Land cover classification accuracy as a function of sensor spatial resolutionThe benefits obtained from sensor systems for monitoring earth resources will depend on the application and interpretation methods used. A frequently used analysis method is supervised per-pixel multispectral classification with a typical application being land cover classification. An investigation is conducted to evaluate the effect of spatial resolution on the ability to classify land cover types with per-pixel digital image classification techniques. Attention is also given to the documentation of changes in scene noise and the percentage of boundary pixels as a function of spatial resolution, in order to improve the understanding of the interrelationship between classification accuracy and spatial resolution. It is found that scene noise varies considerably between land cover categories. Changes in scene noise with coarsening resolution occur at different rates for different categories.
Document ID
19820044130
Acquisition Source
Legacy CDMS
Document Type
Conference Proceedings
Authors
Markham, B. L. (NASA Goddard Space Flight Center Earth Survey Applications Div., Greenbelt, MD, United States)
Townshend, J. R. G. (NASA Goddard Space Flight Center Earth Survey Applications Div., Greenbelt, MD; Reading, University, Reading, Berks., United Kingdom)
Date Acquired
August 10, 2013
Publication Date
January 1, 1981
Subject Category
Earth Resources And Remote Sensing
Meeting Information
Meeting: In: International Symposium on Remote Sensing of Environment