A research report submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in partial fulfilment of the requirements for the degree of Master of Science (Geographical Information Systems and Remote Sensing) at the School of Geography, Archaeology & Environmental Studies) Johannesburg. 2017 / Vegetation mapping using remote sensing is a key concern in environmental application using remote sensing. The new high resolution generation has made possible, the mapping of spatial distribution of vegetation communities. The aim of this research is to test the use of new generation multispectral data for vegetation classification in Ezemvelo Game Reserve, Bronkhorspruit. Sentinel-2 and RapidEye images were used covering the study area with nine vegetation classes: eight from grassland (Mixed grassland, Wetland grass, Aristida congesta, Cynadon dactylon, Eragrostis gummiflua, Eragrostis Chloromelas, Hyparrhenia hirta, Serephium plumosum) and one from woodland (Woody vegetation). The images were pre-processed, geo-referenced and classified in order to map detailed vegetation classes of the study area. Random Forest and Support Vector Machines supervised classification methods were applied to both images to identify nine vegetation classes. The softwares used for this study were ENVI, EnMAP, ArcGIS and R statistical packages (R Development Core, 2012) .These were used for Support Vector Machines and Random Forest parameters optimization. Error matrix was created using the same reference points for Sentinel-2 and RapidEye classification. After classification, results were compared to find the best approach to create a current map for vegetation communities. Sentinel-2 achieved higher accuracies using RF with overall accuracy of 86% and Kappa value of 0.84. Sentinel-2 also achieved overall accuracy of 85% with a Kappa value of 0.83 using SVM. RapidEye achieved lower accuracies using RF with an overall accuracy of 82% and Kappa value of 0.79. RapidEye using SVM produced overall accuracy of 81% and a Kappa value of 0.79. The study concludes that Sentinel-2 multispectral data and RF have the potential to map vegetation communities. The higher accuracies achieved in the study can assist management and decision makers on assessing the current vegetation status and for future references on Ezemvelo Game Reserve. Keywords Random forest, Support Vector Machines, Sentinel-2, RapidEye, remote sensing, multispectral, hyperspectral and vegetation communities / LG2018

Expansion of woody vegetation results in the transformation of a grass-dominated ecosystem to a tree-dominated ecosystem causing land degradation in most semi-arid areas. The imbalance in the natural ecosystem between herbaceous plants and woody vegetation poses a threat to the natural environment. Such changes alter the flow, availability and quality of nutrient resources in the biogeochemical cycle. Most of the dominating woody plants are often unpalatable to domestic livestock. Therefore, the objective is to assess the spatial extent of woody vegetation over time. Knowledge of the spatial and temporal characteristics of woody vegetation dynamics will enable the development of management plans. These characteristics can be derived using remote sensing techniques which have become efficient in such studies. This study aimed to characterize woody vegetation dynamics along the route between Alice and King Williams’s town in Eastern Cape Province South Africa using Landsat data. This aim was achieved by focussing on three specific objectives. The first objective was to compare the performance of multispectral data and Normalized Difference Vegetation Index (NDVI) data of Landsat imagery in mapping woody vegetation cover. The second objective was to investigate the effect of the spatial resolution of remotely-sensed data on discrimination of woody vegetation from other land cover types. The third objective characterised woody vegetation dynamics between 1986 and 2013/2014 using the results from the first objective. The study used Landsat imagery acquired in November or February of 1986, 1994/1995, 2002/2003 and 2013/2014. Due to lack of data which covered the study area two separate dates (November and February) where used for the study resulting in naming the study area western and eastern parts. Unsupervised classification was performed on the multispectral, NDVI and pan-sharpened images to generate four generic land cover classes, namely water, bare land, grassland and woodland. Accuracy assessments of the classified images was done using error matrix. The results showed that the classification based on NDVI images yielded a better overall accuracy than the classification based on multispectral images for the western (83 percent and 75 percent, respectively) and eastern (82 percent and 76 percent, respectively) parts of the study area. Similarly, pan-sharpening resulted in better overall classification accuracy than multispectral, but comparable to the classification of the NDVI images for both the western (82 percent) and eastern (83 percent) parts of the study area. Remote sensing is an effective tool in assessing changes in the physical environment. Landsat imagery is suitable in assessing land cover dynamics given the long-term and free availability of the image. In addition, the large spatial coverage it provides, enables Landsat data to be used on studies that have wide spatial coverage. Classification for the purpose of time-series analysis was then performed on the NDVI images of each date (1986, 1994/1995, 2002/2003 and 2013/2014). Both woody vegetation and grassland experienced changes from 1986 to 2013/2014 with grassland occupying (75 percent) compared to woodland (17 percent) in 1986. In the year 2013/14 grassland occupied 32 percent and woodland occupied 51 percent of the study area. The increase in woody vegetation in the study area can be attributed to livestock rearing and migration of people from the rural to urban areas post-Apartheid. The study output will aid in the development of a database on land cover distribution of the area between King William’s town and Alice town, providing useful information to decision-making and further studies on woody vegetation.

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. Johannesburg 1992. / The aim of the study was to develop understanding of the relationships between vegetation types of the grassland biome of South Africa and the environment, with an emphasis on structural and functional characteristics. The grassland biome in South Africa has traditionally been divided into 'pure' grasslands, assumed to be climatically determined, and 'false' grasslands of recent anthropogenic origin. A review of literature from several disciplines including palaeobotany, archaeology, ecology and biogeography indicates that this is not a valid distinction. It is clear that the distribution of the grassland biome as a whole is poorly understood, but the general correlation between the distribution of biomes and climate elsewhere in the world suggests that this warrants more detailed investigation. A water balance approach was used to develop climatic incices that both predict the distribution of grasslands, and are easy to interpret biologically. The indices are the mean. number of days per annum when moisture is available for plant growth, tbe mean temperature on days when moisture is available for plant growth (wet season temperature),. and the mean temperature when moisture is not available for plant growth (dry season temperature). Based on these three.indices the grassland biome in South Africa call be distinguished from neighbouring biomes. The fynbos and succulent karoo biomes have rainfall in winter. The grassland, nama-karoo and savanna biomes have' rainfall in summer. The forest biome experiences rainfall throughout the year. Of the summer rainfall biomes, the quantity of water available in the grassland biome b greater than in the nama-woo, similar to savanna, but less than forest. Grasslands experience cooler dry season temperatures than savannas. The localised distribution of woody plants within the. grassland biome suggests that it is the effect of climate on the fire regime that may be of overriding importance h'l determining the distribution of the biome as a whole. Woody elements are restricted to sites that are either protected from fire, or experience fires of lower intensity than sites that support- grassland, The unifying feature of the grassland biome is its proneness to fire. The presence of a warm, moist season promotes plant production and leads to a high standing crop close to the ground. The prolonged dry season causes vegetation to dry out annually, rendering it flammable. More arid biomes have plants more widely spaced, making it difficult for fire to spread. In more mesic biomes where rainfall is less sea.sonal than in the grasslands or savannas, fuels do not dry out sufficiently to ignite, A number of additional climatic features may promote burning in the grassland biome, It has the highest lightning density of all South Africa's biomes. 'tVarm, dry 'berg' winds desiccate fuels and 1 omote burning in the more mesic grasslands, The 'curing' of the grass sward due to dry season frost and temperature drop is important in establishing early dry season flammability. Savanna trees are fire tolerant, but they appear sensitive to the cold temperatures prevaient in the grassland biome in. the dry season, The relationship between the distribution of functional characters of grassland plants and environmental conditions was investigated. The distincrion between sweetveld, mixed veld and sourveld was recognised as one of the most Important functional features of South Africa's grasslands, The distribution of these vegetation types was examined in detail. Sweetveld occurs In warm, dry areas; sourveld in cool, moist areas. There Is overlap between these tyP.Js that Is dependant on soil nutrient status. Sweetveld that occurs in climatic conditions that would be expected to support mixed veld and sourveld, is on soils derived from basic parent material, including basalt, dolerite, gabbro and norite. Similarly, sourveld that occurs in areas that climatically would be expected to support sweetveld, is on soils derived from acid parent material such as sandstone and quartzite .. Soil nutrients that are most highly correlated to the occurrence of these three veld types are phosphoms availability and an index of nitrogen mineralization potential. 'l'here is an increase in bot; available phosphorus and the index of readily mineralizable nitrogen from sourveld to mixed veld to sweetveld. These features am inc01).10111tedinto a conceptual model that relates the distribution of these grassland types to carbon and nitrogen metabolism, with the role of phosphorus either similar to nitrogen, or else it may act indirectly by affecting the. rate of nitrogen mineralization, Nitrogen mineralization OCcursat lower water availability than carbon assimilation, and its temperature optimum is higher than that of carbon assimilation. Where nitrogen mineralization is favoured ielative to carbon assimilation, sweetveld is likely to (}C(.1\Xr. Where carbon assimilation is. favoured relative to; nitrogen mineralization, sourveld is likely to occur ....Soil texture affects the balance between these two processes in the degree to wm.r;h it protects soil organic matter, and thereforv the size of the nitrogen and ph_QSPllO_rOll.S pools. Changes in the rlj,stribution of South Africa's b~\omesfor a scenario of climate change are predicted using the biome model developed in this study. This illustrates the value of developing predictive models. / MT2017

As part of a vegetation survey program for the newly acquired farms incorporated into the Loskop Dam Nature Reserve, the vegetation of the Hondekraal Section was investigated. The study provides an ecological basis for establishing an efficient wildlife management plan for the Reserve. From a TWINSPAN classification, refined by Braun-Blanquet procedures, 12 plant communities, which can be grouped into eight major plant communities, were identified. A classification and description of the major plant communities are presented as well as a management plan. Descriptions of the plant communities include characteristic species as well as prominent and less conspicuous species of the tree, shrub, herb and grass strata. This study proves that the extended land incorporated into the Reserve contributes to the biological diversity of the Reserve. / Environmental Sciences / M. Tech. (Nature Conservation)

As part of a vegetation survey program for the newly acquired farms incorporated into the Loskop Dam Nature Reserve, the vegetation of the Hondekraal Section was investigated. The study provides an ecological basis for establishing an efficient wildlife management plan for the Reserve. From a TWINSPAN classification, refined by Braun-Blanquet procedures, 12 plant communities, which can be grouped into eight major plant communities, were identified. A classification and description of the major plant communities are presented as well as a management plan. Descriptions of the plant communities include characteristic species as well as prominent and less conspicuous species of the tree, shrub, herb and grass strata. This study proves that the extended land incorporated into the Reserve contributes to the biological diversity of the Reserve. / Environmental Sciences / M. Tech. (Nature Conservation)

The vegetation of the Nooitgedacht section of the Loskop Dam Nature Reserve resembles Bankenveld vegetation and differs from the other areas of the reserve. This study was undertaken to identify, classify, and describe the plant communities present on this section, and to determine their veld condition. The Braun-Blanquet approach was followed to classify the different plant communities. A total number of 170 sample plots (100m2) were placed in all homogeneous vegetation units in a randomly stratified basis. The Ecological Index Method (EIM) was used to determine the veld condition. Data were collected using the steppoint method and incorporated into the GRAZE model from where the veld condition was calculated. A minimum of 400 step points were surveyed in each community with more points in the larger communities. Plant community data was analysed using the JUICE software program. A total of 11 plant communities were identified. The overall veld condition score indicates the vegetation to be in a good condition, resulting in a high grazing capacity. / Environmental Sciences / M. Sc. (Nature Conservation)

The vegetation of the Nooitgedacht section of the Loskop Dam Nature Reserve resembles Bankenveld vegetation and differs from the other areas of the reserve. This study was undertaken to identify, classify, and describe the plant communities present on this section, and to determine their veld condition. The Braun-Blanquet approach was followed to classify the different plant communities. A total number of 170 sample plots (100m2) were placed in all homogeneous vegetation units in a randomly stratified basis. The Ecological Index Method (EIM) was used to determine the veld condition. Data were collected using the steppoint method and incorporated into the GRAZE model from where the veld condition was calculated. A minimum of 400 step points were surveyed in each community with more points in the larger communities. Plant community data was analysed using the JUICE software program. A total of 11 plant communities were identified. The overall veld condition score indicates the vegetation to be in a good condition, resulting in a high grazing capacity. / Environmental Sciences / M. Sc. (Nature Conservation)