Monitoring large areas for forest change using Landsat: Generalization across space, time and Landsat sensors

Landsat 7 ETM+ provides an opportunity to extend the area and frequency wit h which we are able to monitor the Earth's surface with fine spatial resolu tion data. To take advantage of this opportunity it is necessary to move be yond the traditional image-by-image approach to data analysis. A new approa ch to monitoring large areas is to extend the application of a trained imag e classifier to data beyond its original temporal, spatial, and sensor doma ins. A map of forest change in the Cascade Range of Oregon developed with m ethods based on such generalization shows accuracies comparable to a map pr oduced with current state-of-the-art methods. A test of generalization acro ss sensors to monitor forest change in the Rocky Mountains indicates that L andsat 7 ETM+ data can be combined with earlier Landsat 5 TM data without r etraining the classifier. Methods based on generalization require less time and effort than conventional methods and as a result may allow monitoring of larger areas or more frequent monitoring at reduced cost. One key compon ent to achieving this goal is the improved availability and affordability o f Landsat 7 imagery. These results highlight the value of the existing Land sat archive and the importance for continuity in the Landsat Program. (C) 2 001 Elsevier Science Inc. All rights reserved.