is a thesis that utilises a paper-based technique. The thesis consisted of nine chapters, of which seven chapters detailed the achievement of the given objectives: Chapter 1 provides the background, problem statement, objectives, and outline of the research. Chapter 2 provides a scientometric review of the advancements in the application of geospatial technology in archaeology and cultural heritage in South Africa. The scientometric review highlights significant research growth and technological integration, emphasising global collaboration, key academic platforms, multidisciplinary contributions, and the need for ongoing research to enhance data accuracy and develop innovative solutions for environmental sustainability amidst climate change and degradation challenges. Chapter 3 provides a scientometric review of advances in vegetation mapping through remote sensing and machine learning techniques. The study revealed that the main drivers that propel research in vegetation mapping using remote sensing are advancements in machine learning algorithms and prominent academic platforms such as major journals and conferences. Chapter 4 assessed changes in land cover and landscape structure by applying extreme gradient boosting random forest and fragmentation analysis. The study found that the XGBRFClassifier excels in land cover classification, revealing significant land cover changes and increased fragmentation from 1990-2020 Chapter 5 investigated vegetation growth dynamics using EVI and PCA in the karst environment of the Cradle Nature Reserve. The study found that climatic conditions significantly influence vegetation health, with PCA effectively revealing the combined and individual impacts of the environmental factors Chapter 6 investigated the use of RGB Drone Imagery and vegetation indices for accurate above-ground biomass estimation. The study identified nDSM and RBVI as key predictors of AGB in non-riparian areas, while chlorophyll content and BGVI played important roles in riparian areas. Chapter 7 investigates the impact of topographic parameters TWI, EVI, TRI, TPI, and % slope gradient on plant development, moisture distribution, terrain roughness, and habitat compatibility in the Cradle Nature Reserve. The study found that vegetation health in the Cradle Nature Reserve is significantly influenced by topographic variables, with healthier vegetation typically found on steeper slopes and poorer vegetation in elevated and rugged terrains. Chapter 8 investigated the effects of land use change on annual evapotranspiration in the Cradle Nature Reserve (2000-2023) using MODIS data and XGBoost classification. The study identified fluctuations in evapotranspiration and land cover from 2000 to 2023. Chapter 9 concludes the thesis by highlighting the primary results attained for each target, the research's value to the remote sensing field, and suggestions for future investigations. 11 Chapter 2 ADVANCEMENT IN THE APPLICATION OF GEOSPATIAL TECHNOLOGY IN ARCHAEOLOGY AND CULTURAL HERITAGE IN SOUTH AFRICA: A SCIENTOMETRIC REVIEW This Chapter is based on the following manuscript Matyukira, C., & Mhangara, P. (2023). Advancement in the application of geospatial technology in archaeology and cultural heritage in South Africa: A scientometric review. Remote Sensing (Basel), 15(19), Article 4781. https://doi.org/10.3390/rs15194781 https://doi.org/10.3390/rs15194781 12 Abstract: Geospatial technologies have become an essential component of archaeological research, aiding in the identification, mapping, and analysis of archaeological sites. Several journals have published existing narratives on the development and impact of geospatial technologies in the study of archaeology and cultural heritage. However, this has not been supported by a systematic review of articles and papers, where meticulously collected evidence is methodically analysed. This article systematically reviews the trends in the use of geospatial technologies in archaeology and cultural heritage through the search for keywords or terms associated with geospatial technologies used in the two fields on the Scopus database from 1990 to 2022. Bibliometric analysis using Scopus Analyse tool and analysis of bibliometric networks using VOSviewer visualisations, reveals how modern archaeological studies are now a significant discipline of spatial sciences and how the discipline enjoys the tools of geomatics engineering for establishing temporal and spatial controls on the material being studied and observing patterns in the archaeological records. The key concepts or themes or distinct knowledge domains that shape research in the use of geospatial technologies in archaeology and cultural heritage, according to the Scopus database (1990–2022), are cultural heritage, archaeology, geographic information systems, remote sensing, virtual reality, and spatial analysis. Augmented reality, 3D scanning, 3D modelling, 3D reconstruction, lidar, digital elevation modelling, artificial intelligence, spatiotemporal analysis, ground penetrating radar, optical radar, aerial photography, and unmanned aerial vehicle (UAV) are some of the geospatial technology tools and research themes that are less explored or less interconnected concepts that have potential gaps in research or underexplored topics that might be worth investigating in archaeology and cultural heritage. Keywords: geospatial technologies, geographic information systems, lidar, remote sensing, South Africa, virtual reality 2.1 Introduction The application of geospatial technology in a broad spectrum of fields is gaining momentum worldwide. The widespread application of geospatial technology has resulted in a multiplicity of definitions owing to the distinct perspectives of the stakeholders. Some denote this application as a science, while others consider geospatial technology a group of tools used in various professional fields. Geospatial technology is also recognised as a profession with codes of ethics and specific competence standards (DiBiase et al., 2010). As a profession, geospatial technology, also referred to as geomatics, is a growing multidisciplinary academic field with diverse applications encompassing land surveying, remote sensing, mapping, geographic information systems (GIS), geodesy, and global navigation systems (Aina, 2012). This multidisciplinary academic field constitutes a geospatial industry defined by the United States Department of Labour as “an information technology field of practice that acquires, manages, interprets, integrates, displays, analyses, or otherwise uses data focusing on the context” (Klinkenberg, 2007). From this definition, (Klinkenberg, 2007) deduced that geospatial technologies should be classified as GIS, global navigation satellite systems (GNSS), photogrammetry, remote sensing, cartography, surveying, and other related fields. Aina (Aina, 13 2012) concurred with Klinkenberg (Klinkenberg, 2007) on the definitions. She added devices such as cellular phones, RFID (radio frequency identification) tags and surveillance cameras with embedded technologies that use locational information to the list of geospatial technologies. The fields in which geospatial technologies have been implemented are wide- ranging, including facilities management, precision farming, urban planning, business geographics, security and intelligence, telecommunication, automated mapping, and civil engineering. Archaeology is a further field where such technology has been applied (Aina, 2012). There is an established deep link between geospatial technologies and information communication systems, which has benefitted geomatics fields such as GIS (Aina, 2012). The above discussion elucidates what this literature review refers to as geospatial technologies. As early as the 18th century, archaeologists had grasped the importance of spatial data accompanying archaeological recordings, ranging from relative locations with varying scales to data depicting individual artefacts with excavation contexts (Wheatley & Gillings, 2013). Over the years, the deployment of geospatial technology in archaeology has been used to expound on different spatial phenomena. Such phenomena include seasonal hunter-gatherer camps within a landscape, the hierarchy of settlements within a region, the location of artisanal mines in an area, and many others that exploit spatial links, patterns, and relationships (Wheatley & Gillings, 2013). According to Aina, (2012), and Wheatley & Gillings, (2013), growth in the study of spatial archaeological data has grown from descriptive (tabulated and plotted on simple flat maps) to explanatory (explanation of spatial patterns and relations). Wheatley & Gillings, (2013) further attributed this development to emerging robust geospatial technologies capable of handling high volumes of spatial data with high resolutions and processing speeds. According to Nsanziyera and three other authors (Nsanziyera et al., 2018), the tremendous growth in remotely sensed spatial data has created new horizons and possibilities for archaeological research, such as creating predictive models that are becoming standard tools for investigation in GIS mapping. Further, these advancements in geospatial technologies have created new territories for research in archaeology and many documents in journals of different rankings. Although several journals have published existing narratives on the development and impact of geospatial technologies in the study of archaeology, as noted in this review and others, this has not been supported by a systematic review of articles and papers (Linnenluecke et al., 2020). In a systematic review, evidence is assembled, identified, and critically analysed through systematic procedure so that readers are constantly updated about current literature on the subject (Carrera-Rivera, Larrinaga, et al., 2022). According to Linnenluecke et al., (2020), the world is witnessing an exponential growth of poor-quality scholarly articles, often in open- access formats in predatory journals and good scholarly articles in ranked journals. However, according to Brembs, (2018), the reliability of even exceptionally high-ranking journals could be better, but these journals are currently the only source of credible information on geospatial technologies. This review aims to establish the trend in using and applying geospatial technologies in archaeology globally and in South Africa using the Scopus database. Scopus is an extensive database that provides fully referenced scholarly documents in social sciences, arts, humanities, science, technology, and medicine (Burnham, 2006). The following section gives a framework of the goals and objectives of using geospatial technologies in archaeology. 14 2.1.1 Geospatial Technology in Archaeology Before reviewing the goals and objectives of geospatial technology in archaeology, it is necessary to first look at the field of archaeology itself. The latter has undergone many changes over the past century. Each change attracts a new purpose or modification of the existing ones (Trigger, 1989, p. 370-411, as cited in (Shafer, 2016a)). Establishing temporal and spatial controls on the material being studied and observing patterns in the archaeological records forms the backbone of modern archaeology; however, some aspects have been retained from the older versions of archaeology, such as in archaeologies defined as cultural/historical, new, and or postprocessual (Shafer, 2016a). Historically, the usage of geospatial technologies in archaeology aimed to produce archaeological maps (primarily hand-drawn) meant to record/document site progress and archaeological investigations (Feder, 2016a, 2016b). For instance, the objective of traditional sketch maps was to record the in-situ context or geometrical descriptions of uncovered material evidence (Garrison et al., 2011; Kruger et al., 2016; Reilly, 1991; Shafer, 2016a, 2016b). With advanced geospatial technologies, modern complex maps aim to visualise geometric structures, taphonomic associations and linkages between in-situ and ex-situ fossils to illuminate a landscape signature and detect comprehensive settlement patterns. In this context, the settlement pattern represents how people relate and interact with a landscape, which archaeologists call the landscape signature (Ainsworth et al., 2017; Kruger et al., 2016; Reilly, 1991; Shafer, 2016a). The phenomenal growth in geospatial technologies over the past century has revolutionalised archaeological maps, particularly those used for site surveying and more changes are envisaged in the foreseeable future (Ainsworth et al., 2017; Garrison et al., 2011; Kruger et al., 2016; Reilly, 1991; Shafer, 2016a, 2016b). According to Feder, (2016a), archaeological sites are analogous to extinct animal fossils; a fossil is more than the sum of its parts. However, the juxtaposition and spatial positioning of the bones provide valuable information for palaeontology. Similarly, the juxtaposition and spatial distribution of archaeological remains provide insightful information about the habits of people using the site (Feder, 2016a, 2016b; Lazo, 2005). Archaeological and paleontological excavations, by nature, are destructive; remains left behind by historic and prehistoric people, which inherently are the archaeological source of knowledge, are removed or destroyed in the process of understanding our past (David, 2006; Feder, 2016a; Latham et al., 1999). Artefacts are removed from their original contexts and sent to the laboratory; in the process, spatial relationships are destroyed (Hester, 2016a, 2016b), so how do archaeologists capture and preserve the spatial relationships of the artefacts during site surveys? The answer is to carry out in situ recording (in a reference coordinate system) of the exact horizontal and vertical positions of artefacts. These same positions constitute recorded coordinates of the original location of an artefact, referred to in archaeological terms as provenience (Thomas, 2021). For instance, many items are left in situ to take proveniences before the excavation procedures destroy material evidence when excavating a site (Hester, 2016b). Irrespective of the survey methods employed, the exact location of these proveniences is the subject of geospatial technologies, be they pedestrian walking, subsurface survey, or remote sensing (Feder, 2016a) 15 The following sections are dedicated to expounding these technologies, first by defining form, space, and place, the essential ingredients when defining a spatial location, linking geospatial technologies to spatial archaeology, and tracking the development of these technologies globally and in South Africa. 2.1.2 Understanding Space and Place Before turning to the intricate issues surrounding the development of spatial technologies and how they have revolutionised archaeology in South Africa, a detour to understand the meaning of spatial archaeology is essential. Understanding the recursive linkages between spatial structure and social relations has been the subject of archaeology for years; of primary concern is the interpretation of the meaning of space and place regarding spatial structure in archaeological records (Ashmore, 2002). Albert Spaulding (1954, p. 161-162, as cited in (Ashmore, 2002)) defined space in archaeology as the dimensions used in empirical data operations to characterise and analyse artefacts and assemblages codified as forms with temporal and spatial locus. The description and classification of material evidence archaeologists uncover during their site investigations are called form (Shafer, 2016a). This definition shows that material evidence is diachronic because it has temporal and spatial characteristics (Shafer, 2016a). According to Ashmore, (2002), archaeological studies define a place as a plotted episode on a map through which the past becomes visible. From these definitions, space contains a place, and form exists in place boundaries. On the significance of place, Lewis Binford (1982, p. 5, as cited in (Ashmore, 2002)) implored archaeologists to acknowledge the subtle importance of place; he argued that in understanding the organisation of past cultural systems, it is paramount that archaeologists understand temporal and spatial relations among places which were used differentially during the operation of past systems. He reiterated these sentiments by adding that repeated human actions were essential in forming individual places and constructing social memories; a single place had multiple and mutable roles in society, and past social interactions are essayed thoroughly by understanding space and place. Wherever space and place are defined, (Holdaway et al., 1998) added that archaeologists should analyse artefacts by finding patterns that represent behaviours repeated in common temporal and spatial locus. An average picture of past behaviours is reconstructed from the abandoned artefacts in one location by past individuals (Holdaway et al., 1998). According to Simek, (1984), a conventional approach to intra-site spatial analysis should identify where the pre-historical or historical activity used to happen (place), the spatial distribution of the places, and the tool kit used to perform the actions. Hodder, Jochim and Zimmerman (1978,1976, 1977, as cited in (Shafer, 2016a)) added that “Just as behaviour itself is not random, human beings do not behave randomly in space; that is, they do not use the landscape randomly”. Trigger and Clarke also lauded the importance of defining space; they argued that the past is illuminated by studying the spatial distribution and relative sizes of sites (place) and linking them with relevant evidence, such as political boundaries and regions of influence (Trigger & Clarke, 1977). 2.1.3 Development of Geospatial Technologies in South Africa The use of rudimental geospatial technology for the documentation of fossil deposits in South Africa north of the Vaal River (formerly the Transvaal Province from 1910-1994) can be traced 16 back to the lime-mining activities in the calcitic speleothem caves at Sterkfontein by G Martinaglia in the mid-1890s and David Draper of the Geological Society of South Africa in 1895 (Kuykendall & Štrkalj, 2007). However, the documentation of the fossils in those inception years of archaeology as a discipline lacked accuracy and precision. Most discoveries were from the works of amateur archaeologists, and some of the documented historical fossils had no provenience and no precise spatial context other than a reference to a deposit or member (Kuykendall, 2007; Kuykendall & Štrkalj, 2007). In addition, a significant challenge existed when the fossils were in close contact, and mixing fossil deposits of different ages became inevitable (Armstrong et al., 2018). Furthermore, the apparent disconnection between the provenience of some fossils and museums created misinterpretation of the archaeological record; for instance, in the past, different deposits or samples were grouped as if they came from the same assemblage (Armstrong et al., 2018). An example is in the archaeological documentation of fossils in Sterkfontein, which were referenced as coming from Makapansgat or Taung. However, these prominent localities contain multiple sites spanning millions of years of deposition (Armstrong et al., 2018). Therefore, archaeologists needed to embrace geospatial technologies for detailed recording of all demolished during excavations and exhumations. This is a necessary step in archaeological investigations as it enables future reinterpretation of similar results or for the future archaeologist to experience the site as it was in the first place, arguably making the excavation process virtually reversible (Armstrong et al., 2018; Feder, 2016a; Hester, 2016a). 2.2 Materials and Methods 2.2.1 Overview Systematic reviews are qualitative research methods that follow a predefined protocol such as that provided by Systematic reviews and Meta-Analysis PRISMA statement of 2020, which aim to establish an unbiased and comprehensive summary of available evidence in a particular subject area from existing research studies on some specific topic (Higgins & Green, 2011). Systematic reviews are often conducted to answer specific research questions (Higgins & Green, 2011; Ozturk & Ozen, 2020) and be undertaken in a replicable, scientific, and transparent process with minimum bias from including or excluding studies in the literature review process (Carrera-Rivera, Ochoa-Agurto, et al., 2022). This study aims to reveal the trend in using geospatial technologies in archaeology and cultural heritage from 1990 to 2022 and be insightful in understanding paradigms in archaeological investigations using geospatial technology. Systematic reviews are qualitative analyses often employed with scientometric and bibliometric analyses. The scientometric analysis is a quantitative study of science and scientific research focusing on broader trends across disciplines, research, and the evolution of scientific fields (Bornmann & Marx, 2014). At the same time, bibliometric analysis is a subset of scientometric analysis that mainly focuses on evaluating the performance of researchers, institutions, and journals with specific fields of study (Bornmann & Marx, 2014). This study uses the Scopus search engine to search for scholarly articles containing keywords or terms fundamental to understanding the use of geospatial technology in archaeology and cultural heritage, as listed in Table 2-1. Scopus was launched in 2004 and uses Boolean Syntax to perform bibliometric searches; it combines keywords and operators to yield consolidated and 17 relevant results for a given search criterion (Ozturk & Ozen, 2020). The Scopus database has more information for literature searches than other databases, such as the Web of Science (WOS) (Carrera-Rivera, Larrinaga, et al., 2022; Carrera-Rivera, Ochoa-Agurto, et al., 2022; Ozturk & Ozen, 2020). Table 2- 1: Scopus search engine and queries used for the scope of this study. Search Engine Website Technology Query Scopus scopus.com WORLD GEOSPATIAL (TITLE-ABS-KEY("geospatial technology") OR TITLE-ABS-KEY(GIS) OR TITLE-ABS- KEY("remote sensing") OR TITLE-ABS- KEY(LiDAR) OR TITLE-ABS-KEY("3D scanning") OR TITLE-ABS-KEY("spatial analysis") OR TITLE-ABS-KEY("web mapping applications") OR TITLE-ABS-KEY("augmented reality") OR TITLE-ABS-KEY("virtual reality") OR TITLE-ABS-KEY("geospatial data processing") OR TITLE-ABS-KEY("archaeological predictive modeling") OR TITLE-ABS- KEY("archaeological site mapping") OR TITLE- ABS-KEY("geospatial data integration") OR TITLE-ABS-KEY("geospatial data visualisation") OR TITLE-ABS-KEY("geodetic techniques") OR TITLE-ABS-KEY("satellite remote sensing") OR TITLE-ABS-KEY("geospatial heritage management") OR TITLE-ABS-KEY("GIS-based excavation planning")) AND TITLE-ABS- KEY(archaeology OR "cultural heritage" OR "cultural resource management") AND PUBYEAR > 1989 AND PUBYEAR < 2023 AND PUBYEAR AND ( LIMIT-TO ( DOCTYPE, "ar" ) ) AND ( EXCLUDE ( PREFNAMEAUID, "Undefined" ) ) AND ( LIMIT-TO ( LANGUAGE, "English" ) ) 18 SOUTH AFRICA (TITLE-ABS-KEY("geospatial technology") OR TITLE-ABS-KEY(GIS) OR TITLE-ABS- KEY("remote sensing") OR TITLE-ABS- KEY(LiDAR) OR TITLE-ABS-KEY("3D scanning") OR TITLE-ABS-KEY("spatial analysis") OR TITLE-ABS-KEY("web mapping applications") OR TITLE-ABS-KEY("augmented reality") OR TITLE-ABS-KEY("virtual reality") OR TITLE-ABS-KEY("geospatial data processing") OR TITLE-ABS-KEY("archaeological predictive modeling") OR TITLE-ABS- KEY("archaeological site mapping") OR TITLE- ABS-KEY("geospatial data integration") OR TITLE-ABS-KEY("geospatial data visualization") OR TITLE-ABS-KEY("geodetic techniques") OR TITLE-ABS-KEY("satellite remote sensing") OR TITLE-ABS-KEY("geospatial heritage management") OR TITLE-ABS-KEY("GIS-based excavation planning")) AND TITLE-ABS- KEY(archaeology OR "cultural heritage" OR "cultural resource management") AND PUBYEAR > 1989 AND PUBYEAR < 2023 AND PUBYEAR > 1989 AND PUBYEAR < 2023 AND ( LIMIT-TO ( AFFILCOUNTRY,"South Africa" ) ) AND ( LIMIT-TO ( DOCTYPE,"ar" ) ) AND (