NCRST-E Logo NCRST-E banner National Consortium on Remote Sensing in Transportation - Environmental Assessment
Home Projects Members Publications Committee Data RSPA Meetings Contact

Remotely Sensed Land Cover Classification for Corridor Planning:
Example of Western Alabama Corridor


Introduction
Image Processing
Results
Socio-economic Data Return to Environmental Assessment Home
Return to NCRST-E Home


Introduction
Environmental streamlining brings together the timely delivery of transportation projects with the protection and enhancement of the environment. It is generally assumed that greater efficiency in acquiring and analyzing data used in EIS preparation, and better data standards, would reduce EIS preparation time and possibly result in less controversy over the report's findings. Environmental or development planning requires geospatial information about the distribution of landscape features. Maps, and to some extent aerial photographs, are the traditional sources of this information. Remote sensing offers tremendous potential for planning purposes because it not only affords a picture of the current state, but unlike aerial photographs, multispectral data can be used to provide information classes, such as land cover and land use. Continuity in temporal classification of land cover and its extension to land use can play a significant role in preparation of a comprehensive development plan and are invaluable in the planning stage of road development projects. Data from the Landsat satellite program are invaluable sources of information for regional- and some local-scale planning. Landsat data at 30 m resolution is suitable for relatively general natural resource classes, such as softwood and hardwood forests, pasture, cropland, residential and industrial areas, water bodies, etc. The U.S. government has funded the Landsat program continually as part of an on-going mapping program and data are relatively very inexpensive.

A north-south interstate highway has been proposed for western Alabama primarily to spur development in what's locally known as "the Black Belt," which historically has lagged in jobs and income. The idea is that the new highway would help revitalize the area, provide a direct route from western Alabama south to the Port of Alabama at Mobile, and enhance economic opportunities along the Tennessee-Tombigbee Waterway. The $3 billion cost of that project would be partially offset by leasing the highway's right of way for pipelines, fiber optic cables and other utilities.

The proposed highway would follow U.S. 43 from Mobile north to Muscle Shoals via Tuscaloosa and then possibly enter Tennessee (Figure 1). Federal appropriations were secured to conduct a feasibility study. Originally planned between Muscle Shoals and Tuscaloosa, the freeway will be studied all the way south to Mobile. The total route, approximately 320 miles long, would extend from Mobile roughly along a line following the Tombigbee and Black Warrior Rivers to just west of Tuscaloosa, then northward through Fayette, Marion, Franklin and Colbert counties to Tuscumbia. It would connect to Corridor X/U.S. 78 and Interstate 65. The northern portion of the highway is considered in two segments: from Tuscaloosa to Corridor X (~ 60 mi.) and from Corridor X to Tuscumbia (~ 50 mi.). The boundaries of the corridor to be considered for siting the highway were defined by the Alabama Department of Transportation (Figure 1).

True color image map of the northern half of the proposed Western Alabama Corridor highway.
Figure 1. "True-color" image map of the northern half of the proposed Western Alabama Corridor highway. The boundaries of the corridor to restrict siting are shown. Corridor X divides the Western Corridor into two segments.
(top) Return to Environmental Assessment Home


Image Processing
Acquisition
Four Landsat Thematic Mapper images of the northern two segments of the western Alabama corridor were acquired for the purpose of classifying land cover. Due to the length of the northern portion of the corridor, two images were required to cover the area of interest. To improve classification accuracy, we sought images acquired during the time of year when the leaves are off of the trees and when the vegetation is fully leafed out. A pair of cloud-free images was found in the archives of the U.S. Geological Survey on March 5, 2001 and September 29, 2001.

Georectification
The images were acquired in TIFF format with Universal Transverse Mercator projection with WGS 1984 horizontal datum. However, as is not uncommon, the georectification of the image pairs from the different dates did not agree. Using an independent source, the images from September were determined to be incorrectly rectified by 57 m in the easterly and 57 m in the northerly directions. An appropriate correction was applied to both September images.

Atmospheric Correction
The radiance observed by a space-borne sensor is comprised of a component of energy reflected or emitted from Earth's surface as well as from scattering of energy in the atmosphere. The amount of scattering that occurs is a function of wavelength and must be assessed and explicitly removed from each image band. The atmospheric contribution to each image was removed using a modified form of the dark object subtraction technique.

Segmentation and Classification
In the preliminary classification, the decorrelation stretch images from both March and September 2001 and the tassel cap transformation image from March 2001 were used in the classification. Use of transformation products is sometimes used to reduce the information content of images to a smaller number of bands. In this case, the benefit was minimal so that the final classification was based on the six bands from the March 2001 image and bands 2, 4, and 5 from the September image (Figure 2a).

First the image is segmented. That is, adjacent pixels with similar spectral characteristics are grouped into a segment or image object (Figure 2b). After training by the image analyst, the segments are then clustered into classes (Figure 2c). Thirteen land cover and land use classes were defined for the preliminary classification (Figure 3).

Figure 2a. True color image of a portion of the Western Corridor showing the Tennessee River at Tuscumbia, Alabama. True color image of a portion of the Western Corridor showing the Tennessee River at Tuscumbia, Alabama.
Figure 2b. Segmented image in which adjacent pixels have been grouped based on similar spectral characteristics. The yellow polygons delineate segments. Segmented image in which adjacent pixels have been grouped based on similar spectral characteristics.
Figure 2c. Classified image in which each segment is assigned to a land cover type. Classified image in which each segment is assigned to a land cover type.

Classification system
Figure 3. Classification system used in the classification of the Western Alabama Corridor.
(top) Return to Environmental Assessment Home


Results
True color images of the corridor provide valuable insight into the nature of the terrain and land cover and land use types that the proposed highway might pass through. Floodplains and the number of water bodies that must be crossed or avoided are also readily interpreted from such images. Figure 4 shows the land cover/land use classification results. Most of the area along the northern corridor is in heavily forested land from which timber is harvested. The valleys between areas of relief are commonly occupied by grassland on which cattle are raised. The soil in much of these areas is too rocky for cultivation. Agricultural land dedicated to row crops exists near the northern end in the Tennessee River valley.

Landcover classification for the northern half of the proposed Western Alabama Corridor.
Figure 4. Landcover classification for the northern half of the proposed Western Alabama Corridor.


Socio-economic Data
Complete socio-economic results can be found here.

(top) Return to Environmental Assessment Home


URL:
http://wwwghcc.msfc.nasa.gov/land/ncrst/walacorr.html

Contact: Charles Laymon (charles.laymon@msfc.nasa.gov)
Page Curator: Paul J. Meyer (paul.meyer@msfc.nasa.gov)
Responsible Official: Dr. Steven J. Goodman (steven.goodman@nasa.gov)

Last modified: Thursday, December 2, 2004 10:29:36