Mapping Sahelian Floodplain Vegetation from Satellite Imagery
Loading...
Date
2016-03-30
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The intimate connection of vegetation growth and flood dynamics is an integral component of the Logone floodplain in the Far North Region of Cameroon. Vegetation temporal and spatial dynamics are important to the migration patterns of pastoralists as they move into the region to graze cattle. This project’s aim was to create vegetation maps of the dominant perennial grass species (Oryza longistaminata and Echinochloa pyramidalis) to plot their spatial distribution and area within the floodplain. ‘One-class’ classification of a October 2014 Landsat satellite image was performed using ground-truthed vegetation transect data from the same period (collected by Dr. Paul Scholte, DGIZ) with 39 sites composed of the focus vegetation species and 8 sites of other species. Visibly identifiable features of river, rice fields, settlements, and areas of no vegetation were also included and contributed a further 111 training sites. ‘Leave-one-out cross’ validation was performed to analyze model error. The model was applied to classify the entire floodplain, and results indicate the dominant perennial species covered 2650.28km2. The model may be applied to other Landsat images, but only those captured during periods of similar flooding as dynamics of vegetation growth introduce large variation in spectral signature in images outside of this timeframe. Collecting more sites specifically for training may improve the model. Classified maps can illustrate the extent of the perennial grass species and inter-annual vegetation dynamics of the floodplain, which can be used along to investigate the relationship of vegetation within the region’s coupled human and natural system.
Description
Agriculture/Ecological/Environmental Science (The Ohio State University Denman Undergraduate Research Forum)
Keywords
Remote Sensing, Vegetation, Classification, Satellite Imagery