Estuarine Landcover Along the Lower Columbia River Estuary Determined from Compact Ariborne Spectrographic Imager (CASI) Imagery, Technical Report 2003.

Abstract

This report was funded by the Bonneville Power Administration (BPA), U.S. Department of Energy, as part of BPA's program to protect, mitigate, and enhance fish and wildlife affected by the development and operation of hydroelectric facilities on the Columbia River and its tributaries. The views in this report are the author's and do not necessarily represent the views of BPA.

Change in Land Cover along the Lower Columbia River Estuary as Determined from Landsat Thematic Mapper (TM) Imagery, Technical Report 2003.

“…The C-CAP protocol was used to interpret land cover classes from one Landsat TM scene for creation of the TM1992 data set (NOAA Coastal Services Center, 1997). The TM2000 data set was created by Earth Design Consultants, Inc. (Garono et al, 2003b). Both data sets were classified using ERDAS Imagine image analysis software.…”

“…The cover class to which a pixel is assigned during the classification process can be affected by the spatial error. For example, the spatial error reported by Garono et al (2003b) for classification of two Landsat 7 ETM+ scenes were 0.31 and 0.45 pixels, representing an on the ground distance of 9.3 to 13.5-m, respectively. A 10-m shift in the area that a pixel covers can, in some circumstances (see discussion above), influence the cover class to which that pixel is assigned during the classification process.…”

“…The "from" data set was created by Earth Design Consultants, Inc. (EDC) for the Estuary Partnership as part of a larger habitat mapping project (Garono et al, 2003b;Garono et al, 2003a). It is a classified Landsat 7 ETM+ scene (47/28) acquired on March 24, 2000, and will be referred to as the TM2000 data set.…”

“…It is a classified Landsat 7 ETM+ scene (47/28) acquired on March 24, 2000, and will be referred to as the TM2000 data set. Although two Landsat 7 ETM+ scenes were classified in the EDC study (Garono et al, 2003b), we selected only the western portion of the TM2000 data set for this change analysis because comparable C-CAP data only exist in this area (Fig. 1).…”

“…Fourteen cover classes were identified, using a combination of unsupervised and supervised classification, from spectral information alone with data from 244 training sites. Remaining cover classes were identified from the spectral data plus other ancillary information including, a diked area mask, digital elevation data, digital orthoquad photographs, aerial photographs, and local knowledge (Garono et al, 2003b). …”