Based on the decline of the population of the sockeye and chinook salmon along the Snake River, the Federal Government listed the two fish as endangered under the Endangered Species Act. As one possible means of preventing extintion of these species, the federal Government is investigating the permanent drawdown of the four dams along the Snake River. These dams are the Ice Harbor, Lower Monumental, Little Goose, and Lower Granite. This paper presents a portion of the methodology used to evaluate the impacts on transportation systems by this proposed alternative.
Constraints were also included in the model to represent maximum allowable rail traffic and barge traffic at selected location. These allowed the before and after analysis of impacts on the transportation system. The model first found the optimal means of transporting all wheat to Portland at minimal cost under existing conditions, then the constraints were modified prevent any barge traffic upstream of the four dams and re-run. The difference between the two runs were then used as potential impact of the proposed drawdown on the transportation system. HDR Engineering, Inc. was hired to review the previous model and make recommendations to update the model and include revised figures available in 1998. It was decided first to rewrite the model in an Excel spreadsheet using a commercially available optimizer add-in called "What's Best". The ease of developing and modifing the model in Excel proved invaluable in revisions and analysis that would have been far more difficult in GAMS. As an added benefit, the spreadsheet format is easier to read, print, and integrate into a database, especially to the user unfamiliar with GAMS.
Several revisions were made to the model uaing the ArcView GIS program. The previous model had used ArcPlot to find routes from each production center to various destination locations along a road coverage. The arcs traversed along these routes were summarized in table which consisted of each route number and every arc id associated with it. This table had almost 1.5 million records. Display of the previous model was done using ArcPlot and dynamic segmentation.
For the revised model, numerous routes had to be added to the model to assure the full impact of the drawdowns were represented. The Network Analyst extensison was use to find the shortest route from the two points specified and an Avenue script identified each arc associated with that route. These values were then added to the original table for later analysis.
Once the optimization model was run in Excel, the results were summarized in a table consisting of the total amount of wheat for each available route in the model. The result required the total amount trucked (or shipped by rail or barge) for each segment of the road network. This was quickly done using an Access database. The two tables involved were the table listing the routes with assciated road arcs and the table of total traffic by route. These tables were then joined via the route field and summarized by the arc id. Figure-4 shows the design of the Access query that accomplished this operation.
Through the use of an ODBC connection to Access, the table listing the total traffic for each road segment was joined to the road coverage of the area. This allowed a visual display of results on a standard ArcView interface. The display used a graduated red line symbol to show the total traffic keyed to the width of the line symbol. A dramatic display of the pattern of truck traffic was easily created in this manner. Figure-5 shows the traffic under the existing conditions while Figure-6 show the traffic pattern after the proposed drawdown. The obvious relocation of traffic to the west or downstream of the Snake River is evident.
In addition to the standard display of total truck traffic on the ArcView interface, the difference in total truck traffic for each segment could also be displayed. Figure-7 shows this difference with an incerase in traffic shown in red with a decrease in traffic shown in green. This allows a visual display of the areas most impacted by the proposed drawdown.
As an added result of the linear programming model, a dual value is also computed for each constraint in the model. For the contraint of total wheat shipped must equal the total wheat produced, the dual value gives the transportation cost from each production zone. These values can be directly joined to the shapefile of the production zones to show the increase in transportation costs for each area in the state. Figure-8 shows the relative increas in transportation cost by production location. The maximum increas shown is approximately 9 percent. The display shows what was expected, i.e. the greatest impact occurs the further upstream along the Snake River.
The model was also used in a similar fashion to analyze and display the impacts of the proposed drawdown all the other commodities currently barged along the Snake River. Similar methods were used to display the impacts on rail traffic and barge traffic.
Using the results of the analysis described in this paper the total costs associated with the proposed drawdown of the four dams along the Snake River in Eastern Washington were estimated due to numerous factors including infastructure improvements, paving thickness, interchange improvements, and impacts to local traffic in the cities located downstream of the dams in question. Improvements required to upgrade the railroad system in the area were also analyzed to estimate the cost to handle increases in their systems. The use of an integrated approach using Excel, Access, and ArcView GIS allowed a simplified interface for analyzing and displaying the results and comparing various alternatives in the project. The methods described here could also be used for other studies and would not be limited to transportation studies.