Landcover Classification of the Nyack Floodplain Using Remote Sensing and Comparing MultiSpec and ENVI Software

For the past two summer's, I have had two internships studying the Nyack Floodplain south of Glacier National Park in Montana. The internships were with the Flathead Lake Biological Station and with the Landsat Data Continuity Mission at NASA's Goddard Space Flight Center. During the summer of 2003, my internship with the Flathead Lake Biological Station for the biodiversity study to complete a landcover classification of the Nyack floodplain. That same summer my NASA internship was a three-week training course on the basics of Arcview GIS and learning remote sensing using MultiSpec software. That same summer, I used a Trimble GPS backpack unit in the field to collect polygon data of the different landcover types and downloaded the data into Arcview. I also assisted three University of Montana Graduate students with their field research on the same site. During my internships in the summer of 2004, I continued my landcover classification of the Nyack floodplain for the Flathead Lake Biological Station, and I attended a five-week internship at NASA's (Goddard Space Flight Center). During this internship, I continued my landcover classification of the Nyack floodplain. I had two goals for my internships: to identify stressed stands of cottonwood trees along the losing reaches and gaining reaches of the Nyack floodplain using Landsat 7 imagery, and to compare the MultiSpec and ENVI software. The MultiSpec software is a freeware provided through the University of Purdue. This software is very easy to use, with an abundance of tutorials and web pages to make it very user friendly. While the ENVI software is available by purchase. ENVI is a more advanced software that provides more functionality than MultiSpec. Each software did an exceptional job of classifying the Nyack floodplain. I was able to use both MultiSpec and ENVI to classify the landcover types of the Nyack floodplain, but the ENVI program gave me more options and made the supervised classifications easier then the MultiSpec program. A combination of classifications and fieldwork allowed me to locate specific areas on the imagery like the losing and gaining reaches and the cottonwood stands. The cottonwood stand were identified by the use of hyperspectral aerial photographs taken by Dr. F. Richard Hauer of the Flathead Lake Biological Station, and the losing and gaining reaches of the river were identified through research by Dr. Jack A. Stanford of the Flathead Lake Biological Station. Research by Dr. Jack A. Stanford, indicated that the cottonwood stands in the losing reaches of the floodplain were stressed compared to the cottonwood stands in the gaining reaches of the floodplain. This was due to the losing reaches of the floodplain removing nutrients from the substrate of the river, and dispersing them throughout the floodplain through ground water flow paths. While the gaining reaches of the floodplain had less stressed cottonwood stands, due to the nutrients added to these areas from the ground water forced back up into the substrate of the river channel. Using Landsat 7 imagery, MultiSpec software, and ENVI software to identify these areas, I was unable to determine what stands of cottonwood trees were the ones stressed and what ones were not. This is due to the 30-meter pixel size resolution of the Landsat 7 imagery. Higher resolution imagery may be able to identify these stressed stands of cottonwood trees along the losing and gaining reaches of the Nyack floodplain. Since my results came back inconclusive using the Landsat 7 imagery, my next step will be to use higher resolution imagery to see if it can determine these stressed cottonwood stands.