Methodology

 

·       Each band of the imagery was downloaded from the University of Maryland’s website. After converting generic binary formats to image files, the individual bands were then stacked to complete images using the Erdas Modeler. The image p170r60 03/06/2000 contained no projection information, so it was georeferenced in Erdas.

 

·       A watershed AOI was created using a reference image from the World Agroforestry Center. All images were then subset to the Nyando Basin AOI.

 

·       Normalized Difference Vegetation Index (NDVI) was calculated for all images in the Erdas Modeler    using the algorithm: (NIR-IR)/(NIR+IR)

 

for Landsat 5 and 7 images bands 3 and 4 are used:

                                    (b4-b3)/(b4+b3)

 

·       A difference image was then created by subtracting the 2000 NDVI images from the 1986 NDVI images.

 

      

               Difference image p169r60

 

·       Convert NDVI images from DN to 8-bit for use in ArcView

   (n-min)x255/max-min  è Image scaled to 0-255.

 

·       Convert to grid in ArcView

 

·       Identify areas with extreme change (as highlighted in above image)