Electronic Theses and Dissertations (PhDs)

Permanent URI for this collectionhttps://hdl.handle.net/10539/38008

Browse

Now showing 1 - 2 of 2

Integrating Sentinel-1/2 and machine learning models for mapping fruit tree species in heterogeneous landscapes of Limpopo
(University of the Witwatersrand, Johannesburg, 2024-10) Chabalala, Yingisani Winny; Adam, Elhadi
From ancient times to this century, Africa has relied chiefly on agriculture for survival. Crop type maps are crucial for agricultural management, sustainable farming systems, and realizing food security. Agronomists, agricultural extension officers, policymakers, and the government rely on crop type spatial distribution information to make informed decisions and optimize resource allocation for sustainable agricultural management. Attaining food security for all is an urgent need in Africa. However, the farming landscapes predominately comprise fragmented smallholder heterogeneous farms. The farming systems include intercropping and cultivating different crops that require different management strategies. This results in within-class spectral similarities and intra-spectral variability due to similar canopy structures and different phenologies, which complicates the application of remote sensing in crop type mapping. The free availability of Copernicus products such as Sentinel 1 and 2 have high temporal, spectral, and spatial resolution suitable for mapping smallholder agriculture. Thus, this research aimed to integrate Sentinel-1/2 and machine learning models for mapping fruit tree species in heterogeneous landscapes of Limpopo. First, the research tested the applicability of sampling techniques and five mapping classifiers (i.e., Random Forest (RF), Support vector Machine (SVM), Adaptive Boosting (AdaBoost), Gradient Boosting (GB), and eXtreme Gradient Boosting (XGBoost) in mapping fruit trees and co-existing land use types. The original dataset was under-sampled randomly into two balanced datasets (i.e., Dataset 1 and Dataset 2) consisting of 100 and 150 sample points. Furthermore, the imbalanced ratio from the original dataset was reduced by applying different sampling strategies to extract four imbalanced datasets (i.e., at 40%, 50%, 60%, and 70%), which resulted in the formation of Dataset 3, Dataset 4, and Dataset 5, respectively. These samples, together with the original dataset (i.e., Dataset 7), were used as input to Sentinel‑2 (S2) data using adaptive boosting (AdaBoost), gradient boosting (GB), random forest (RF), support vector machine (SVM), and eXtreme gradient boost (XGBoost) machine learning algorithms. The results showed that reducing the amount of imbalanced ratio by randomly under-sampling the original imbalanced dataset could increase the classification accuracy to 71% using the SVM classifier and 60% of the original dataset. Individually, the majority of the crop types were classified with an F1 score of between 60% and 100%. Secondly, the research independently assessed the effectiveness of Sentinel-1 (S1) and Sentinel-2 (S2) data for fruit tree mapping using random forest (RF) and support vector machine (SVM) classifiers. Four models were tested using each sensor independently and fusing both sensors. From the fused model, features were ranked using the RF mean decrease accuracy (MDA) and forward variable selection (FVS) to identify optimal spectral windows to classify fruit trees. The best fruit tree map with an overall accuracy (OA) of 0.91.6% with a kappa coefficient of 0.91% was produced using the RF MDA and FVS model and SVM classifier. The application of SVM to S1, S2, S2 selected variables and S1S2 fusion independently produced OA = 27.64, Kappa coefficient = 0.13%; OA= 87%, Kappa coefficient = 86.89%; OA = 69.33, Kappa coefficient = 69. %; A = 87.01%, Kappa coefficient = 87%, respectively. The green (B3), SWIR_2 (B10), and vertical horizontal (VH) polarization bands were identified as the optimal spectral features for S2 and S1 data, respectively. The third part of the research identified the optimal growth window period in which fruit trees can be detected with high accuracy. Phenological metrics were extracted from 12 months (i.e., January to December) of Sentinel-2 (S2) data and were used to classify fruit trees using a random forest (RF) classifier in a Google Earth Engine environment. The results showed that fruit trees can be detected and mapped with high accuracy during winter months (i.e., April-July) with an overall accuracy (OA) of 84.89% and a kappa coefficient of 83%. The user accuracy ranged from 62 to 100%, while the producer accuracy ranged from 60 to 100%. The fruit trees were mostly differentiated from co-existing land use types using the short infrared and the red-edge bands. The fourth part of the thesis attempted to increase fruit tree classification accuracy by classifying optimal Sentinel-2 images acquired during the fruit trees' critical growth stages using a Deep Neural Network (DNN) model. This was achieved by applying phenological metrics derived from Sentinel-2 images acquired during optimal crop-growing seasons (i.e., flowering, fruiting, harvesting). The DNN models were optimized by tuning the hyperparameters to achieve the best classification results. The DNN produced an OA of 86.96%, 88.64%, 86.76%, and 87.25% for April, May, June, and July images, respectively. The results indicate the DNN models were robust and stable across the selected fruit growth periods. This research has shown that earth observation (EO) data such as Sentinel 1 and 2 can be used to map fruit trees in fragmented sub-tropical horticultural landscapes characterized by different environmental conditions and different crop cultivars operating under different management practices. The research results will assist agricultural stakeholders (i.e., farm managers, agronomists, agricultural extension officers, and policymakers) in allocating agricultural resources, devising effective agricultural management strategies, and attaining sustainable agriculture and food security.
Remote sensing-based assessment of mangrove forest changes and related regulatory frameworks for the sustainability and conservation of coastal ecosystems in Zanzibar Island, Tanzania-East Africa
(University of the Witwatersrand, Johannesburg, 2024-10) Mohamed, Mohamed Khalfan; Adam, Elhadi
Mangroves are vital components of the world's coastal ecosystems, yet they face significant threats from storm surges, tidal waves, commercial aquaculture, and expanding human settlements. These challenges have heightened the need for accurate mangrove maps to gauge ecosystem degradation. However, mapping mangroves at species and community levels is challenging due to the inaccessibility of these environments. Remote sensing offers an efficient alternative to conventional field-based methods by enabling data collection in these challenging ecosystems. This study aimed to apply remote sensing techniques to map mangrove forest changes and species in two protected bays in Zanzibar, Tanzania. The thesis focuses on four key areas. First, it examines the history of mangrove management in Zanzibar, from colonial times (1890) to the present, highlighting policies, laws, and community involvement in conservation. The colonial authority implemented several land administration laws and regulations to protect mangrove forests. However, mangrove forests suffered significant degradation from 1930 to the end of World War II. The post-independence policy framework established the legal foundation for the introduction of community involvement in mangrove conservation. The legal foundation for introducing community participation in mangrove protection was established by post-independence policy structures such as the National Forest Conservation and Management Act of 1996. Nevertheless, sustainable mangrove use remains inadequate. Second, the study compared community perceptions of mangrove ecosystem services using chi-squared tests and one-way ANOVA. Household surveys showed that provisioning services (PS) were the most identified (84%). Supporting (SS), regulating (RS), and cultural services (CS) were rated by 46.2%, 45.4%, and 21.0%, respectively. Statistical analyses indicated significant differences in the awareness of RS (χ2 = 6.061, p = 0.014) and SS (χ2 = 6.006, p = 0.014) between Chwaka, Charawe, Ukongoroni, Unguja Ukuu, and Uzi wards. There were no significant differences in the identification of PS (χ2 = 1.510, p = 0.919) and CS (χ2 = 1.601, p = 0.901). The study found that residents’ occupations did not determine their reliance on mangrove ecosystem services (χ2 = 8.015; p = 0.1554). Third, changes in mangrove cover in Menai Bay and Chwaka Bay between 1973 and 2020 were analyzed using Landsat data. TerrSet geospatial software was used to classify land cover. The SEGMENTATION module grouped pixels based on spectral similarity, and the images segments were transformed into training sites and signature classes using the SEGTRAIN module. Finally, the segments were classified with the SEGCLASS module into a pixel-based land cover map. Separation of land cover classes was determined using the Jeffries–Matusita (J-M) distance and the transformed divergence (TD) index. For Chwaka Bay, overall classification accuracy ranged from 82.5% to 92.7%, while for Menai Bay, it ranged between 85.5% and 94.5%. Producer and user accuracies ranged from 72% to 100%, with kappa coefficients (κ) between 0.72 and 0.90. Menai Bay experienced a 6.8 ha yearly decline in mangrove cover between 1973 and 2020, while Chwaka Bay saw a 48.5 ha annual decrease. Fourth, the study aimed to map mangrove species in Menai Bay using metrics extracted from the Landsat 9 OLI-2 dataset, i.e., vegetation indices (VIs) and gray-level co-occurrence matrices (GLCMs). A critical step in this study was identifying the contribution of vegetation indices and texture features to classifying mangroves. Training data from very high-resolution (VHR) unmanned aerial vehicle (UAV) data covering parts of the study area helped identify five major mangrove species, i.e., Rhizophora mucronata, Ceriops tagal, Sonneratia alba, Avicennia marina, and Bruguira gymnorrhiza. Results showed that textural features attained overall classification accuracy of 68.29% (kappa = 0.62) and 67.07% (kappa = 0.60) for random forest (RF) and support vector machine (SVM), respectively. Vegetation indices (VIs) recorded overall accuracy of 72.64% (kappa = 0.67) and 67.78% (kappa = 0.61) for RF and SVM. Overall, this study demonstrates the potential of remote sensing technologies for mapping mangrove forest changes and species in challenging environments like Zanzibar’s protected bays. By integrating historical policy analysis with modern geospatial techniques, the research highlights the significant role of both legal frameworks and community involvement in mangrove conservation. The community surveys underscore the varying perceptions of mangrove ecosystem services across different wards, with provisioning services being the most recognized. These findings underscore the importance of advancing remote sensing applications and refining conservation strategies to ensure the sustainability of mangrove ecosystems. Additionally, the analysis of long-term changes in mangrove cover from 1973 to 2020 reveals a concerning decline, particularly in Chwaka Bay. Lastly, the study’s classification of mangrove species using Landsat 9 OLI-2 data, vegetation indices, and texture metrics achieved notable accuracy, emphasizing the value of remote sensing in distinguishing species-level characteristics.