In April I presented a paper at the 2008 meeting of the Association of American Geographers in Boston, MA. This annual meeting is an opportunity for geographers in the public, private, and academic sectors to get together to discuss their research (theoretical and applied) and hot topics in Geography. Approximately 8,000 geographers attended this year's meeting!
My presentation titled, "Characterizing Land Use/Land Cover within the Galapagos Islands using an Object Based Image Analysis Approach" was included in a session of other papers on the Galapagos Islands. Despite being located more than 600 miles off the coast of Ecuador, the human population of the Galapagos Islands has been steadily increasing in the last 15 years. Human population growth in the islands has been accompanied by increasing numbers of introduced plants and animals - some of which have become invasive (weeds) - which can alter land cover (what is on the ground) and land use (how the land is used).
The goal of my research was to characterize land use and land cover within an agricultural area of Isabela Island by applying an object-based image analysis (OBIA) approach to high spatial resolution satellite imagery. What does that mean? Well, I basically take satellite images of
Isabela Island (just like you might see in Google Earth) which are divided into a grid of small cells called pixels ("a" in image above). I use specialized software that groups the pixels of the image into "objects" based on rules that I set ("b" in image above). For example, I might group together pixels that look green in the image. Then, I create and apply a set of rules to classify those objects - basically naming each one ("c" in image above). So, those green pixels that are grouped together might be generally called vegetation, and then broken down into sub-groups - like crops when they form a rectangular shape, or mangroves when they form small patches and are close to the coast. The tricky part is deciding on the rules, which should be based on knowledge about the study area and how the landscape is organized.
In this study I primarily focused on identifying a particularly aggressive invasive plant, guava (Psidium guajava) within the agricultural area where it was intentionally introduced and cultivated on local farms. The plant has now spread across the agricultural area and into the Galapagos National Park, and has become difficult to control (like many weeds elsewhere in the world). My results demonstrate that the OBIA approach is generally well-suited to classifying land use and land cover in the Galapagos Islands and for identifying guava - and may be useful for locating other invasive plants in high-resolution satellite imagery.
I plan on continuing work on this topic (among others) as part of my dissertation research over the coming year.
