1. Abstract

1
1. Abstract The Okavango Delta, located in
northern Botswana, is RAMSAR-status wetland home
to 650 bird species and over 1000 floral species.
The Delta provides critical habitat and resources
to wildlife (including large mammal populations)
and humans. But it faces potentially ecologically
damaging consequences due to natural and
anthropogenic change. Changes in land use such
as extraction of natural resources (water, fish,
wood and reeds), increased burning, over-grazing
of domestic livestock, and a growing tourism
industry increasingly pressure the
wetland-savanna ecosystem. The Delta experiences
two remotely observable disturbance regimes,
flooding and fire. Additionally, oscillations in
precipitation cycles of 3, 8, 18, and 80 years
have been reported for southern Africa. The
impact of those oscillations on flooding (amount
and distribution) across this alluvial fan is
unclear. This research utilizes 85 Landsat TM
and ETM scenes from 1989 through 2002 covering
the southeastern distal portion of the Delta.
Extracted patterns of flooding and fire were
tested against a 2000 Landsat-based vegetation
structure classification created by local
researchers at the University of Botswanas Harry
Oppenheimer Okavango Research Center. Preliminary
results suggest that 1) flooding and fire regimes
manifest very different spatial and temporal
patterns, 2) the co-occurrence of these
disturbances occurs primarily in floodplain
grasses, 3) fire regimes differ between
management regimes (photography versus wildlife
concessions), and 4) climatic trends reported in
the literature are moderately correlated with
Landsat-derived vegetation indices. These early
findings suggest that seasonal, annual, and
longer-term anthropogenic and climatic impacts on
ecologically critical disturbance regimes can be
effectively assessed with seasonally rich optical
time-series data.
Multitemporal Assessment of Vegetation
Disturbance in the Okavango Delta, Botswana
Amy L. Neuenschwander (amy_at_csr.utexas.edu)
Advisor Kelley A. Crews-Meyer Department of
Geography and the Environment University of Texas
at Austin
4. Remotely Sensed Data An extensive set of
field data acquired for the Okavango Delta are
available for this research through the
University of Texas Center for Space Research,
collaboration with the University of Botswana
Harry Oppenheimer Okavango Research Center (HOORC
-located in Maun, Botswana), and other
international partners. This project is utilizing
85 Landsat TM and ETM images beginning in April
1989 and extending through October 2002 with an
average time step of every 2-3 months. Research
Question Can vegetation structure be mapped
using multitemporal, multispectral remote sensing
data? Vegetation structure is defined as the
horizontal and vertical distribution and
organization of the vegetation and for the
purposes of this research it is equivalent to the
ecological definition of vegetation formation.
Vegetation formations (e.g. open woodland or
grass/shrub savanna) are defined as the
hierarchical level below vegetation types (e.g.
grassland or woodland). A common approach to map
land cover change from medium resolution
multispectral data (such as Landsat) is the
from-to change analysis where relative changes
of land cover between two scenes are extracted.
Rather than mapping static land cover change
between two distinct times, this research will
utilize the total temporal signal from the
available Landsat dataset and map temporal
classes, or classes that spectrally behave
similarly throughout the time series. The use of
remotely sensed time series data will capture the
subtle stages of vegetation structure on the
landscape that is driven by flooding and fire as
well as identification of vegetation formations
due differences in phenology. By characterizing
the successional changes in vegetation formation,
it becomes possible to track land cover change
trajectories due to disturbance and gain insight
into these landscape processes. For example, a
change of floodplain vegetation to upland
shrub-dominated vegetation can be an indicator of
an important shift in the functioning of the
system and have subsequent land use impacts
(Ringrose et al., 2005). These drying shifts can
impact the carbon source/sink pools (Hudak and
Wessman, 1998 Asner et al., 2004) as well as
affect the hydrology and soil biogeochemistry
(Asner et al., 2004).
8. Next Steps Multitemporal analysis of remotely
sensed data using wavelets appears to be a
promising methodology for detecting the dominant
trends and periodicities in ecosystem cycles.
This research will continue by examining patterns
in the wavelet power spectrum for all detected
temporal clusters and linking to landcover
trajectories observed in the field. Once
behavior of ecosystem response is identified and
known, it is possible to model future scenarios
of stressors to the Okavango including climate
change and increased anthropogenic pressures.

• References
• Asner, G.P., Elmore, A.J., Olander, L.P., Martin,
  R.E., and Harris, A.T., 2004. Grazing systems,
  ecosystem responses, and global change. Annual
  Review of Environmental Resources, 29261-299.
• Heinl, M., Neuenschwander, A., Silva, J., and
  Vanderpost, C., 2006. Interactions between fire
  and flooding in a southern African floodplain
  system. Landscape Ecology 21(5) 699-709.
• Hudak, A.T. and Wessman, C.A., 1998. Textural
  analysis of historical aerial photography to
  characterize woody encroachment in south African
  savanna. Remote Sensing of Environment 66(3)
  317-330.
• McCarthy, T.S., G.R.J. Cooper, P.D. Tyson, and
  W.N. Ellery, 2000. Seasonal flooding in the
  Okavango Delta, Botswana recent history and
  future prospects. South African Journal of
  Science, 96 25 33.
• Ringrose, S., Jellema, A.,Huntsman-Mapila, P.,
  Baker, L., Brubaker, K., 2005. Use of remotely
  sensed data in the analysis of soil-vegetation
  changes along a drying gradient peripheral to the
  Okavango Delta, Botswana. Int. J. Remote Sensing
  26(19) 4293-4319.
• Torrence, C. and Compo, G.P., 1998. A practical
  guide to wavelets analysis. Bull. Of Amer.
  Meteorological Society 79 61-78.

Acknowledgements This work was supported by the
National Aeronautics and Space Administration
under the Earth System Science Fellowship
Program data provided by the EO-1 program and
SAFARI 2000 program and Technische Universitat
Munchen. Field travel was made possible by the
National Science Foundation under the Doctoral
Dissertation Research Initiative. Many thanks to
Susan Ringrose, Wilma Matheson, Thoralf Meyer,
and Lawrence Drotsky for their field assistance
this summer.