Ruiliang Pu home page

 

Project3


Using digital elevation model (DEM) data, its derivatives, and forest cover data, we constructed neural networks to classify 27 land ecosystem classes at Duck Mountain,  Manitoba, Canada in 1995.  Training and testing of those neural networks were done using an existing land-systems map prepared through airphoto interpretation and field studies.  The method can increase classification accuracy by 10-30% with a limited database compared to traditional methods.  Land ecosystems classification provides data set used to represent the spatial patterns of landscape physical properties.  With these land systems classification results people can exploit land resources more effectively.

 


Please go back my main homepage