SUMMARY

The GIS analysis that was performed identified aspen contributed forest stands within northeast Michigan.  These stands were identified from a 1992 Landsat 5 Thematic Mapper image by processing the image within the Erdas Imagine 8.3 remote sensing software package.  The steps that were performed included, overlaying and registering the raw imagery, collecting GPS ground coordinates to identify the position of an aspen stand similar to the desired type, classifying the imagery based upon the signature collected at the area of interest delineated by the GPS point, clumping the resultant classification together, and performing a sieve operation to weed out any polygons that represented less than five acres. 

Because of the limitations of experience and file transfer size, additional processing steps were performed with Arcview's Spatial Analyst Extension.  Spatial Analyst was able to convert the image file to a grid file (its own format) and then, based upon the attributes within the grid file, was able to select all polygons except the background polygon and convert it to an Arcview shape file.  This process transferred the polygons that represented the aspen stands from the image format to a usable format for the GIS. 

At the same time the processing was happening to the image files, the GIS data was being manipulated to 'meet' the image data for future integration.  The original data taken from the Environmental Systems Research Institute's (ESRI) 1998 GIS Map Data CD had to be reprojected from its native projection (Geographic, WGS84) to the projection that matched the data from the satellite imagery (UTM, Datum Unknown).  The data that was reprojected included Michigan counties, rivers, and roads.

The GIS step was to create the 1500 ft buffer around the rivers layer that would eventually help to identify the desired aspen stands.  This buffer represents a grouse's affinity for water.

At the point of overlaying all the data, the first 'real' problem came up.  When all the data was put together, everything except the aspen locations aligned with one another.  After investigating this for some time, the conclusion was reached that the error must have been with the original identification of the image's datum.  The header file that contains this type of information, specified the datum as code 8, not knowing what this was, the datum best fitting the project's needs was selected.  This appears to have been the mistake, having the improper datum caused for a misalignment with all the other data.  To correct this problem, a roads layer containing all the major roads in northeast Michigan was screen digitized from the image.  The next step was to create a 'link' file within ESRI's Arc/Info GIS that would adjust the newly digitized roads to the same coordinate space as the GIS roads that were aligning with all the other data.  The same link file was then applied to the image classified aspen stands to move them into a relative, in relation to the other layers, correct, coordinate space.

At this point the two main data layers were complete and the final analytical step could take place.  This step was to overlay the image classified aspen contributed forest stands with the rivers proximity buffer to produce a final layer that met the selection criteria.  These criteria were that the aspens had to match the signature found at the desired GPS location, they had to be greater than five acres in size, and had to be within 1500 feet of a river.  These are the parameters that should produce locations that contain high grouse concentrations for the analysis described in the introduction of this web page.