Title: Spatio-temporal distribution of Mimosa pigra and its relations with ecology and hydrology in the Kafue Flats Zambia - A multispectral satellite and Unmanned Aerial Vehicle Approach
Author: Lilian Rudo Nyashanu
Supervising Institution: IHE Delft - Institute for Water Education
The Kafue Flats which is a floodplain system along the Kafue River has had two dams constructed resulting in a shift of the historic seasonal flooding regime. The Flats are recognised Ramsar sites and important areas for migratory birds and the endemic antelope; the Kafue Lechwe. The rapid spread of the invasive species, Mimosa pigra, related to the changes in hydrology of the Kafue River as a result of the hydropower production dams along it has been both an ecological and economic problem for the region. The shrub forms an impenetrable thicket which restricts the growth of native grassland; because it is unpalatable to most wildlife and livestock, reduces suitable habitat for native antelope and increases competition with grazing lands. The study used field data from Unmanned Aerial Vehicles (UAVs) and plant community coverage data to ground truth and develop land cover maps and in particular Mimosa pigra for Itezhi-Tezhi, Namwala and Lochinvar National Park in the Kafue Flats, Zambia. Six Sentinel 2 images from September 2015 and 2017 were classified using the Maximum Likelihood Classification algorithm. Mimosa pigra could not be classified for Itezhi-Tezhi for 2015 and 2017 even though it was identified on the ground with an overall accuracy of 81 % and 80 % respectively. In Namwala the classified map had overall accuracies of 70 and 75 % for 2015 and 2017 respectively: in Lochinvar the overall accuracy of the classified maps was 85% and 84 % for the 2015 and 2017 maps. Mimosa pigra coverage had no significant effect on species richness. Cover of grass species reduced with an increase in Mimosa pigra cover.