Mapping land-cover change in a haitian watershed using a combined spectral mixture analysis and classification tree procedure
Severe deforestation in the Caribbean nation of Haiti is a long-standing concern in Haiti and internationally. There are, however, few studies measuring the amount, type, rate or location of this deforestation and related land-cover changes. This study measures the loss of pine forest over three decades from one watershed in Haiti. The study employs an image processing method that draws on the strengths of spectral mixture and classification tree analyses. Results show 54% of the watershed was forested in 1979 compared with 22% in 2000. For the 2000 map, overall accuracies range from 81 to 91% and user's mean per-class accuracies range from 71 to 90%. Overall map accuracies range from 73 to 83% for the 1979 land-cover map with user's mean per-class accuracies ranging from 71 to 84%. For 2000, the combined classification procedure yields more accurate results than a classification tree alone. © 2010 Taylor & Francis.
Versluis, Anna and Rogan, John, "Mapping land-cover change in a haitian watershed using a combined spectral mixture analysis and classification tree procedure" (2010). Geography. 671.