Merged GIS, GPS data assist siting for gulf gas line

Dawn R. Scott
ESRI
Redlands, Calif.

Jon A. Schmidt
ENSR
Northborough, Mass.

A GIS-based decision-support system was developed for a U.S. Gulf of Mexico onshore and offshore pipeline that has assisted in locating a cost-effective pipeline route based on landcover type, wetland distribution, and proximity to other environmentally sensitive resources. (GIS = geographical information system)

Fugro West Inc., Ventura, Calif., and ENSR, Acton, Mass., worked with Destin Pipeline Co. LLC, Birmingham, Ala., to plan and provide environmental documents for the 214-mile Destin pipeline project that stretches from the gulf into Mississippi.

Extensive data were needed for the project, including maps of various sizes and scales as well as data concerning the impacts of the new pipeline on surrounding land uses. A consistent land-use map for the entire study region was also necessary to evaluate pipeline routing alternatives.

Fugro developed a large-scale satellite image processing system that classified land-cover types in Mississippi and Alabama at a resolution of 20 m. Multispectral SPOT imagery was used as the primary source.

Additional sources of digital map information were collected from federal and state agencies (roads, sensitive resources, county boundaries, etc.), as well as offshore features such as existing pipelines and platforms. These data were combined with digital maps of the proposed and alternative routes.

Fugro also collected highly accurate field information using GPS (global positioning system) receivers. These data were entered directly into the pipeline-support system to refine estimates of environmental impact on wetlands and to precisely locate stream, road, and utility crossings.

Fugro then developed analytical routines within the GIS, based on dynamic segmentation techniques, to provide stationing numbers as the route crossed sensitive land-cover types (such as wetlands).

Described here are the methods used to integrate various sources of available GIS data with satellite imagery and surveyed information. Costs of collecting and processing these data are compared with benefits of the system over use of manual methods.

Company reluctance

GIS, GPS, and satellite-image processing techniques are no longer new methods for analysis of spatial data. However, these tools are not used by many companies and consultants involved in utility mapping and modeling.

The reasons are complex, relating to acceptance of computer-based techniques and an overemphasis in the GIS "world" on creation of extremely costly, yet not always extremely useful, data bases.

One approach to improving acceptance of GIS technology is to focus on costs and benefits, before the project begins, to convince the project manager or client that a more cost-effective and better product will be produced through the use of these methods.

One way to do this is to focus attention on the unique analytical capabilities of GIS as opposed to focusing on standard map production or inventory processes.

A second, related approach is to make use of the wealth of existing digital data that exist. These data, ranging from utility information to scanned quadrangles to satellite imagery, can "jump start" a project implementation and win early support from decision-makers.

The layers can then be integrated with more-accurate information later, if the data base is properly structured. The project described here is an example of an integration of general-accuracy information which solved a pipeline company's immediate needs cost-effectively. Later, more-accurate data were added to enable further analyses.

The project

The objective of the project, begun in May 1996, was to develop a siting-constraints data base, perform an alternatives analysis study, and prepare a FERC (U.S. Federal Energy Regulatory Commission) Environmental Review for the Destin pipeline.

The line runs between Main Pass Block 260, offshore Gulf of Mexico, comes ashore near Pascagoula, Miss., and terminates at Southern Natural Gas Co.'s South Main Line at the Enterprise compressor station in Clarke County, Miss.

This approximate route includes 120-130 miles of pipeline onshore and 80-90 miles offshore. Fugro was asked to conduct the study and to develop both onshore and offshore supporting GIS data bases. Fig. 1 [331,538 bytes] presents the study area and pipeline location.

A large data base was needed to conduct this project, including environmental and infrastructure information covering the offshore area and the nearshore areas of Mississippi and Alabama from Biloxi, Miss., to Mobile Bay, Ala. At this level of planning, the positional accuracy of the data base needed to be fairly general, ±50 ft.

U.S. Geological Service (USGS) quadrangles were used as a base for the pipeline alignment sheets that were initially digitized. Some of the questions that would have to be answered with the GIS data base included:

• For each pipeline alignment, how many miles of each type of land use were traversed, on a county-by-county basis?
• Are there any drinking water wells within 150 ft of each alignment (centerline)? At what milepost?
• Does the pipeline cross any wetlands? If so, what is the beginning and ending milepost for each wetland?
• Are any streams or rivers crossed? If so, what is the milepost of the crossing?
• Are any highways, roads, or utilities crossed? If so, what is the milepost of each crossing?
• What is the total length of each alignment?
• What is the total length within each county? At what milepost does the pipeline cross a county boundary?
• What is the milepost of landfall?
• At what mileposts does the pipeline cross into or out of public lands?
• At what mileposts does the pipeline cross within a mile of a listed special-status species? How many times did this occur along each alignment?
• At what mileposts does the pipeline cross within a mile of hazardous waste sites, landfills, or underground storage tanks? How many times did this occur along each alignment?

Many of these questions require some type of spatial analysis, including overlay analysis, intersection, buffer analysis, and dynamic segmentation. Fugro's GIS staff, basing an economic analysis on time and resources invested in previous projects, estimated that a manual approach, involving locating (or creating) paper land-use maps from counties, cities, or other public agencies, and digitizing the maps, would require a budget of more than $150,000, or more than $1,250/mile.

To save money and to provide a more-accurate, current product with potential utility in the long term, an automated method involving GIS, satellite-image processing, and eventual use of GPS satellite positioning, was devised.

The process

A four-step process designed to conduct this project included acquisition and processing of satellite imagery, development of a GIS data base, data base analysis, and map and table production.

Acquiring, processing imagery

The first step in the project was to develop a good land-use map for the entire study area. Fugro contacted various satellite-image vendors and chose SPOT Image, Toulouse, France, based on availability of recent imagery and an acceptable resolution.

Eight scenes of multispectral imagery, with a resolution of 20 m, were chosen from SPOT Image's 1996 archives to cover the region. These data were imported into the image-processing software and rectified to a standard coordinate system. Each scene was approximately 50 mb in size.

The imagery was then classified by land use/land cover categories, including wetlands (forested, scrub shrub, and emergent), industrial/commercial, agriculture, pasture, water, barren/clearcut, scrub brush, deciduous forest, open water, and planted pine.

Classification is an iterative process. First, training samples are developed using known locations of various land-use types. The imagery is then classified based on these training samples. Map output is produced, and the draft maps are reviewed by biologists and specialists having knowledge of the area. The classification is then modified.

This process was conducted until the biologists and in-field personnel were satisfied with the accuracy of the classification. The outcome of this process was a consistently developed land-use map covering the entire study area with good, general accuracy suitable for regional analyses.

This process took approximately 2 months (1 month for data acquisition and 1 month for processing).

Developing GIS data base

At the same time, a GIS data base was being constructed to support environmental analyses.

A large number of regional data layers were needed. To keep costs low, various existing data sources were obtained, such as U.S. Census Bureau's TIGER files, based on information from the U.S. Census Bureau, for general geographic features, U.S. National Resource Conservation Service (NRCS) data for soils, and U.S. Environmental Protection Agency (EPA) information regarding hazardous waste sites.

Many of these sources were located and obtained through standard internet searches and downloads. Other data were obtained from vendors and agencies in the states of Mississippi and Alabama.

Finally, offshore data for the study area were obtained from the John E. Chance GIS data base, a proprietary source of pipeline, structure, and other offshore information.

Once digital data were obtained, the format was modified through standard data-conversion commands to create a single, consistent Arc/Info data base. (Arc/Info is database software produced by ESRI.) Projections were changed as necessary.

The data base was then provided to the project manager in the form of an ArcView project file. This allowed the project manager to visualize and demonstrate use of the data base to project staff and the client.

Because the GIS staff for this project was located on the West Coast of the U.S. and the project manager, environmental staff, and client were located on the East Coast, the ArcView data base was used as a tool to get the data into the hands of those requiring it to make project decisions.

A listing of source materials used to support the project is included in Table 1 [265,858 bytes].

Data base analysis

Three primary types of analysis were conducted with the data base and GIS processes:

• Buffer analysis. This was used whenever information regarding proximity to the pipeline was needed.
  By creating a buffer around the pipeline data layer at a user-specified distance and overlaying the buffer with the data layer in question (wells, agricultural land uses, sensitive species locations, etc.), it is possible to determine the number and identity of these features or areas within the buffer distance.
  In another example, it was necessary to determine the distances of archaeological sites within a certain proximity (such as 150 ft) from the pipeline centerline. This was accomplished by writing a program that ran a buffer analysis, calculated straight-line distances between the sites and the pipeline, and wrote these distances and the site-identification to an output file.

• Overlay analysis. This type was employed whenever questions came up regarding the location of two or more types of data.
  For example, the question regarding the length of pipeline in each county was answered by overlaying the pipeline data layer and the county boundary layer. It should be noted that overlay analysis is not the same as graphical analysis, as performed by a standard CAD system.
  An overlay analysis within a GIS creates a new layer of data, which in this example is the combination of pipeline arcs and county polygons. A quick summary analysis on this layer then yields the length of pipeline within each county.

• Route analysis. Most of the questions posed by the project manager related to the length that each pipeline route alternative traversed various sensitive resources, such as wetlands or stream crossings. These questions were answered by use of the route functions within Arc/Info.
  The concept of dynamic segmentation is of interest in this type of study because it affords a great deal of analytical flexibility. Dynamic segmentation allows association of attributes to a linear feature, such as a pipeline, based on a route-measure format.
  It was therefore possible to work with attributes of the pipeline in terms of mileposts rather than x,y coordinates. Furthermore, attributes could be modified by changing mileposts without affecting the underlying pipeline coordinates.
  Because we were dealing with satellite imagery, our data were in grid, or "raster," format. We used the extensive grid-processing commands in Arc/Info to create a "route" representing the pipeline alternatives.
  The route was then segmented on a milepost basis according to land use/land cover type. Finally, the mileposts/land-use tables were summarized and printed, allowing comparison between the pipeline route alternatives.
  We found this type of data-integration tool (raster and vector) to be highly valuable in answering the analytical questions posed by the project manager.

Map, table production

A series of maps showing the study area, pipeline route, and environmental features were produced to illustrate the report. During the initial phase of the project, the maps were of interest but were not as useful as the tabular products.

The calculated lengths of traversed features and milepost information, generated as described previously, provided quantitative information which could be used in comparing pipeline route alternatives.

Data integration

During the second phase of the project, more-accurate information was needed regarding the location of the pipeline and the environmentally sensitive features potentially affected.

Over the course of several months, surveyors used GPS equipment to obtain precise coordinates of the pipeline centerline, wetlands crossed by the pipeline, and archaeological sites.

Each week, a data delivery was made to the GIS support group. Data were obtained in ASCII format (x,y coordinates) and imported into the system using routines that automatically created lines from the coordinate data. During this phase, issues of projection and datum were critical in integrating the survey data into a unified GIS data base.

Following data entry, analyses similar to those previously described were conducted.

Route analysis allowed the production of milepost information to indicate where the pipeline crossed wetlands, streams, or other sensitive features. Buffer analysis located environmentally sensitive features, such as threatened and endangered species, wells, and archaeological sites within a particular query distance.

Again, these data proved equally or more valuable than the maps that were generated to illustrate the pipeline route.

Costs, benefits

The use of GIS/GPS/remote-sensing techniques on this project proved valuable in several areas.

• Use of satellite imagery provided a quick method of developing a consistently formatted land-use map in a relatively short time.
• Second, use of existing data sources kept the initial costs to a minimum while providing a good overall visualization tool for the project team.
• Use of GIS analytical processes enabled relatively quick data in response to questions related to environmentally sensitive features in and around the proposed pipeline.
• GPS provided highly accurate survey data during later phases of the project in a format which could be easily integrated with the GIS data base.

Understanding and quantifying the benefits of these technologies are revealed by their costs in relation to costs of other, similar projects.

Land-use mapping

In a study example developed by Spot Imagery, several alternative sources of land-use information were compared.

The objective of this project was to produce a map showing crop distributions in northern California (a region of 2,800 sq miles).

One option is aerial photo interpretation, requiring 125 frames (color infrared, 5-m resolution, obtained semimonthly over a 6-month period). These photos would have to be obtained on a semimonthly basis, and then geometrically corrected, scanned, and resampled.

The overall cost of the photography and processing was estimated to be $500,000. Alternatively, satellite imagery (10-m resolution) could be acquired and interpreted in a much shorter time (days, rather than months for the photos) for the estimated cost of $72,000.

In this study, the cost of the satellite-image acquisition and interpretation was less than $40,000, or approximately $300/mile. This compares favorably to the earlier estimate of manual land-use map generation for approximately $1,250/mile.

In addition, because the satellite imagery was georectified consistently, the overall product tends to be more accurate than relying on individual maps or interpreted photos with varying degrees of distortion across the region.

Several new companies have entered the geospatial imagery market recently. These new companies (EarthWatch, Space Imaging EOSAT, TRW), along with existing firms, such as SPOT Image, and such government agencies as the National Aeornautics & Space Administration (NASA) and USGS, are advertising more accessible sources of imagery at much greater resolution.¹

The imagery provides great potential for new GIS users to "jump start" their GIS applications with readily available, low-cost data sources.

Environmental analysis, mapping

Another study describes a similar mapping/analytical project for the Sitco/SunShine Pipeline Co. ²

This study involved environmental mapping for a proposed 600-mile pipeline between Pascagoula, Miss., and Polk County, Fla. It involved use of satellite imagery early on, as well as environmental mapping from existing GIS sources and GPS survey/integration. The study authors estimated a total cost of $2,000/mile.

The approximate GIS cost of the Destin project, including the types of mapping and analyses described in the Sitco project, is $625/mile. This cost does not include the GPS survey or archaeological survey-only the GIS-project costs. GIS was a cost-effective tool because of the proliferation of high quality, accessible data sources that have become available in the last few years.

Another factor in the low cost of GIS on this project is the reliance by Fugro on automation routines (Arc/Info AML programs) that analyzed the data in batch mode to minimize operator input and increase overall efficiency.

Acknowledgment

The authors wish to acknowledge the contribution of John Barfield, Southern Natural Gas Co., Birmingham, to this project.

References

1. Carlson, George R., and Patel, Beni, "A New Era Dawns for Geospatial Imagery," GIS World, Vol. 10 No. 3, March 1997, pp. 36-40.
2. Schmidt, Jon A., and Kiefer, K., GIS/GPS Data Acquisition for SITCO/SunShine Pipeline Projects, Ecology & Environment Inc.

The Author