A research report submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in partial fulfillment of the requirement for the degree of Master of Science (Environmental Sciences) at the School of Geography, Archaeology & Environmental Studies
March 2017 / Accurate estimates and mapping of grass quality is important for effective rangeland management. The purpose of this research was to map different grass species as well as nutrient Phosphorus (P) and Sodium (Na) concentration across grass communities using Sentinel-2 imagery in Telperion game reserve.
The main objectives of the study were to: map the most common grass communities at the Telperion game reserve using Sentinel-2 imagery using artificial neural network (ANN) classifier and to evaluate the use of Sentinel-2 (MSI) in quantifying grass phosphorus and sodium concentration across different grass communities. Grass phosphorus and sodium concentrations were estimated using Random Forest (RF) regression algorithm, normalized difference vegetation index (NDVI) and the simple ratios (SR) which were calculated from all two possible band combination of Sentinel-2 data.
Results obtained demonstrated woody vegetation as the dominant vegetation and Aristida congesta as the most common grass species. The overall classification accuracy = 81%; kappa =0.78 and error rate=0.18 was achieved using the ANN classifier. Regression model for leaf phosphorus concentration prediction both NDVI and SR data sets yielded similar results (R2 =0.363; RMSE=0.017%) and (R2 =0.36 2; RMSE=0.0174%). Regression model for leaf sodium using NDVI and SR data sets yielded dissimilar results (R2 =0.23; RMSE=16.74 mg/kg) and (R2 =0.15; RMSE =34.08 mg/kg). The overall outcomes of this study demonstrate the capability of Sentinel 2 imagery in mapping vegetation quality (phosphorus and sodium) and quantity.
The study recommends the mapping of grass communities and both phosphorus and sodium concentrations across different seasons to fully understand the distribution of different species across the game reserve as well as variations in foliar concentration of the elements. Such information will guide the reserve managers on resource use and conservation strategies to implement within the reserve. Furthermore, the information will enable conservation managers to understand wildlife distribution and feeding patterns. This will allow integration of effective conservation strategies into decisions on stocking capacity. / MT 2017
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:wits/oai:wiredspace.wits.ac.za:10539/23495 |
| Date | January 2017 |
| Creators | Mashamba, Tendani |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Format | Online resource (viii, 58 leaves), application/pdf |
Page generated in 0.0015 seconds