Identification of Landform Gully Sites Using Remote Sensing

William Brown (Dec. 2005)

ES 771 Remote Sensing

 

Introduction:

Landform gullies are channels formed in the Earth by running water usually after a rainfall event. Gullies start off as small channels called rills but as erosion continues to the extent that the channel becomes too deep to cross in a vehicle, they are classified as gullies. Landform gullies are formed by land disturbance events such as cultivation, tree clearing, improper road construction and maintenance, and new building construction. When these gullies have a hydrologic connection to a receiving water body such as a stream there is a great potential for degradation of water quality due to transport of sediment and potential contaminants.

Remote sensing is a technology that may be used to identify and inventory these erosion features. This document will focus on a methodology for the identification of landform gullies with a hydrologic link to distinguish them from other disturbed land features such as clay pits, new development, and off-road recreation. Remote sensing and GIS technology will be used to interpret aerial photography and Landsat satellite imagery for potential gully features.

Gully Photos:

The photographs below are examples of known gully systems identified in the Panhandle region of Florida. Figures 1 and 2 are of a large gully formed by the removal of the vegetation and soil from a clay pit site. Figure 1 is an upstream view leading towards the clay pit excavation. Figure 2 is the downstream view leading into a stream. This appears to be a highly erosive site and could lead to significant water quality degradation and changes to the profile of the channel. Figure 3 is a photo of a roadside gully, which formed in an unpaved road ditch that was designed for sediment control but is now a source for erosion. Figure 4 is an example of a gully that formed likely due to timber harvesting activities.

 

Figure 1.  Upstream view of a gully created during the excavation of a clay pit (Photo by Michael Rainer).

Figure 2. Downstream view of the same gully as in Figure 1 showing its connection into a stream (photo by Michael Rainer).

 

Figure 3. Gully formation in a roadside ditch (Photo by Michael Rainer).

 

Figure 4.  Gully formation in a longleaf pine plantation (photo by Michael Rainer).

 

Gully Aerial Views and Interpretation:

The figures below are of known landform gullies identified in the Panhandle region of Florida from digital ortho quarter quads (DOQQs) downloaded from the Land Boundary and Information System  (LABINS, 2005).  The aerials were flown in February 2004 which helps to reduce vegetation cover on the sites.  The DOQQs are in true color and are 1 meter in horizontal resolution.  

Site 1 shows several gullies which likely formed from excavation activities from the clay pit (Figure 5).  The gullies are draining into the large stream west of the excavation pit.  Also visible in white are the two roads leading to the clay pit from the north and one road heading south away from the pit.  Site 2 is north of site 1 and is an example of a gully system formed in a roadside ditch along the unpaved road running east to west in the image.  Once again there is a hydrologic connection with the nearby stream.  Sites 3 and 4 are land disturbance sites of unknown origin.  The site 3 gully runs southwest into the stream.  Site 4 is a very large gully running northeast and depositing into a stream bed.  At the interface with the stream an alluvial fan has developed and the stream has actually been redirected northwards of the alluvial fan.

Figure 5.  Site 1 gully system associated with a clay pit excavation site and draining into a nearby stream located in Santa Rosa County, FL.

Figure 6.  Site 2 gully system associated with a roadside ditch along an unpaved road located in Santa Rosa County, FL.

Figure 7.  Site 3 gully developed from a land disturbance area and draining into a nearby stream located in Okaloosa County, FL.

 

Figure 8.  Site 4 large gully developed from a land disturbance area<SPAN> and draining and depositing sediment into a nearby stream bed in Walton County, FL.

Landsat Imagery Interpretation:

A Landsat TM image was obtained from January 7, 2003.  The image is a scene which covers the same general extent as the DOQQs previously described but with a 30 meter horizontal resolution.  As with the DOQQs, a winter image was chosen in order to reduce vegetation cover which might make potential gully sites more difficult to identify (Figure 9).  

In order to automate the identification process over a larger geographic extent, an unsupervised classification technique was applied.  Using the IDRISI Kilimanjaro CLUSTER module, Landsat bands 1 through 5 were used, along with a 1% saturation level, and a broad generalization level as input parameters.  The resulting image produced 8 distinct clusters which were further reclassified into 3 distinct categories; gully, disturbed land, and other.

Sites 1 through 4 are presented below (Figures 10-12).  The cyan color represents potential gully sites, red is disturbed lands, and yellow is all other categories.  The blue outlined areas represent the digitized gully based on the DOQQs and true color Landsat image.  Sites 1, 2, and 3 show a fair agreement with the reclassified image.  The site 4 reclassification identifies the area as mostly disturbed land versus gully area.  

Figure 9.  Landsat TM imagery from January 7, 2003 in true color with the four previously described gully sites identified.

   

Figure 10.  Sites 1 and 2 based on the clustered and reclassified Landsat TM imagery from January 7, 2003.

Figure 11.  Site 3 based on the clustered and reclassified Landsat TM imagery from January 7, 2003.

Figure 12.  Site 4 based on the clustered and reclassified Landsat TM imagery from January 7, 2003.

Conclusion:

Using remote sensing to classify landform gully sites has potential for success based on the limited data analyzed.  Limiting factors include the resolution of the imagery and the time of the year.  Landsat imagery is limited to 30 m resolution versus the 1 m resolution of the DOQQ.  The Landsat imagery may not be adequate for small or newly formed gully sites.  Analyzing the DOQQ for gullies produces more accurate results but is a very tedious process.  Running a proximity analysis to collect the potential sites near hydrologic features will further reduce the number of sites which are connected hydrologically, making the process more manageable. Further research will need to be conducted to fine tune this methodology.

References:

Aber, James (2005),  Emporia State University, ES 771 Remote Sensing, Lab 8 Vegetation Analysis, http://academic.emporia.edu/aberjame/remote/lab08/lab08.htm, Accessed October 23, 2005. 

Land Boundary Information System (2005), Bureau of Survey and Mapping, Florida Department of Environmental Protection, http://www.labins.org/, Accessed on December 5, 2005.

Rainer, Michael, Soil Scientist, Personal Communication, December 2, 2005.