Sanity-Checking My HSR Ridership Model

In previous posts about modeling high-speed rail ridership, I used a gravity model for the estimation. While poking around with spreadsheets, I figured out that a good way to sanity-check the model is to run it on existing high-speed rail systems with known ridership. It turns out that the model fits the data decently but not amazingly, and tends to overestimate ridership at long distances (800 km+) and underestimate it at short ones.

The model

The model I use is a gravity model, with the constant trained on Shinkansen data from JR East and JR Central (PDF-p. 4):

\mbox{Ridership } = 75000\cdot\mbox{Pop}_{A}^{0.8}\cdot\mbox{Pop}_{B}^{0.8}/\max\{\mbox{distance}, 500\}^{2}

The populations of metro areas A and B are in millions, distance is in km, and ridership is in millions per year in both directions combined.

He blocks the door and doesn’t move to let people on or off. Credit: Barroa Artworks.

The data

I’ve tested the model on two datasets: Shinkansen, and Taiwan HSR. These are island systems with a finite, controllable number of stations; Taiwan, a single-line system, is especially easy to model. The km-points are taken from line lengths; but mini-Shinkansen lines have artificially inflated lengths to account for the greater travel time, by a factor of about 2.7, to be compatible with an average express train speed of about 220 km/h. This means the model will overrate their passenger-km, but it’s not a significant source of error as they are fairly small cities – were they bigger they’d get full Shinkansen.

Metro areas are combined, and when a metro area has several stations, they are merged and only the most prominent is depicted, such as Tokyo, Shin-Osaka, and Taipei.. In Japan I use the broader category of major metropolitan area wherever possible, with the exception of Shizuoka-Hamamatsu, which are not merged as they were distinct until recently and remain two separate city cores that only share suburbs on the margins. Otherwise I use the smaller metropolitan employment area, as the MMA is only defined for the largest cities, and not for (say) Aomori or Kanazawa.

In Taiwan there’s no real definition of metro area. The secondary cities are single-tier municipalities encompassing the metro area plus some rural areas; I take what Wikipedia calls the urban part, which is nearly the same as the municipality. Taipei and New Taipei are merged – there’s a stop in New Taipei but New Taipei is really a suburb of Taipei spreading in all directions; but Taoyuan is kept separate, as it tries to develop its own core and lies only in one direction from Taipei, to its west. Outside the cities I use county populations where the stop seems to serve the center of the county, but Chiayi is expansive and I focus on the independent Chiayi City plus the suburb the station is in, and Changhua’s station is very peripheral to the county, most of which is closer to Taichung.

Both countries charge similar fares – Wikipedia has Taiwan charging, in PPP terms, $0.25/p-km, which is close to the Shinkansen average, and compares with about $0.15/p-km in Continental Europe. In addition, both have linear population distribution, Japan along the Taiheiyo Belt and Taiwan along the west coast.


The model massively underrates the ridership of THSR. It believes ridership is 26 million a year, with a total of 4.465 billion p-km; the actual numbers are 67 million and 12 billion respectively as of 2019, per Wikipedia. I have not seen ridership by city pairs, only boardings per station. The numbers do not make it obvious if there is more very short-distance ridership than I expect. The average trip length I predict is 172 km; the actual average is 178. Taichung has slightly more ridership than Zuoying, where in reality Taichung and Kaohsiung have the same populations, but Zuoying is not quite at city center whereas Taichung also draws from Changhua County, whereas the Changhua station has very low ridership. Overall, to the extent the shape of the model is correct, the minimum of 500 km in the denominator cannot be too wrong – or, if it is, the minimum must be more than the Taipei-Kaohsiung distance of 339 km or not much less than it.

In Japan, the situation is less clear. Total Shinkansen ridership is 438 million as of financial year 2018-9, per Wikipedia; this is the last year before corona, as the years end on 3-31 and in March of 2020 Japanese ridership was already suppressed due to social distancing. Passenger-km on JR East, JR Central, and JR West totaled around 100 billion, with Hokkaido and Kyushu adding scant numbers, but these are railroad-km, and the Shinkansen charges based on the distance along the legacy line and not the Shinkansen, inflating p-km by somewhat less than 10%.

In contrast, my model thinks total Shinkansen ridership is 389 million and p-km sum to 170.815 billion. The 389 vs. 438 discrepancy is easy to explain – my model ignores intra-metropolitan trips, and we know that they exist because there are some Shinkansen commuters in towns like Mishima. However, 100 vs. 171 billion p-km is harder. For this, there are several explanations, all plausible, and yet none completely satisfactory:

  • About 40 billion of the p-km involve riding through Tokyo, of which 21 billion are from the Tohoku Shinkansen and 19 from the Joetsu and Hokuriku Shinkansen. There are no through-trains, and the through-trips via Joetsu and especially Hokuriku are circuitous.
  • Yamagata and Akita between them generate around 6 billion p-km per the model; this is an overestimate, as the spreadsheet does not distinguish km that are really stand-ins for trip time from km that are actually traveled.
  • A total of 6.5 billion p-km per the model are diagonal between the Tohoku, Joetsu, and Hokuriku Shinkansen; in reality, connecting at Omiya or Takasaki is so circuitous that I expect nearly everyone drives.
  • Inter-island trips are especially likely to be done by air. Tokyo-Fukuoka has a rail-air modal split of 7.4-92.6, over a distance of 5 hours, and Nagoya-Fukuoka is only 51-49, over a distance of 3:20. This is bad for rail by European standards, where 5 hours is typically 20-30% for rail and 3:20 is a clear majority, and even by intra-Honshu Japanese standards, where Tokyo-Hiroshima at 3:55 is 68-32 and Tokyo-Okayama at 3:15 is 70-30.

All trip categories above are disproportionately long, helping explain why the model underpredicts ridership while overpredicting p-km. Subtracting all of the above one gets to not much more than 100 billion.

The model does nail certain aspects of Shinkansen ridership. Tokyo-Sendai, Tokyo-Hiroshima, and Tokyo-Okayama are easy – the model was trained in part on those specific city pairs. But in adition, overall ridership out of Tokyo and Osaka is very close to total JR Central ridership in these two regions. The model slightly overpredicts Osaka but that is expected since it lumps the Keihanshin region together whereas JR Central would not count Kobe.

Nagoya is more overpredicted, and it is possible that it is uniquely auto-oriented and this slightly reduces rail ridership, by maybe 25% below modeled prediction. If that is what is happening, then the constant 500 in the denominator of the model as well as 75,000 in the numerator should be adjusted – the reason for the choice of 500 is that Tokyo-Nagoya and Tokyo-Osaka ridership levels both follow the same model if the exponent is 0.8 and distance is ignored; if in fact Nagoya has a 25% malus then to countermand it the constant in the maximum should be lowered slightly, to 430 or a little less.

It’s tempting to rewrite the model in terms of travel time and then set the constant at 2 hours (and not 2.5 hours as I did when trying to model Germany). But note that it’s far from enough to explain the model’s gross underprediction of Taiwanese HSR ridership, an underprediction that exists across all distances in Taiwan. Nor is it possible to lower the 75,000 constant in the numerator and address any of the underprediction of Taiwan.


  1. Sassy

    > Tokyo-Sendai, Tokyo-Hiroshima, and Tokyo-Okayama are easy – the model was trained in part on those specific city pairs. But in adition, overall ridership out of Tokyo and Osaka is very close to total JR Central ridership in these two regions.

    I wonder if a focus on trips with Tokyo on one end misses something like “destination centralization” on a larger scale.

    It might also explain Taiwan, since Taipei (depending on how you want to draw the metro area borders) is more primate of a city than Tokyo is.

    South Korea might not be an island, but in the context of intercity travel, it might as well be, and Seoul is also a very primate city. I wonder how the model works there.

    • Alon Levy

      KTX is a bit more open than THSR and the Shinkansen, and I don’t even have ridership by station, but I do have ridership by line. I poked around but without a spreadsheet, and KTX overperforms the model but not massively. Go here and look for the link called 기본통계표: 고속철도 여객 수송동향 (연, 2004 ~ 2020). Ridership on the Gyeongbu HSR line before corona was 36.004 million on the KTX mainline to Seoul Station and 17.096 on the SRT reverse-branch to Gangnam, connecting Seoul (26.2), Daejeon (1.5), Gumi-Gimcheon (0.6), Daegu (2.5), Gyeongju (0.3), Ulsan (1.1), and Busan (3.5). The model predicts 42.1 million trips between pairs of these cities; this nails the pre-SRT ridership, and underpredicts 2019 ridership by 21%.

      The question is what to do about diagonal trips, i.e. Honam-Gyeongbu. The system pretty clearly counts branch trips like Seoul-Pohang as Donghae trips and not Gyeongbu trips – from all above cities to Pohang, pop. 0.5 million, the predicted ridership is 3.78 million and the actual in 2019 was 5.566 million (actually down from 2018, I don’t know why). The sum total of ridership from Gyeongbu cities from Daegu north to Changwon and Jinju is predicted at 7.78 million, actually a bit over the 2019 actual of 7.028 million, but with so much traveling at low speed, the one-way trip time is 3 hours on Seoul-Changwon, at which point the model starts docking off ridership (then again, Seoul-Busan is usually around 2:30), and also the frequency is kinda meh.

      The general conclusion I draw from this is that I slightly underpredict Korea, just as I massively underpredict Taiwan, and also the 500 km bit in the denominator should be turned into a slightly lower number, maybe 400 km or 2 hours of trip time (and the KTX is on the slow side for HSR; SRT is faster).

  2. Phake Nick

    – For through traffic via Tokyo, while it probably make sense most passengers coming to or from Tokaido line wouldn’t use this option to reach Hokuriku, given that conventional express trains offer similar or better travel time with much lower fare, it is still the best option for travellers to Niigata or Tohoku region, considering how seamless the transfer process is. Railway still manage to obtain 62.9% modal share from Nagoya to Niigata and 56.6% from Nagoya to Miyazaki (Sendai) against airlines.
    — Such model share split is in itself quite interesting, because Nagoya to Niigata via Shinkansen is longer and is more circuitous than to Sendai, but probably because there are better flight options between Nagoya and Sendai that airlines still manage to gain higher share on the route than the Nagoya-Niigata route?
    – Have transfer traffic via conventional trains be considered? For example, despite Oita and Kochi have not been served by Shinkansen directly, the modal share of Osaka to Oita and Osaka to Kochi for railway are 67.0% and 39.9% respectively (only comparing between train and air flight) and most passengers who travel on such route would do the Osaka to Kokura or Osaka to Okayama leg via high speed trains.
    – For inter-island trip, performance of those routes might be impacted by Fukuoka’s convenient airport location and Hakodate’s station being faraway from city core.

    • Phake Nick

      In fact the model over-estimated inter-island traffic not just on long trip but also on shorter trip.
      For example, according to this data:
      In FY2018, South Hokkaido to Aomori have 290.4k outbound and 295.4k inbound passengers on JR, and South Hokkaido to Miyagi hav 84.4k outbound and 83.9k inbound passengers, with Shinkansen being the only mode offered by JR between these city pairs.
      While the Hakodate to Aomori passenger count significantly exceeded the model’s predicted 55k passenger per year, on Hakodate to Sendai it’s still significantly less than the model’s predicted ridership of 280k by half. This is despite the route have no direct air service and drivers need to take a connecting ferry to travel between the two islands
      Likewise, Hiroshima to Kumamoto have 107.0k outbound and 109.1k inbound passenger on JR, but the spreadsheet is predicting a 751k passengers between the two, which is an over-prediction by a factor of 3.5x

      • archie4oz

        The Hiroshima to Kumamoto probably over predicts because Hakata is probably the main destination when crossing into Kyushu.

        • Phake Nick

          Note that, the estimation for Aomori and Hiroshima in the original comment didn’t include traffic from/to Hachinohe or Fukuyama, which would be counted as part of Aomori prefecture/Hiroshima prefecture in the statistical figure.
          The spreadsheet is predicting another 347k passengers between Fukuyama and Kumamoto which make the estimation deviated even more greatly from the actual performance

    • Phake Nick

      Non-interisland trips seems to have some problem too? For example, Shizuoka prefecture total inbound and outbound passengers to Kyoto, Osaka, and Kobe via JR is about 3.4 million in FY2018, but the formula estimated total traffic between Shizuoka+Hanamatsu and Osaka on High Speed Rail to be 8.4 million per year.

      • Andrew in Ezo

        The formulas likely don’t take into account the cultural and business “spheres of influence” that effect intercity transport figures. For example, Shizuoka is considered eastern Japan culturally and as such in the Tokyo/Kanto sphere of influence, i.e. most business and leisure travel is Tokyo dominant as a destination. Hakodate has long relations with Aomori, but very weak ones with Sendai as they are completely different regions culturally. Hakodate is connected more with Sapporo and Tokyo.

      • borners

        Would Aichi-Shizuka underperformance this be the effect of the Nozomi service avoiding any Shizuoka/Eastern Aichi stops? I mean does the model doesn’t consider line capacity which is a problem for the Tokaido line, so JR Central has focused on the Big three stops. Of course JR Central is a horrible monopolist and doesn’t provide proper conventional expresses between Toyohashi and Atami just because they can afford to punish those on a budget (disclaimer I have taken those local trains several times over the last year). Add to that Shinkansen stations in Aichi have bad access to the No.2 employment centre which is Toyota city, no regular 1 seat connection (although that may change with Meitetsu chiryu upgrades). JR Central is choked on Nozomi profit they haven’t realised they kinda screwed themselves with how they made the Aichi Loop line a disappointment.

        Sorry I have grudges against JR Central, the best bad train operator in the world. Shizuoka should use the Maglev dispute to take the conventional Tokaido line away from them.

        • Andrew in Ezo

          Apparently JR Tokai is promising Shizuoka Prefecture more Hikari stops once the Chuo Maglev line is open. Don’t expect any faster services on the Tokaido Line zairaisen though. They want you taking the shinkansen for intercity trips longer than a hop to the next regional city- I know how inconvenient and uncomfortable it can be taking the longitudinal seat locals, having done the Atami-Toyohashi stretch several times on the Seishun 18 ticket. Notice how the service gets better once you reach Toyohashi, what with Meitetsu running parallel from there, more or less.

          • borners

            That’s why they just force JR Tokai to give up the whole damn Tokaido main line between Toyohashi and Atami. Like what happened to conventional lines in the Hokuriku/Tohoku shinkansen corridors. If you force Kasai Yoshiyuki* to chose between the neglected sewers of his Empire and getting to see the Maglev built before he dies he’s gonna choose the latter. Instead the silly governor of Shizuoka just wants hikari services and a station “near” the new airport they built in the middle of nowhere. Add in the potential for infill in the city centres, proper integration of the mini-private rail lines …..yeah I may have spent a few hours obsessing over this.

            *For those of you who don’t Japanese rail politics inside baseball Kasai is the eternal president of JR central (i.e. he’s “retired” but still in charge) whose control of the Tokaido money machine has made him the last great Rail Baron who dreams of a Maglev future and was instrumental into not only getting it approved but putting Abe back in power so he could approve it. Right now Maglev construction is severely delayed because they didn’t butter up Shizuoka prefecture earlier while planning to tunnel through its water sources.

            So I guess the lesson is separate legacy conventional operators from High-speed operators? I mean maybe Japan is a special case because of the guage difference making interoperability a pain?

          • Phake Nick

            In the Maglev controversy, the Shizuoka governor is presenting itself as representative of rural area against urban, using arguments like the coronavirus pandemic would nullify the need for people to travel quickly across urban area for meeting and thus money shouldn’t be spent on building new high speed line connecting urban areas together and instead funding should be directed to improve level of service in rural area outside major cities.

        • Phake Nick

          But then, things like Aomori prefecture to Fukushima prefecture in JR East’s territory is also overestimated.
          Traffic between Aomori + Hachinohe and Fukushima + Koriyama is estimated as 263k/year, but outbound + inbound passengers between Aomori prefecture and Fukushima prefecture on all JR lines combined is only 99k/year

          • borners

            I wonder if service pattern differences strike again. The Nasuno service only goes as far as Koriyama so its basically the commuter Shinkansen taking people from Oyama and Utsunomiya especially. Underperformance in the far North and overperformance closer to the centre is somewhat symmetrical. One might also wonder about the effect of the Omiya-Ueno bottleneck.

            Fukushima’s main towns are on my list of “should be trying out S-bahny style services on its branches” prefectures which is basically all of the non-megacity ones.

          • Alon Levy

            Yeah, the interesting thing is that my model nails Kanto-Aomori and Kanto-Iwate (Kanto = Tokyo + Kanagawa + Chiba + Saitama), neither of which factored into the model – I didn’t even start computing metro area populations north of Sendai until I made this spreadsheet. I guess Kanto-Shiga and Kanto-Mie underperform because the Shinkansen doesn’t stop there much – there’s Gifu-Hashima but it’s at a peripheral location.

          • Phake Nick

            What I am thinking about is that might be the model tend to overestimate traffic from/to smaller cities, especially traffic between city pairs that are both small

        • anonymouse observer

          What JR Central is doing with the conventional rail line makes business sense because none of conventional rail lines operated by JR Central is profitable even before this pandemic. If a for-profit organization operates the high-speed rail service in one of the largest metroplex of the world in terms of population and output, you cannot argue against what they do for profit unless they are subsidized for the conventional rail operations (none) or have commitment (none). Given the depopulation going on all over the Tokai region except in Nagoya and its suburb and decline of Hamamatsu (have you been to Kajimachi in Hamamatsu lately? It is sad.) as well as advanced motorization (home of Suzuki, Yamaha, Honda, etc.), I have no expectation for JR Central to improve the conventional rail service in Shizuoka region.

          Also, I imagine the majority of the complaints for missing rapid service in Shizuoka region are coming from Seishun 18 Kippu users, who are occasional, very small-budget travelers.

          Mikawa-Anjo has been helping non-Toyota factories around the area (Denso, Makita, etc.) and attracted businesses to open new and expand in the immediate surrounding area. It provided a way to quickly get to Nagoya so that passengers can transfer to Nozomi trains. Toyota? I do not think they care about the access from/to the Shinkansen that much because they have moved several core functions of the corporation which require a lot of business traveling to Nagoya and Tokyo already. For instance, their international department used to be based in Tokyo but moved to Mei-eki district in Nagoya when the Centair opened and then consolidated most of those satellite headquarters offices into the Midland Square:

          Meitetsu Chiryu upgrade would not change the situation significantly in terms of north-south through running capability because:
          – Mikawa Line portion will be still stub-end with Nagoya-bound connection (makes sense if they still have strong passenger flow from/to Nagoya)
          – Mikawa Line does not go anywhere close from Mikawa-Anjo Station (the nearest Meitetsu Station would be Minami-Anjo Station on Nishio Line)

          Again, with advanced motorization in the region, it would be difficult to make significant difference from the passenger rail improvements in the region, especially in terms of Shinkansen access, unless the improvements are massive/significant.

          • Phake Nick

            According to Toyo Keizai calculation, in normal years Tokaido Main Line have an operation coefficient of less than 100 so it should not be making losses. Although they have only computed the figure for the entire line and thus it does not show how much of this is contributed by area around Nagoya which presumably would have higher traffic density thus perform better than portions within Shizuoka.

          • borners

            First I was making two arguments. One was trying to explain why Shizuoka/Aichi underperforms Shinkansen given current population, Gunma and Tochigi are poorer than Shizuoka despite its recent decline.

            And yes I had a brainfart on the Mikawa Anjo and misunderstood whats going on in Chiryu. So you’re right on that. And Toyota city may not be headquarters but its still the largest and richest city in Aichi after Nagoya. For mode share it matters.

            My second argument was that JR Central is managing its assets poorly. Sure its never going to be Shinkansen or Tokyo commuter rail in terms of income or profitability. But its the central trunk of the national freight network and connects an urbanised region of nearly 4 million people. If Entetsu’s been able to keep going for this long then it should be fine. And for goodness sake the only reason the Maglev passes buck is because the Tokaido is at capacity so there are plenty of passengers.

            And “advanced motorisation” is a cop out. Its Shizuoka you’re crammed into a super narrow coastal plain that happens to also be congested because its got the spine of the national highway. Learn from Central Europe you can get higher mode share in poly-centric regions with better services, and if JR Central isn’t up to it then Shizuoka Prefecture should take it away from them. You don’t need any new ROW, infill here and there, a couple of overtaking platforms, through-running with Tenryu, Oikawa, Shimizu, Gakuen lines and at least build a goddam concourse to Entetsu. Also use the compact city style ordinances to encourage clustering around train stations. Most Kansai private rail companies would champing at the bit for so many assets.

            Also so the profit argument would work if JR Tokai weren’t the most profitable railway company on the face of earth and if it hadn’t wasted two decades not building a second Shinkansen with government support because management was high on Gadgetbahn. Compare with JR East’s hard work on the Mini-shinkansen and north-of-Morioka work and they managed to still do some conventional express services in and out of Sendai!

          • Phake Nick

            1. The reason why JR Tokai funding the Chuo Shinkansen in itself is exactly because they don’t want to wait behind the queue for government funding, which would have taken even longer times. And according to my understanding, nationally the discussion on an alleviation to the Tokaido Shinkansen was mainly centered around the construction of Hokuriku Shinkansen which have nothing to do with JR Tokai.
            2. Are you aware of the situation that JR East would have axed the through service of limited express on Joban line to Sendai according to plan in year 2011, which did not happens as planned only because the earthquake in that same year have shifted the national focus of Japan to the area’s recovery and thus it is desirable to keep direct service from Tokyo to coast of Fukushima?

          • borners

            1. That’s part of it. Though the super-normal profits that JR can use to invest have been around for two decades. Anyway re-fighting the Maglev vs Conventional argument is water under a bridge now. The backup Kansai-Kanto connection is a reason but the reality is that the Tokaido line is at capacity and super-profitable and the underperformance of Tokyo-Fukuoka suggests upping the speed with a better alignment/faster trains can steal market share from other transport modes too. I wonder if it also makes a Shikoku/Sanin mini-shinkansen possible. It might also make JR Central less cagey about having Tokyo station through service. I have no disagreements with building a second HSR between Osaka and Tokyo. But I do flutter between whether a more conventional set up would be better and going for maglev glory.

            Also a little more research on the Governor, who clearly dislikes railways! It took 4 years of the legislature bullying him to authorize a freight bypass in a factory town. Though I can understand the airport request its literally built on top a Shinkansen tunnel even if its not a very busy airport. He’s clearly in the Ishihara/Hashimoto mold of reactionary gasbag.

            2. I was actually referring to the rapid service that runs between Fukushima and Sendai. Yeah I heard about the discussion of cancelling the Hitachi service. And they should, Tokyo-Sendai services only happen 4 times a day. And the Soma coast was poor and sparsely populated before 2011. Its very different from Shizuoka. Add in it having a really overbuilt road system. And that service is probably a net negative for locals since between Iwaki and Iwanuma the Joban line is single track with few passing loops which means local trains have to wait sometimes more than an hour at Haranomachi (I have done this) because the Hitachi needs headways and freight has even fewer passing loops. Add in the higher operating costs of a limited express and the resulting higher fare…its really quite pointless. Just run local services between Iwaki and Sendai, with maybe a semi-express for Sendai commuters in the Northern half. I was there on a September weekend in pandemic year, it was actually quite busy and not just with school kids. Cheap and regular should be fine.

          • Tonami Playman

            @borners, Regarding using conventional HSR instead of Maglev for Chuo Shinkansen. I’ve always wondered this myself. The new alignment is super straight with minimal stops. JR Tokai could have built the tunnel as is and run conventional HSR at 360km/h. If we use the 303km/h average speed of the Beijing-Shanghai HSR as a reference, then the 438km Chuo Shinkansen alignment can be traversed in 87mins, 20mins more than the estimated time of 67mins with 505km/h Maglev. But using steel rails means the Chuo Shinkansen trains can though-run into the Sanyo Shinkansen towards Hakata reducing the trip time from the current 4:57 to 4:03. This is slower than a hypothetical time of 3:43 for Maglev plus Sanyo combo to Hakata. However this does not include the transfer penalty. If the Shin-Osaka station is sited deep underground like in Shinagawa and Nagoya, then we’ll be looking around 5-10mins for the transfer reducing the Maglev advantage to about 10mins out of a 4hr trip.

            Since Sanyo Shinkansen still has spare capacity, The through running Chuo Shinkansen trains could take over the Nozomi service between Tokyo and Hakata. I can see Maglev making sense in China further reducing trip times between Beijing – Shanghai and Beijing – Guangzhou, but with Japan it seems to that it’s more about the technology showpiece.

          • Phake Nick

            1.a. If done 20 years ago they might not have sufficient technology to tunnel through all those mountain range, and combined with the use of conventional Shinkansen technology, it would mean the speed up benefit from current Tokaido Shinkansen could be more limited than the Maglev and thus isn’t as worthy for the money to be invested.
            1.b. That Tokaido Shinkansen is at-capacity suggest it is not really meaningful for them to try gaining any additional market share from competing travel modes, since doing so would only further stress the line. So while strengthening Tokaido Shinkansen could indeed help capture more passengers travelling to Fukuoka, it wouldn’t be too beneficial to JR Tokai due to lack of extra capacity
            1.c. Fukuoka, Shikoku, Sanin, are all JR West/Shikoku’s territories. While JR Tokai will also be able to get through passengers, it is not something they can decide.
            1.d. Sanin Mini Shinkansen have always been proposed. The biggest obstacle is you have to stop traffic on the line between Sanin and Okayama for months in order to change the line’s gauge. And then due to the problem of Tokaido Shinkansen is at capacity and thus it wouldn’t be possible to allow through train operation from Mini Shinkansen lines into Tokaido Shinkansen, such Mini Shinkansen retrofit can only get them direct train service to no further than Osaka, and thus benefit is also relatively limited.
            1.e. To allow through operation of Tokaido Shinkansen trains to Tohoku, it would require engineering work at Tokyo station that is not possible in normal times, and can only be considered after the opening of Chuo Shinkansen relieving demand on Tokaido Shinkansen.
            2.a. JR East have just discontinued the rapid service between Fukushima and Sendai last month
            2.b. According to my understanding JR Freight no longer operate on most of Joban line?

          • Phake Nick

            @Tonami Playman According to my understanding, Chuo Shinkansen can have maximum gradient up to 40‰ thanks to the maglev technology, for conventional Shinkansen the maximum gradient allowed is only 15‰, although up to 20‰ is permissible for short distance and special permission have been granted to build train on rail Shinkansen with up to 35‰ gradient. But that still likely mean conventional Shinkansen might not be able to run the super straight route being constructed currently with Maglev performance in mind.

          • Herbert

            The Cologne Frankfurt line has a 4 % gradient as do some S-Bahn tunnels, so that gradient is not an absolute obstacle to wheel on rail technology

  3. adirondacker12800

    tends to overestimate ridership at long distances (800 km+) and underestimate it at short ones.
    Because it’s not taking door to door trip time in account like you said in your other posts? Passing through an airport takes the same amount of time whether you are catching a short flight or a long flight.
    Southwest pulled out of the Providence-Philadelphia market, the legacy carrier raised fares to “we are a monopoly again” and ridership on Amtrak nearly doubled. It’s more complicated than population numbers. But they are good to look at because no matter what you do there isn’t going to be a lot of ridership from Laramie Wyoming because there aren’t a lot of people there.

    • Herbert

      Virtually nobody lives at la defense, but it will still generate a lot of ridership for any transport that serves it…

          • adirondacker12800

            More people work in La Defense than live in micropolitan Laramie and metropolitan Cheyenne combined. I’d hazard a guess that the majority of them live in metro Paris.

    • James S

      I came to post about Providence and air competition as well. In my experience on Amtrak, a ton more people than I ever expect get off at Providence. Makes sense, flights to NYC arent a thing. But Boston DOES have a lot of NYC flights. Pre-COVID, there were 3 airlines offering flights hourly. The model must include competitors.

      In CA, I would expect very good rail ridership between LA and Bakersfield, as there are zero flight options, much higher than the population number would suggest.

      • Phake Nick

        I think there isn’t that much need to explicitly include competition, as barring cases where a place have no nearby airports for domestic flights, such as Kyoto or Northern Kanto, if the city is large enough and have an airport then airlines will automatically be attracted into a market, and depends on demand there might be a single airlines ran monopoly or more airlines would want to join reaching a competitive market with low air fare, which is indirectly a factor of how many demand exists in-between different markets, and thus should be model-able from demand factor without explicitly considering them, even though there could be fluctuation in air market due to joining and departing of various airlines in different times due to business model of each of them, but when speaking in term of time span of decades which is the time span you would be looking at for HSR investment, such fluctuation in competition is relatively transient and shouldn’t be taken into consideration.

      • adirondacker12800

        Looking at population is good place to get rough estimates. The estimates will be off because it involves people who make more than one decision. …… Penn Station New York is too crowded, no one wants to go there, according to some. No one is going to be stopping those people from flying and getting stuck in traffic in a cab. Or getting stuck in traffic along I-95 and it’s alternates between Maine and Virginia. I know people who drive to Florida a few times a year. I think they are nuts. Which is why I can get flights from Albany to multiple destinations in Florida.

        When I looked a few days ago there are no non-stops to Cleveland across New England-not-Logan or Upstate New York. If flying to Cleveland involves getting to Logan from Worcester a four hour train ride looks better. Providence, Bradley and Upstate were all connecting flights at 3:45-ish. That makes a four hour train ride look good too. But it could be a three hour train ride or less from Albany and points west. The places where a bus to the high speed rail station, along with the metro areas in Upstate New York that will have a station or stations is 5 million people. Or since there is no traffic in Upstate New York, driving to the high speed rail station.

        Boston-New York has a lot of flights because Acela service sucks and the Regional is even slower. DC-NY used to have a lot more flights. Sometimes ALB-DCA is reasonably priced. Sometimes it’s not. If I have to fly into BWI or IAD to get a low fare taking the train looks better. And sometime in the 23rd Century it might be possible to do it in two half hours by train. Which will kill off flights because it’s a bit over an hour to fly and it takes an hour to pass through an airport. And a lot longer to get out of one compared to a train station.

        ……. it also assumes air service is evenly distributed. Binghamton with a metro area smaller than Utica has air service. Utica doesn’t. Neither does Worcester which is bigger than both of them combined. Utica is likely to have high speed tracks right outside it’s ornate Union Station. Binghamton is unlikely to have rail service of any kind…… Population is a good place to get a rough estimate.

      • Phake Nick

        As for ridership between LA and Bakersfield, the distance between the two are quite a lot less than the floor distance of this formula, and also I feel like it should be more like “regional” type instead of “intercity” type of trip, that I don’t think the model for HSR could predict the demand accurately. Instead one should judge its demand using the principles of regional commuter railway, which the fare and schedule would become a very important factor

        • adirondacker12800

          Bakersfield is as far away from Los Angeles as Wilmington Delaware or Hartford Connecticut are from New York. Or Hartford is from Boston. Give or take a few.

  4. Herbert

    Would the model work better if instead of “metro areas” it were to use “x amount of time around a train station with hsr service” as the population basis?

    • Alon Levy

      I haven’t tried it! But it can get weird pretty fast:

      1. The values of X are different for origins and destinations. At this level of detail ridership isn’t perfectly symmetric.
      2. Looking at origins, different values of X have their own problems. X = 30 minutes means the Gare de Lyon commute shed doesn’t even cover all of Paris, let alone the suburbs; X = 1 hour is better for this, but then tends to bleed over in the presence of HSR, since sub-hour intercity trip times are common.

    • Henry Miller

      How can you calculate this? It seems really hard. I can figure that out for crows easily enough, but for ground bound humans it gets complex. Because of road crossing it often is the case that walkers can get much father in one direction than the other. Those same roads often mean in some directions cars can get a lot father than others (freeway, vs a highway with badly timed stop lights, vs residential streets). And transit might be running weird routes (which might or might not be sensible) where some people are outside that half hour zone while someone farther out the way is within it.

      Given the raw data and access to google maps API I can do this. (note that my access to Google maps is probably something not allowed in the license). However it isn’t easy.

      Which is probably why models don’t do well for short trips: local factors matter a lot more. At 600km none of the above matters as the speed of the HSR will make up for whatever time is lost getting to/from the train. However for shorter trips the alternatives start to look better quickly.

      If someone knows how to get better population within X time to some point in a city let me know. There are a lot of cities where it seems like HSR from the suburbs to the downtown(ie about 50km between stations) could be useful, and subsidize service to the next city. However I don’t know how to calculate this.

      • Phake Nick

        There are some software and APIs that can help you calculate, using public transport or other methods of mobility, how far can you get to within x minutes amount of time, from any given point in a city, and how much people/employment opportunity/etc would that cover. But a problem with those tools is that many of those require some specific form of open data be available for all transportation means in a city as well as a number of other open data for each cities, which in cases like Japan is usually not available

  5. John

    It’s worth noting that Taiwan’s HSR station locations are complete crap outside of Taipei and Kaohsiung–the intention was to develop new towns around the station sites, which has happened around Taoyuan and Hsinchu but not the other ones. But all the stations are very far from downtowns and only Taichung has a semi-decent transit option connecting the core to the HSR station. So THSR is leaving a lot of ridership on the table. (On the other hand, it could also mean that hand-wringing over station sites just doesn’t matter that much in the long run.)

    • Alon Levy

      Yeah, that’s the interesting thing about it. Taipei has a good location, but the majority of trips are not to or from Taipei Station, and ~40% aren’t to Nangang or Banqiao either.

      • Luke

        I’m not sure of the degree of influence of this factor, but Taiwan has relatively low rates of motorization. The Taiwanese MOT has the 2016 number at one roughly equal middle-income countries like Mexico, Turkey, and Hungary, 2/3 of Korea’s and western Europe’s, and only about half of Japan’s.

        • borners

          Maybe its to do with Kaohsiung-Taipei being just the right distance to make domestic flight uncompetitive? And Taipei is the only destination with an airport close to the CBD. Indeed eyeballing it looks like HSR stations are definitely better placed than airports. This points to airline share decline.

          Click to access 62%20Demand%20Adaptation%20towards%20New%20Transport%20Modes%20-%20Case%20of%20Taiwan%20HSR%20Trans.%20metrica%20-%20unformatted.pdf

          Bus share surprised me, are Taiwanese buses good?

          • Phake Nick

            Buses are indeed a main intercity transportation mean in Taiwan. I guess it is a natural consequence to countries like Taiwan and South Korea with extensively laid expressway but still rely on public transport?

          • barbarian2000

            This is supported by Latin America, which has seen a lot of investment in road transport, disinvestment in rail, and has lower car ownership than the US/Europe – And where buses are king.

          • Herbert

            To my knowledge the only domestic flights remaining in ROC are to/from outlying islands.

          • Jason

            For “bus” think Greyhound. For intercity travel, the buses are privately operated highway coaches. They are cheap, slow, and reasonably comfortable. It’s basically your only choice if you need to save money, which means you don’t own a car, and perhaps at most own a 50cc scooter unsuitable for highway travel.

          • John

            Legacy rail on Taiwan’s west coast is stuck in a no man’s land where it’s not cheap enough to compete with buses but also not nearly fast enough to compete with HSR. It’s also very poorly run by East Asian standards, with awful timetabling and frequent delays making transfers extremely unreliable. This in turn means that expresses are often forced to add stops by local demand, slowing them down even further. There’s also been a number of high-profile deadly accidents over the past few years, which has further destroyed its brand image.

            It’s very frustrating because a better-run regional rail would improve public transit by a LOT outside of Taipei, but at this point I don’t know how it could even be done.

          • John

            Kaohsiung’s regional rail tunnel is done, and apart from infill stations it has done absolutely nothing to improve service. It has actually made service worse for many passengers due to the extra stops. Governments have talked about “metro-izing” the TRA line in urban cores for more than a decade now, but the TRA still doesn’t have enough drivers or trains to run more than 2 local trains per hour in urban Kaohsiung and 3 in urban Taichung–and not on anything close to a clockface schedule, so for example in Kaohsiung northbound there’s a random 17-minute headway followed by a 38-minute one. It’s concrete-before-organization planning.

        • Phake Nick

          Many in Taiwan, including in larger cities like Taipei, also commonly use motorbikes as their main transportation method. But as motorbike isn’t suitable for long distance trip over expressway, I guess that could in turn help boost the use of HSR?

  6. Herbert this particularly bad article bemoans the state of railroading in Germany and seems to have identified a culprit: an unnamed cabal forces DB to build tunnels where above ground solutions would work. They even quote Vieregg Rössler…

    Thing is: the number one reason tunnels are dug where above ground solutions would work are NIMBYs…

  7. Pingback: Midwestern Urban Geography and High-Speed Rail | Pedestrian Observations
  8. Pingback: Make Marne-La-Vallée a European High Speed Rail Hub – Métro vélo dodo
  9. Pingback: Eastern European High-Speed Rail | Pedestrian Observations
  10. Pingback: More Sanity Checks on My High-Speed Rail Model | Pedestrian Observations

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.