Malahlela, O. E.Murungweni, F. M.Mavhina, Murendeni Sharon2025-08-212025-08-212025-05-16Mavhina, M.S. 2025. Evaluating the biomass of palatable grass species in the communal rangeland using the integration of UAV data with high-resolution satellite data. . .https://univendspace.univen.ac.za/handle/11602/2899M.Sc. (Ecology and Resource Management)Department of Ecology and Resource ManagementThe biomass of palatable grass species plays a crucial role in sustaining livestock within rural and communal rangelands. Accurate biomass mapping techniques are essential in providing insights into the quantification and distribution of palatable grass biomass. Assessing the distribution of biomass and the effectiveness of biomass mapping is one of the important steps for the provision of crucial information on the health of rangelands while facilitating effective land management and sustainable resource utilisation. This study aimed to evaluate the biomass of palatable grass species in communal rangeland by integrating the use of unmanned aerial vehicle (UAV) obtained in December 2022 data with high-resolution PlanetScope(PS) satellite imagery (3 m spatial resolution) obtained in October 2022. The Supervised Maximum Likelihood Classification (MLC) algorithm was employed in mapping land cover classes across the study and achieved overall accuracy of 99.29% and a kappa of 0.98, indicating high reliability in mapping grazing areas. The grazing area size was quantified as 33 695 hectares (ha) from the total 66 982 hectares (ha) of the study area, with grazing areas accounting for 50%, followed by built-up areas (26%) and water bodies (0.21%). The study also evaluated the performance of linear and non-linear regression models for mapping the Palatable grass biomass by comparing the model performance using remotely sensed data derived from the PS imagery. According to the results, non-linear regression models outperformed the linear models, and obtained a root mean squared error (RMSE), mean absolute error (MAE) and coefficient of determination (R2) of 2.15, 1.69 and 0.11, respectively. Additionally, the study tested the performance of linear and non-linear models for mapping palatable grass biomass using the unmanned aerial vehicle (UAV) and PlanetScope data. The comparison between the PlanetScope subset data and the UAV data obtained an RMSE of 2.36 , MAE of 3.22 and R2 of 0.35 for the subset image, and an RMSE, MAE and R2 of 9.83, 6.57 and 0.23, respectively for the UAV data. Both the satellite imagery and the UAV imagery were used to develop and test models for predicting and mapping palatable grass biomass with the use of integration of spectral bands, vegetation indices, and grey level co-occurrence matrix (GLCM) features. These findings demonstrate possibilities offered by remote sensing techniques in improving rangeland management and sustainable resource planning.1 online resource (vii, 65 leaves): color illustrationsenUniversity of VendaCommunal grazing areasUCTDLinear regressionMaximum Likelihood Classification (MLC)Non-linear regressionPalatable grass biomassPlanetScope imageryUAV image633.2GrassesForage plantsGrasslandsPasturesRangelandSavannasForest biomassPlant biomassDissertationMavhina MS. Evaluating the biomass of palatable grass species in the communal rangeland using the integration of UAV data with high-resolution satellite data. []. , 2025 [cited yyyy month dd]. Available from:Mavhina, M. S. (2025). <i>Evaluating the biomass of palatable grass species in the communal rangeland using the integration of UAV data with high-resolution satellite data</i>. (). . Retrieved fromMavhina, Murendeni Sharon. <i>"Evaluating the biomass of palatable grass species in the communal rangeland using the integration of UAV data with high-resolution satellite data."</i> ., , 2025.TY - Dissertation AU - Mavhina, Murendeni Sharon AB - The biomass of palatable grass species plays a crucial role in sustaining livestock within rural and communal rangelands. Accurate biomass mapping techniques are essential in providing insights into the quantification and distribution of palatable grass biomass. Assessing the distribution of biomass and the effectiveness of biomass mapping is one of the important steps for the provision of crucial information on the health of rangelands while facilitating effective land management and sustainable resource utilisation. This study aimed to evaluate the biomass of palatable grass species in communal rangeland by integrating the use of unmanned aerial vehicle (UAV) obtained in December 2022 data with high-resolution PlanetScope(PS) satellite imagery (3 m spatial resolution) obtained in October 2022. The Supervised Maximum Likelihood Classification (MLC) algorithm was employed in mapping land cover classes across the study and achieved overall accuracy of 99.29% and a kappa of 0.98, indicating high reliability in mapping grazing areas. The grazing area size was quantified as 33 695 hectares (ha) from the total 66 982 hectares (ha) of the study area, with grazing areas accounting for 50%, followed by built-up areas (26%) and water bodies (0.21%). The study also evaluated the performance of linear and non-linear regression models for mapping the Palatable grass biomass by comparing the model performance using remotely sensed data derived from the PS imagery. According to the results, non-linear regression models outperformed the linear models, and obtained a root mean squared error (RMSE), mean absolute error (MAE) and coefficient of determination (R2) of 2.15, 1.69 and 0.11, respectively. Additionally, the study tested the performance of linear and non-linear models for mapping palatable grass biomass using the unmanned aerial vehicle (UAV) and PlanetScope data. The comparison between the PlanetScope subset data and the UAV data obtained an RMSE of 2.36 , MAE of 3.22 and R2 of 0.35 for the subset image, and an RMSE, MAE and R2 of 9.83, 6.57 and 0.23, respectively for the UAV data. Both the satellite imagery and the UAV imagery were used to develop and test models for predicting and mapping palatable grass biomass with the use of integration of spectral bands, vegetation indices, and grey level co-occurrence matrix (GLCM) features. These findings demonstrate possibilities offered by remote sensing techniques in improving rangeland management and sustainable resource planning. DA - 2025-05-16 DB - ResearchSpace DP - Univen KW - Communal grazing areas KW - Linear regression KW - Maximum Likelihood Classification (MLC) KW - Non-linear regression KW - Palatable grass biomass KW - PlanetScope imagery KW - UAV image LK - https://univendspace.univen.ac.za PY - 2025 T1 - Evaluating the biomass of palatable grass species in the communal rangeland using the integration of UAV data with high-resolution satellite data TI - Evaluating the biomass of palatable grass species in the communal rangeland using the integration of UAV data with high-resolution satellite data UR - ER -