The Okavango Delta, an internationally recognized wetland, is undergoing natural and anthropogenic change at a variety of spatio-temporal scales. The objective of this research was to utilize remotely sensed imagery to assess the spatio-temporal distribution of flooding and fire and their subsequent influences on vegetation as represented by vegetation index trajectories in the Okavango Delta. The characterization of the spatiotemporal dynamics of vegetation spectral response via a time-series of remotely sensed data not only informs ecosystem and disturbance theory but also presents new methodological applications for multi-temporal change analysis. Disentangling these components from a signal is critical for better assessing the interrelationships among climatic oscillations, disturbance regimes, and human management on ecosystem response. This research tested six hypotheses regarding flooding and fire, and found that the largest number of fires occurred either within 5 km of the border to the Wildlife Management Areas or within the active (flooded a minimum of every two years) floodplains. These hypotheses indicate that burning is highest where people have accessinto the management areas and where the natural resources are plentiful. Periodicities from vegetation signal time-series did not confirm published climate-driven periodicities of 3, 8, and 18-years but did reveal seasonal (6 month) and quasi-decadal periodicities. Vegetation trajectories were more predictable with increasing flood frequency and duration, but were less predictable with increased fire frequency. The fact that increased burning resulted in less predictable behavior indicates the potential of quantifying the anthropogenic influence on the landscape using remotely sensed imagery. Flooding and fire were not statistically correlated to the residual dynamics, refuting the conceptualization of flooding and fire as disturbance and supporting the interpretation of flooding and fire as disturbance regimes. This research thus contributes methodologically and theoretically to the ecology literature by operationalizing tests for disturbance versus disturbance regimes via spatio-temporal characterization. Further, this work extends change detection techniques typically implemented with coarser spatial resolution but more frequently acquired imagery by using harmonic regression and wavelet analysis with Landsat data. Lastly, this work provides a temporally rich assessment of recent vegetation, flooding, and fire trends for improving management efforts of the Okavango Delta.

Thesis (M.Sc.)--University of the Witwatersrand, Faculty of Science, 1993.