PARENT SESSION
Oral Session #71: Remote Sensing and GIS.
Presiding: J.D. Allan
Wednesday, August 7. 1:00 PM to 4:45 PM. Palo Verde Room, Radisson.

Linking aerial imaging to remote sensing-based mapping of the vegetation of the Udzungwa Mountains of Tanzania.

Muchoney, Douglas^*,1, Juhn, Daniel¹, ¹ Center for Applied Biodiversity Science, Washington, DC

ABSTRACT- Satellite-based mapping of vegetation is incumbent upon spatially explicit examples of vegetation classes that can be used to calibrate (train) and validate (test) classification algorithms. Aerial imaging can be used to supplement existing plot data especially where access is restricted due to access and field costs. This paper reports on the use of aerial imaging to sample the vegetation of the Udzungwa Mountains of Tanzania and to develop a set of site/plot data to map its vegetation. Landsat data, existing aerial photography, a digital elevation model, environmental gradients and expert opinion were used to design aerial transects and survey sites. Using an aerial system that links logs data from an onboard GPS to time-coded video, we acquired and processed aerial digital video 35 mm and digital photography imagery for the Udzungwa Mountains and translated these observations into a set of calibration and validation data. These data were used to calibrate and validate supervised artificial neural network and maximum-likelihood (Bayesian) classification algorithms. The resulting vegetation map is being used to assess the current protected area system for the Udzungwa Mountains and to evaluate expansion of the Udzungwa Mountain National Park. Current efforts involve using the remote sensing data, vegetation map, digital elevation model and plot/site data to generate and test predictive models of species distribution.

KEY WORDS: remote sensing, aerial imaging, vegetation, mapping