North Coast Hiawatha Restoration Study released by the Rail Passengers Association

Synopsis of projected North Coast Hiawatha ridership in the September 30, 2021 study done by the Rail Passengers Association for the Big Sky Passenger Rail Authority:

RPA’s North Coast Hiawatha route study proudly touts a projection of handling 426,000 passengers annually. Ridership drives economic benefit and cost recovery, which was also included in the study. Projections did not consider components such as train schedule, train equipment, amenities at stations, the number of stations, required new mechanical facilities, track and signal infrastructure that would be demanded by host railroads, and crew districts along the route. While exclusion of these components are indeed fatal flaws, this analysis considers the results at face value. And while RPA repeats that the study was not meant to be comprehensive and that other studies need to be done in the future, this analysis considers only the product (the ridership figures) and not the process.

RPA ridership is indicated by county. Proposed station stops are indicated within the county, but if the county has more than one stop, specific station ridership is not conveyed – only the county ridership. One huge flaw in the RPA system is that it evidently only considers the county indicated (see chart below) without taking into account adjacent counties. The population indicated is usually the 2019 U.S. Census estimate. The biggest gaffe in this regard is indicating the population for only Ramsey County for the St. Paul/Minneapolis stop (St. Paul is in Ramsey county) while ignoring (evidently) much-more-populous Hennepin County (Minneapolis). For three of the counties in the study, the proposed station stop is not even in that county.

With this in mind, ridership in each of the 43 counties with their 51 projected station stops will be analyzed. Below is the ridership projections table for the North Coast Hiawatha from the study, with references to the most-recent station ridership figures for the Empire Builder from the RPA website.

The RPA study is at:

2020 Montana county population is shown:

Empire Builder ridership:

Note about Empire Builder ridership: Amtrak lists station ridership by “boardings and alightings” which means that every passenger entraining or detraining is counted at that station. This is not an accurate representation of the actual number of people riding because that passenger would again be counted at their origin station for an alighting passenger and at their destination station for a boarding passenger. Therefore, to compensate, Empire Builder ridership used in the critique of the North Coast Hiawatha study is 50% of stated Amtrak station ridership, rounded up to the nearest whole person. For example: 2019 ridership at Wolf Point was 4,603; to make that comparable with figure on the NCH study, the ridership will be considered as being 2,302, half the indicated figure, rounded up to the nearest whole number.

King County (Stations Seattle, Auburn), Population 2,252,782, ridership: 30,406:

The population figure ignores that the Seattle/Tacoma metro area has a population of over 4 million. The Auburn stop especially would attract a large number of patrons from Pierce County (Tacoma), but the entire metro area should be considered. As an endpoint, Seattle should generate a large amount of patronage, but ignoring much of the population artificially reduces projected ridership. With only 30,406 riders annually, the study projects on average only 83 passengers daily departing Auburn on an eastward trip (assuming no Seattle-Auburn trips). For the Seattle metro area, the Empire Builder logged 42,407 passengers at its Seattle, Edmonds, and Everett stops in 2019 or 139% of the patronage projected that Seattle and Auburn for the North Coast Hiawatha.

Kittitas County (Stations: Cle Elum, Ellensburg), Population 47,935, ridership: 9,709. With a population of just over 2% of the understated population for the Seattle and Auburn stops combined, the study projects ridership to be 32% of King County.

Yakima County (Stations: Yakima, Toppenish), Population 250,873, ridership: 13,852. While the population of Yakima County is 523% that of Kittitas County, ridership is projected to be only 143% thereof. Projected ridership in Yakima County is similar to Empire Builder ridership in Ward County, North Dakota (Minot) at 13,641, but Yakima County has 359% of the population of Ward County.

Benton County (Station: Prosser), Population 204,390, ridership: 11,285. Ridership for the Prosser stop is unrealistic. Prosser has only about 13,000 people in the immediate area, and is at the far west-end of the county. Most of the population resides in Richland and Kennewick at the far east-end of the county, and would use the Pasco stop or more likely, a new stop in Kennewick. The study does seem to take this (an anomaly in the study) into account as ridership in Franklin County is higher than Benton county with its much-higher population.

Franklin County (Stations: Pasco, Connell), Population 95,222, ridership 19,288.

If the projected ridership in Benton and Franklin Counties (the Tri-Cities) is added together, the 30,573 figure is greater than that projected at Seattle/Auburn (30,406). Not a likely scenario.

Adams County (Station: Ritzville), Population 20,220, ridership 4,096. Ridership for this county is in line with similar populations in the study, but actual ridership at Ritzville would be only a fraction of the projection, since most of the county would use the Connell stop. Connell is much closer to the center of population of Adams County, Othello, where the area population is about 80% of the entire county. It’s only 24 miles from Othello to Connell, but 53 miles from Othello to Ritzville. Evidently the RPA model doesn’t take population distribution within the county into consideration, but it is obvious the only reason for a “stop” is to capture that county’s population. In theory, the model “still works” because it captures the population one way or the other IF the adjacent counties have a stop (if they don’t, like Pierce County, Washington, Kootenai County, Idaho, and Clay, Strearns, Hennepin, Carver, and Washington counties in Minnesota, the directly available population is not captured). But then this raises the question of why even designate specific station stops if county population is the primary driver?

Spokane County: (Stations: Cheney, Spokane), Population 522,798, ridership 7,056. Again, the one-county-only model ignores that the Spokane-Coeur d’Alene metro area has a population well over 700,000. Still, even with the model’s population base of well over 500,000, usage at Spokane (and Cheney) is projected to be fantastically low. Put another way: While the population of Adams County is only about 4% that of Spokane County, the projected ridership in Adams County is at 58% that of Spokane County. Also, the model projects NCH ridership to be only 30% that of current (2019) Empire Builder ridership. While ridership is projected to be relatively robust at the stations between Cle Elum and Pasco inclusive, ridership beyond Pasco in Washington State is anemic, and suggests that any “East-West” service train operate only Seattle to Pasco.

Bonner County: (Station: Sandpoint), Population 45,739, ridership 9,265. The study projects ridership at Sandpoint to be 131% that of Spokane-Cheney, with a stated population base of only 9% that of Spokane! The model projects NCH ridership to be 275% that of the Empire Builder at this same station. While this is obviously unlikely, it does show a difference between a projection and reality. Empire Builder ridership at Sandpoint (2019) was only 12% that of Whitefish though the communities share similar demographics and that they’re major tourist destinations which include a ski area. But one of the reasons for the anemic ridership at Sandpoint is scheduled train times – around midnight and 230 AM versus 900 PM and 730 AM at Whitefish. Arrival times dramatically affect usage, but such a consideration was not included for the North Coast Hiawatha ridership at Sandpoint or anywhere else; just an example of how the model can over-represent patronage if operational realities are not part of the considered input.

Sanders County: (Stations: Thompson Falls, Plains), Population 12,113, ridership 2,454. Ridership appears to be similar to other counties with a like population.

Lake County: (Station: Arlee), Population 30,458, ridership 6,169. Ridership appears to be similar to other counties with a like population, but Ravalli or Dixon would be better stations to serve most of Lake County’s population. (Ravalli is closer to most of the population; Dixon would be nearly equally so, and has existing track structure that would reduce station costs.)

Missoula County: (Station: Missoula), Population 119,600, ridership 6,604. The birthplace of the Big Sky Passenger Rail Authority, Missoula county’s projected ridership compared to the population is one of the worst on the route. Again, not all population was considered, notably Ravalli County (44,000 residents) where 92% of all travel outside the county is via Missoula County (much more in the winter). Yet, Missoula County with 393% of the population of Lake County is projected to have only 107% of the ridership. Broken down on a city-to-city comparison, Arlee has .87% (as in less than 1 percent) of the population of Missoula, but its projected ridership is 93% that of Missoula!

Granite County: (Station: Drummond), Population 3,379, ridership 684. Ridership appears to be similar to other counties with a like population.

Powell County: Actually, Powell County is not indicated in the study, which is odd because it’s a BSPRA member. Since the creation of the railroad here, passenger trains operating on this route via Helena have stopped at Garrison to serve patrons in Powell County (county seat Deer Lodge), Anaconda-Deer Lodge County, and Butte-Silver Bow County. Exclusion of a Powell county station fails to tap significant ridership potential.

Lewis and Clark County: (Station: Helena), Population 69,432, ridership 14,064.

The ridership is more in line with area population, though that, too is understated considering Jefferson and Broadwater counties are part of the metro area. For whatever reason, Helena is projected to have 212% of the ridership of Missoula, but has only 58% of the stated population. (Again, the population of both the Helena and Missoula metro areas is understated, but Missoula more so.)

Broadwater County: (Station: Townsend), Population 6,237, ridership 1,263. Ridership appears to be similar to other counties with a like population.

Gallatin County: (Station: Bozeman), Population 114,434, ridership 6,318. Worst-performing ridership station along the NCH route by far, especially considering it’s the gateway to Yellowstone National Park and the Big Sky Resort complex, is the home of the largest university in Montana, and is the largest tourism-oriented county in the state of Montana. Bozeman is projected to have only 79% of the ridership Amtrak currently has at East Glacier Park/Browning, one of the stops for Glacier National Park along the Empire Builder route (and is largely seasonal). The combined ridership projected at Bozeman and Livingston (9,682) as the primary stops for Yellowstone National Park is but 81% that of the combined actual 2019 ridership at Glacier National Park stops of East Glacier, Essex, and West Glacier (11,987). Gallatin County has 3.5 times the population of Stark County in North Dakota (Dickinson) – which has almost no tourism draw – but is projected to have slightly LESS ridership.

The Bozeman airport (at Belgrade) is Montana’s busiest. It serves Yellowstone National Park (along with other airports like West Yellowstone, Cody, Billings, and Jackson), and has near exclusivity serving the Big Sky Resort complex. The NCH study projects the proposed train will produce only .8 percent (less than 1%) of the travel than does the Bozeman airport. This is in contrast to the Empire Builder at Whitefish, which with its two daily trains (one east and one west) attracts 8% that of the ridership at Glacier Park International airport southeast of Whitefish. In other words, comparing the Kalispell(Whitefish) and Belgrade(Bozeman) airports, the Empire Builder at Whitefish currently attracts ten times the patronage than the study projects would occur at Bozeman compared to airline boardings. (Amtrak ridership at Whitefish is for Whitefish only and does not include West Glacier, Essex, or East Glacier.) Considering passenger train ridership alone, Bozeman is projected to generate only 23% of the patronage of the existing Whitefish stop.

Park County: (Station: Livingston), Population 16,606, ridership 3,364. Ridership appears not to be affected positively by its proximity to Yellowstone National Park. (With no regular public transportation between Livingston and the Park, this is likely true.) The population of Park County is almost identical to that of Hill County, Montana (County Seat: Havre), a county that receives relatively little economic impact from tourism; but 2019 Empire Builder ridership in Havre is 152% of the projected NCH ridership at Livingston.

Sweetgrass County: (Station: Big Timber), Population 3,737, ridership 757. Ridership seems to be in line with population. The RPA database incorrectly identifies the county name, it’s Sweet Grass – two words, not one. Not a big deal – unless you’re in Sweet Grass County. Montanans know that Sweetgrass is a community on Interstate 15 at the Alberta border.

Stillwater County: (Station: Columbus), Population 9,642, ridership 1,953. Ridership seems to be in line with population in the study, but still understated. For example, ridership in at Columbus stop is projected to be only be 92% of the existing Malta stop in Phillips County, but Stillwater County has 236% of the population of Phillips County, and would logically draw significant patronage from adjacent Carbon County.

Yellowstone County: (Station: Billings), Population 161,300, ridership 8,906. Another underperforming station for ridership. With 232% of the population of Lewis and Clark County, Yellowstone County’s projected ridership is but 63%. In fact, Billings has projected ridership only 46% of “Mandan/Bismarck” though its metro population stands at 143%. And with 981% of the population of Hill County, the Billings stop in Yellowstone County is projected to have but 174% of the ridership at the current (2019) Havre stop.

Treasure County: (Station: Hysham), Population 696, ridership 141. Ridership in line with population.

Rosebud County: (Station: Forsyth), Population 8,937, ridership 1,745. Ridership in line with population.

Custer County: (Station: Miles City), Population 11,402, ridership 2,310. Ridership in line with population.

Prairie County: (Station: Terry), Population 1,077, ridership 218. Ridership in line with population.

Dawson County: (Station: Glendive), Population 8,613, ridership 1,745. Ridership in line with population.

Wibaux County: (Station: Wibaux), Population: 969, ridership 196. Ridership in line with population. Note: The combined populations of the counties east of Yellowstone County in Montana (Treasure, Rosebud, Custer, Prairie, Dawson, and Wibaux) is 31,694. Projected NCH ridership in these six counties is 6,355. Contrast this with three counties directly north along the Empire Builder route in Montana: Phillips, Valley, and Roosevelt with station stops of Malta, Glasgow, and Wolf Point respectively. Together, these counties have a combined population of 22,242, and a combined ridership at Empire Builder station stops of 6,810. The RPA study is projecting that while counties along the NCH route in Eastern Montana have 142% of the population of the three easternmost Montana counties along the Empire Builder route, NCH ridership would only be 93% thereof. For whatever reason, the RPA study projects the NCH to have much less utility in Eastern Montana than the current Amtrak service.

Billings County: (Station: Medora), Population 923, ridership 188. Ridership in line with population, but it shouldn’t be. Medora is a “political stop” as the community has shown interest in the passenger train due to its being a tourist destination (by North Dakota standards) with its proximity to Theodore Roosevelt National Park. Yet, the ridership doesn’t reflect this, and it’s about the same as Wibaux County, Montana with a similar population. Again, the RPA model is evidently unable to acknowledge stops with tourism potential and project ridership accordingly.

Stark County: (Station: Dickinson), Population 31,489, ridership 6,378. Ridership in line with population in this study, but as far as demographics (and referencing the Bakken Oil Boom), Stark County (Dickinson) is very similar to Williams County (Williston) along the Empire Builder route. Yet, the study projects ridership at Dickinson to be only 43% that of Williston. (Clearly the model is not reading actual travel patterns correctly – and Williston has far superior air service).

Burleigh County: (Station: Mandan/Bismarck), Population 95,626, ridership 19,369. It’s curious that the station is “Mandan/Bismarck” when Mandan is in Morton County and Bismarck in Burleigh County. In this case, choosing Burleigh as the “study” county is good, was it has much more population, but overall, the population of the Bismarck metro area is understated (i.e. Morton County not considered). Interestingly, though the study indicates that “Mandan-Bismarck” has about half the population of Fargo, it predicts twice the ridership.

Stutsman County: (Station: Jamestown), Population 20,704, ridership 4,194. Ridership in line with population.

Barnes County: (Station: Valley City), Population 10,415, ridership 2,110. Ridership in line with population.

Cass County: (Station: Fargo), Population 181,923, ridership 10,045. Ridership in line with the population, but the population is understated. Fargo-Moorhead metro area has 245,000 people, or about 121% the population of the St. Cloud metro area, but the study projects ridership at Fargo to be only 51% that of St. Cloud.

Becker County: (Station: Detroit Lakes), Population 34,423, ridership 6,973. Ridership is in line with population, but is 297% that of the Empire Builder at the same station, even though 6 of the top 10 stations for ridership at Detroit Lakes would be served by both trains….

Wadena County: (Station: Staples), Population 13,682, ridership 2,771. Unlike Detroit Lakes, projected ridership is only 92% of the Empire Builder. Again, not all area population is considered, and in fact, the RPA model places the Staples station in the wrong county; it’s actually in Todd (population 25,262). In operational reality, the Staples stop for the Empire Builder is actually busier than the Detroit Lakes stop (contrary to the North Coast Hiawatha RPA projection); this is because Staples attracts riders from the Brainerd micropolitan area and its 96,000 residents.

Sherburne County: (Station: St. Cloud), Population 97,238, ridership 19,696. Projected ridership is a whopping 430% that of the Empire Builder. Again, the model has incorrectly identified the county where the St. Cloud station is: Benton County; In reality, most of the 203,000 people in the St. Cloud metro area live just across the Mississippi River from the station in Stearns County. An inconsequential number of the residents reside in Sherburne County, which is probably indicative of why the ridership projection is so skewed.

Ramsey County: (Station: St. Paul), Population 560,321, ridership 30,386. Another misrepresentation of the nearly 3.7 million people in the Minneapolis/St. Paul metro area. Moreover, this shows that the study just “rubber-stamped” Empire Builder stops from Fargo and east; instead, it should have added a stop at somewhere like Coon Rapids (where a Northstar commuter rail station exists) to include much of the Minneapolis population. Not surprisingly, the RPA study projects ridership at St. Paul to be a dismal 67% that of the Empire Builder.

From St. Paul to Chicago inclusive, the station stops will be shown together to indicate how erratic the NCH ridership projections are and how most do not coincide with actual travel patterns for the Empire Builder.

The majority of the patronage between Chicago and St. Paul/Minneapolis is local travel. Comparing existing Empire Builder ridership with proposed North Coast Hiawatha ridership in this segment:

  1. NCH ridership at St. Paul is 67% that of the EB.
  2. NCH ridership at Red Wing is 236% that of the EB.
  3. NCH ridership at Winona is 155% that of the EB.
  4. NCH ridership at La Crosse is 51% that of the EB.
  5. NCH ridership at Tomah is 164% that of the EB.
  6. NCH ridership at Wisconsin Dells is 73% that of the EB.
  7. NCH ridership at Columbus/Portage is 119% that of the EB.
  8. NCH ridership at Milwaukee is 318% that of the EB.
  9. NCH ridership at Chicago/Glenview is 94% that of the EB.

(And will mention again that ridership at Detroit Lakes for the NCH was 297% that of the EB and at St. Cloud 430% of the EB)

Given that the primary destinations for travel between St. Paul and Chicago are stations between St. Paul and Chicago inclusive, the projected ridership for the North Coast Hiawatha should be similar to that of the Empire Builder, but only at Chicago-Glenview and Columbus-Portage is this the case. In all the other cases, the North Coast Hiawatha is significantly more or significantly less, with ridership figures for Milwaukee and Red Wing being off-the-charts unexplainable. While one would not expect the NCH ridership projection for these stations to be the same or even similar than the Empire Builder, the amount more or less compared to other stations in this segment should be proportionally the same and it’s not – it varies widely.

As an aside, the RPA study incorrectly places the station at Wisconsin Dells in Juneau County. It’s really in Columbia County – that’s three stations that whatever database this program is using isn’t even able to correctly identify a known geographic location.

The Big Sky Passenger Rail Authority acknowledges the importance of the Empire Builder to Northern Montana, and in doing so also mentions that it does not serve the majority of Montana, most of whom reside in the Southern part of the state. Logically, one would then expect the study to then reflect ridership associated with that much-greater population. To expose some interesting aspects of the study in this regard, here’s a synopsis of state ridership for both the NCH and EB (2019).

Projected NCH ridership and current EB ridership by state:




























Stations served by the EB: 46, but only 45 at any one time. Five stations on the Portland section of the train, giving a total of 40 stations from Chicago to Seattle inclusive. Stations on the Portland section of the Empire Builder (Pasco, Wishram, Bingen, Vancouver, and Portland) are not included in ridership, though ridership Spokane and east include riders which may be using the Portland section of the Empire Builder west of Spokane.

Number of proposed stations for the NCH in the study: 51

As stated above, ridership at many stations in the NCH proposal range from fantastically large (like St. Cloud, MN and Milwaukee, WI) to unusually small (like Spokane, WA). The North Coast Hiawatha projection is for 45,548 more passengers annually than for the 2019 Empire Builder, but 94% of the advantage for the NCH occurs in just three states: Wisconsin, Minnesota, and Idaho. The remainder would be in Washington State, but is largely offset by the Empire Builder with more riders in Illinois, North Dakota, and Montana. Yes, Montana. The study projected that a revived North Coast Hiawatha would serve 1,720 FEWER patrons in Montana than actual ridership on the Empire Builder. It’s not a surprise considering that, other than the very strange ridership spike at Arlee and Helena, the remaining major Montana NCH stops vastly underperform (and as have been explained in earlier text in this document); Based on Montana station patronage alone, the Empire Builder serves 90 passengers per route mile in Montana annually, while the North Coast Hiawatha – according to the study – would serve only 76. (The NCH route – if operated via Dixon – is 103 miles further crossing the state.) In fact, of the seven states that would be served by the NCH, the model clearly states that the least amount of utility (determined by station usage only) for the train is within Montana. (North Dakota is 115.5 passengers annually per route mile, and Idaho is 106.5, with all others much more.) The train has many more route miles in Montana than any of the other six states.

While the NCH projects exceptionally poor ridership at major communities like Billings, Bozeman, and Missoula which are likely the main contributors to Montana ridership being below actual numbers registered by the Empire Builder in Montana, the difference is likely even greater than indicated. Much like poor Empire Builder ridership (relative to other stations) at places like Spokane, Sandpoint, and Fargo due to middle-of-the-night departure times, actual Amtrak trains are subject to reduced ridership due to service interruptions. In the case of the Empire Builder specifically, it routinely loses many thousands of riders each year due to service interruptions – mostly due to weather, but sometimes due to derailments or other events such as recurring annual service interruptions to one its major connecting trains, the Coast Starlight. Were the North Coast Hiawatha an actual operating train, its ridership too, would be subject to reductions due to service interruptions; but as a simply a computer model, its ridership is only affected by input variables, not by operational reality. In this case, some of its ridership is probably overstated, but in too many cases, the lack of proper consideration of population understates potential ridership. And apparently nowhere – in locations such as Bozeman. Missoula, and Medora – is tourism a component of the study. Or, at least the model doesn’t show it.

For your consideration.

Mark Meyer

(not representing any passenger train advocacy group or any other entity)