WIReDSpace

Welcome to WIReDSpace (Wits Institutional Repository on DSpace)

For queries relating to content and technical issues, please contact IR specialists via this email address : openscholarship.library@wits.ac.za,
Tel: 011 717 4652 or 011 717 1954

Photo by @inspiredimages

Communities in WIReDSpace

Select a community to browse its collections.

Now showing 1 - 5 of 19

Recent Submissions

  • Thumbnail Image
    Item type:Item,
    Estimating biophysical variables of the winter wheat crop using multispectral remotely sensed data
    (University of the Witwatersrand, Johannesburg, 2025) Zikhali, Siyabonga; Adam, Elhadi
    Winter wheat is a vital crop for ensuring global food security. For example, wheat accounts for about 21% of the world's food production. However, its cultivation faces threats from environmental stress. Furthermore, significant climate changes, coupled with the necessity for sustainable agriculture and the global imperative to address food security issues, increase the urgency of adopting sustainable agricultural practices. This can be achieved by monitoring and estimating crop biophysical parameters such as leaf area index (LAI) and chlorophyll a and b (Cab), which offer valuable insights into crop health and development, helping farmers make better-informed decisions. Nonetheless, traditional and recent UAV-based crop monitoring methods have limitations, including the requirement for field presence and poor temporal resolution. Therefore, this study aimed to estimate biophysical parameters of winter wheat (i.e., LAI and Cab) using multispectral Sentinel-2 imagery. The support vector regression (SVR) and random forest (RF) algorithms were utilised in conjunction with Sentinel-2 imagery to estimate the winter wheat LAI and Cab, and both models performed sufficiently well for most parts. However, the RF model performed better than SVR. RF LAI models exhibited an average predictive R2 of 0.87, with RMSE of 0.11 and MAE of 0.09, while the SVR LAI model exhibited an average predictive R2 of 0.63, with RMSE of 0.18 and MAE of 0.13. Similarly, the RF Cab model had an average predictive R2 of 0.89, with RMSE of 1.24 and MAE of 0.95; while the SVR Cab model had an average predictive R2 of 0.57, with RMSE of 1.96 and MAE of 1.29. As a result, RF was used to create estimation maps for LAI and Cab across different growth stages. These results indicated that open-access data from Sentinel-2 and machine learning algorithms, such as RF, provide economical means in terms of both time, computer, and financial costs to produce realistic estimations of crop variables such as LAI and Cab, even with minimal resources available
  • Thumbnail Image
    Item type:Item,
    A radiogenic and stable isotope investigation into the petrogenesis of Proterozoic massif-type anorthosites from the Kunene AMCG Complex in Angola”
    (University of the Witwatersrand, Johannesburg, 2025) Ntuli, Simphiwe Polar
    The petrogenesis of Proterozoic massif-type anorthosites is a topic of controversy, particularly regarding the origin (mantle vs. crust) and modification (e.g., fractional crystallisation and crustal contamination) of the parent magmas that form these rocks. The Mesoproterozoic (1500–1375 Ma) Kunene Complex (KC) in Angola and Namibia is the largest (~42,500 km2) example of Proterozoic massif-type anorthosites globally. The KC anorthosite suite has been divided into six geochronologically and structurally distinct domains, each one emplaced into Paleoproterozoic crust consisting of Regional Granite and the Namibe Zone in Angola and the Epupa Zone in Namibia. This study reports high-precision bulk-rock radiogenic (Sr-Nd- Hf) and internal (plagioclase and pyroxene) stable (O) isotopes in conjunction with petrography and bulk- rock major and trace element data to constrain the petrogenesis of the main KC anorthosite suite in Angola. The anorthosite suite sensu lato across all domains is enriched in Ba, Pb, Sr, and K and depleted in Rb, Nb, Ta and Ti relative to the primitive mantle. Rock-types across all domains show positive Eu anomalies (Eu/Eu*) and fractionated trace element patterns (i.e., enrichment in light/heavy rare earth elements). There is variation in Sr-Nd-Hf-O isotopic compositions across all domains and lithologies. All radiogenic Sr-Nd- Hf isotopes were corrected to a mean age of 1418 Ma for the KC. Domains 1 (ca. 1380 Ma) and 2 (ca.1412– 1400 Ma), situated in the north of the KC, show slightly radiogenic 87Sr/86Sr1418Ma and unradiogenic εNd1418Ma. The olivine-dominant anorthosites of Domain 1 show 87Sr/86Sr1418Ma between 0.704222– 0.704639 and εNd1418Ma of -1.85 to -5.23. Two olivine-leucogabbros from this domain show δ18O of 6.60 and 6.72‰ as well as radiogenic εHf1418Ma values of +5.24 and +5.72. Domain 2 shows 87Sr/86Sr1418Ma between 0.704323–0704918 and εNd1418Ma values of -1.33 to -13.23 and δ18O 4.56 and 7.27‰ in magnetite- bearing anorthosites and a very radiogenic εHf1418Ma value of +11.76 for one leucotroctolite. Domain 3 (ca. 1391 Ma), emplaced south of Domain 2, is characterised by 87Sr/86Sr1418Ma between 0.703574 to 0.704345 and εNd1418Ma values of -6.67 to +6.25, δ18O values of 6.26 to 8.39‰ and an unradiogenic εHf1418Ma value of -7.76 for one leucogabbronorite. Domains 4 (ca. 1390 Ma) and 5 (c. 1438 Ma), situated in the southwest and east of the complex, respectively, show relatively juvenile 87Sr/86Sr1418Ma and εNd1418Ma isotopic compositions. Domain 4 shows 87Sr/86Sr1418Ma between 0.703813–0.703944 and εNd1418Ma values of +0.015 to +6.72 and δ18O values of 6.46 to 7.12‰. A gabbroic anorthosite from Domain 5 shows 87Sr/86Sr1418Ma of 0.703591 and an εNd1418Ma value of -1.75, while two plagioclase separates show δ18O values of 4.88 and 6.25‰. Olivine-dominant anorthosites, regardless of domain, generally show juvenile Sr-Nd isotopic compositions along with high Mg# and An# of olivine and plagioclase compared to anorthosites that are pyroxene- dominant. This relationship may simply be controlled by the amount of crustal contamination, as is proposed for many massif-type anorthosite complexes globally. Binary mixing models were constructed to iii assess suitable crustal contaminants and the amount of contamination required to modify the parent magmas to the KC anorthosite suite, assuming the parent magmas were sourced from a depleted mantle source. These models show that the parent magmas were variably contaminated (up to 10%) by their adjacent crustal domains during emplacement, with the Regional Granite and Epupa Zone likely contaminants. Some anorthosite samples with strongly unradiogenic Nd with moderately enriched Sr isotopic compositions require contamination with a component with long-term enrichment in light rare earth elements. This can be modelled with an Archean crustal component, suggesting some interaction with Archean crust at depth beneath the KC was important in anorthosite petrogenesis. Paleoproterozoic model ages (TDM) for the Nd isotope compositions of the KC anorthosite also support the binary mixing models. Therefore, the Sr–Nd- Hf–O isotopic data for the KC anorthosite suite in Angola are consistent with crustal contamination of mantle-derived parent magmas in a continental arc-like setting.
  • Thumbnail Image
    Item type:Item,
    A simulation study to explore the dangers of ignoring competing
    (University of the Witwatersrand, Johannesburg, 2026) Nephawe, Mafanedza; Nasejje, Justine
    Abstract This study investigates the consequences of incorrectly handling competing risks in survival analysis, particularly applied to simulated customer churn and multimorbidity data. Traditional survival analysis methods often neglect competing risks, potentially leading to biased estimations and inaccurate conclusions. This research aims to evalu- ate the impact of neglecting competing risks using comprehensive simulation studies, addressing a significant gap in practical methodological validation. To achieve this, synthetic datasets representing customer churn scenarios and multi- morbidity contexts were generated, encompassing varying sample sizes of 1,000, 2,000, and 3,000.The Churn data was simulated to be linear and the multimorbidity data was simulated to be non-linear. Four competing risks survival models Cause Specific Cox, Random Survival Forest (RSF), DeepHit, and the Fine-Gray model were employed and rigorously compared. Each model was implemented to evaluate discriminative per- formance, calibration accuracy, and predictive consistency using key metrics: C-index, Brier score, and ROC AUC, with five-fold cross-validation applied to ensure robustness. The results indicated significant variations in model performance and clearly demon- strated the implications of ignoring competing risks. In the customer churn scenar- ios, the Cox proportional hazards model consistently exhibited superior discriminative performance (C-index and ROC AUC values generally above 0.8) and superior calibra- tion (Brier scores around 0.1) across all dataset sizes (N = 1000, 2000, 3000). This can be attributed to the churn data being simulated to be linear and the presence of competing risk being negligible. Random Survival Forest (RSF) and DeepHit models performed comparably well, though without substantial gains from increased sample sizes, whereas the Fine-Gray model consistently showed the weakest performance, with lower discriminative abilities (C-index and ROC AUC around 0.6) and poorer calibra- tion (Brier scores around 0.2). In multimorbidity scenarios, RSF and DeepHit emerged as the best performing models, achieving high discrimination (C-index and ROC AUC around or exceeding 0.8) and excellent calibration (Brier scores close to 0.1), particularly with larger dataset sizes. Traditional methods (Cox and Fine-Gray) displayed moderate to low performance, par- ticularly for Disease B (C-index around 0.4-0.6 and higher Brier scores around 0.2-0.3), indicating limited effectiveness in complex health scenarios when compared to machine learning methods. The study underscores the substantial risks of neglecting competing events in survival analysis, clearly showing that failure to properly account for competing risks can inflate event probabilities and distort predictive accuracy. This research reaffirms the need to carefully select analytical approaches aligned with the specific aims and available data in real-world applications. In conclusion, accurately modeling competing risks is crucial across domains such as business analytics and clinical research. This work provides empirical evidence to guide researchers and practitioners in adopting appro- priate competing-risk methodologies, ultimately promoting more precise, reliable, and domain-specific survival analyses.
  • Thumbnail Image
    Item type:Item,
    Graph duals and operators
    (University of the Witwatersrand, Johannesburg, 2025-06) Khampepe, Ramasedi; Mphako-Banda, Eunice Gogo; Kriel, Christo
    This dissertation addresses several open questions concerning the iteration of the mincut graph operator. Central to this study are fundamental questions about periodicity and convergence under the mincut operator. We propose a construction of a class of graphs and provide general properties of these graphs in terms of its order and edge-connectivity. We then prove that it is 2 − periodic under the X(·) operator. These findings add on to the exploration of periodicity, convergence of the X(·) operator. We further explore the interplay between cycles, minimal edge cuts, and graph operators, with a specific focus on planar graphs. The dual relationship between cycles and minimal edge cuts in planar graphs serves as a guiding framework for our analysis, enabling us to investigate the iterative behavior of the mincut operator in greater depth.
  • Thumbnail Image
    Item type:Item,
    A field, petrographic and geochemical investigation of norite and anorthosite rocks related to the potholed Merensky Reef sub-facies at Northam Platinum Zondereinde mine, Western Bushveld Complex
    (University of the Witwatersrand, Johannesburg, 2025-10) Keikabile, Zanele; Hayes, Ben
    This research study is an investigation of norite and anorthosite bodies identified within potholed Merensky Reef in the southwestern extension of Northam Platinum’s Zondereinde Mine, situated in the Western Limb of the Bushveld Complex, South Africa. This study aims to understand the nature and impact of these bodies on the reef as well as constrain their effect on PGE grade and mining. Data from underground geological mapping and borehole logging is combined with petrography and bulk rock trace element geochemistry. Comparisons are also made with similar norites in the hanging wall and footwall of the Merensky Reef. Underground mapping and core logging demonstrate the unpredictable spatial distribution of these bodies, with both sharp and gradational contacts being observed. Potholing and slumping of the reef also appear to be closely related to the norite-anorthosite bodies. Microstructural observations from thin sections indicate there are textural similarities between the norite bodies and the hanging wall norite to the Merensky Reef, which is also supported by the bulk rock trace element compositions. These similarities between the norite bodies and the layered stratigraphy above the Merensky Reef suggest they are connected (i.e., cogenetic) to the layered stratigraphy and they are not of an intrusive or exotic origin. The possible relationship between the norite-anorthosite bodies and Iron-Rich Ultramafic Pegmatoids is also explored, as the latter commonly replace the norite and anorthosite bodies at Zondereinde Mine. It is also shown that the norite-anorthosite bodies are barren in terms of PGE grade, and they therefore significantly dilute the PGE grades of the mined Merensky Reef, such that they pose substantial challenges to mining operations. A recommendation from this study is that there should be increased diamond drilling to better delineate the spatial distribution of these bodies. This will inform mining strategies, increase efficiency, and minimise geological losses in the southwestern extension of Zondereinde Mine.