Categorized climbs: new trends

[Danskebjerge]

In recent years, something has happened to the climbs of the Tour de France: The percentages of the slopes have gone up, while the length of the climbs has been decreasing.

That's one of the conclusions from a small study I made. The study involves 427 Tour climbs from the years 2006-2009 and 2012-2014.

In the first period, an average categorized climb had a steepness of 5.8 percent and a length of 9.4 kilometers. In 2012-2014, the steepness is 6.6 percent, and the length 7.8 kilometers. In a nutshell: The climbs are steeper, but shorter. And this development is seen in all of the five categories - from the HC category down to category 4.

Read the entire analysis at Danskebjerge.dk, and see diagrams with the results. (Sure, it's for nerds, I'm fully aware of that...) ;-)

 

[Libertine Seguros]

I think the trend towards shorter but steeper climbs is universal; the way the style of racing has changed has meant riders race much more conservatively, and the success of the USPS/Banesto/Sky template has meant that gaps are produced more by attrition than anything else on the classic lower gradient climbs at present. Therefore in order to increase attacking opportunities there has been a trend towards steeper climbs; the Tour still lags behind the Giro and Vuelta in terms of ultra-steep climbs, but the days of long tempo grinders producing fireworks are long since over and the race design has had to accommodate that.

 

[Linkinito]

Think it's a good place to post my own analysis I've done on Velowire eight months ago. I should update it with 2014 data (I'd have liked to get the 2006-2007 data).

I wanted to see if there was clear categorization bounds. I've based my curves on the SCC squared cotation criteria (L * G² with L being the length of the climb in kilometers and G the average gradient in %).

Therefore I've came up with these curves, with x being the distance of the climb in kilometers, and y the required average gradient to get in a particular category:

HC: y = 27/√(x) (729 points for the SCC)
1C: y = 20/√(x) (400 points for the SCC)
2C: y = 13.5/√(x) (182 points for the SCC)
3C: y = 9/√(x) (81 points for the SCC)
4C: y = 3/√(x) + 1 (between 16 and 30 points for the SCC depending on length)

Was pretty surprised with the results with only a few climbs being badly categorized.

 

[Eshnar]

Think it's a good place to post my own analysis I've done on Velowire eight months ago. I should update it with 2014 data (I'd have liked to get the 2006-2007 data).

I wanted to see if there was clear categorization bounds. I've based my curves on the SCC squared cotation criteria (L * G² with L being the length of the climb in kilometers and G the average gradient in %).

Therefore I've came up with these curves, with x being the distance of the climb in kilometers, and y the required average gradient to get in a particular category:

HC: y = 27/√(x) (729 points for the SCC)
1C: y = 20/√(x) (400 points for the SCC)
2C: y = 13.5/√(x) (182 points for the SCC)
3C: y = 9/√(x) (81 points for the SCC)
4C: y = 3/√(x) + 1 (between 16 and 30 points for the SCC depending on length)

Was pretty surprised with the results with only a few climbs being badly categorized.

That's a very good study. I guess the only weakness lies in the fact that the average gradient isn't sufficient to determine the difficulty of the climb. Take Angliru for istance. 9.5% average looks fairly hard, but it doesn't tell the whole story...

 

[murali]

wht abt max gradients and the position of the climb in the race?

 

[Eshnar]

wht abt max gradients and the position of the climb in the race?

That would help, surely. Also the altitude.

 

[Netserk]

Arcalis is quite the outlier :O

 

[Linkinito]

That would help, surely. Also the altitude.

Categorization is a combination of factors. Of course, distance and average gradient are the main criteria. Irregularity can help a climb to get a higher category: for example, Glandon south being HC for the first time last year.

A MTF can also benefit from a better categorization. For example, Pla d'Adet, HC since 1982 - 10.2 km @ 8.3 % - while similar climbs in difficulty like col de Menté (9.3 km @ 9.1 %) or col d'Agnes (10.2 km @ 8.1 %) are 1C.

Here's also a big problem.

Compare the col des Chevrères (3.5 km @ 9.5 % from Belfahy), the col de Menté (9.3 km @ 9.1 % from Saint-Béat), the col de la Colombière (16.3 km @ 6.8 %) and the Port d'Envalira (27.5 km @ 5 % from Andorra).

These 4 climbs are very different, but they are all 1st category climbs.

Basically, we are putting on the same level a 10-minute effort, and a 45+ minute effort. Envalira definitely takes more than an hour, it's probably one of the longest climbs alongside Galibier north.

Some people consider that Galibier from Briançon shouldn't be categorized as HC (22.8 km @ 4.8 %), but it tops at 2645 meters. Altitude is a definitive factor, even for climbs that are long but not that hard. And of course, there's the "prestige" of a climb, just take Aubisque east as another example (quite irregular 29.2 km @ 4.2 % with tons of flat parts). It's HC, though.

 

[Netserk]

Imho Petit Saint Bernard and Envalira are closer to HC status than Galibier is to cat 1. Given the high altitude (although not as high as Galibier) I'd argue the former two should've been HC.

 

[Linkinito]

Imho Petit Saint Bernard and Envalira are closer to HC status than Galibier is to cat 1. Given the high altitude (although not as high as Galibier) I'd argue the former two should've been HC.

I agree for Envalira, probably a little bit less for Petit Saint Bernard but is also HC-worthy. However i'd definitely put Allos north as HC - but I'll talk about it later on another thread...

 

[Krzysztof_O]

Very good chart. Altitude is also an important factor. Maybe if you add 0.1 point for each meter of col altitude over 1000 meters then these climbs could suit even better to their cathegories. As Eshnar pointed out, average gradient doesn't tell the whole story when the climb is not steady. I'm curious how it would look if the formula is used for each climb section separately (as in my ranking of 10 hardest GT climbs: http://forum.cyclingnews.com/showthread.php?p=1476624#post1476624).

 

[Cannavaro]

Think it's a good place to post my own analysis I've done on Velowire eight months ago. I should update it with 2014 data (I'd have liked to get the 2006-2007 data).

I wanted to see if there was clear categorization bounds. I've based my curves on the SCC squared cotation criteria (L * G² with L being the length of the climb in kilometers and G the average gradient in %).

Therefore I've came up with these curves, with x being the distance of the climb in kilometers, and y the required average gradient to get in a particular category:

HC: y = 27/√(x) (729 points for the SCC)
1C: y = 20/√(x) (400 points for the SCC)
2C: y = 13.5/√(x) (182 points for the SCC)
3C: y = 9/√(x) (81 points for the SCC)
4C: y = 3/√(x) + 1 (between 16 and 30 points for the SCC depending on length)

Was pretty surprised with the results with only a few climbs being badly categorized.

I also use L*G^2 (original G*height difference) to get a rough idea how difficult a climb is. My impression is that one could use simpler bounds:

HC: 800 points
1C: 400 points
2C: 200 points
3C: 100 points
4C: 50 points

Of course that bounds are not that accurate as yours, but you have the nice interpretation that one categorie higher means that the climb needs to be twice as hard. By the way, i think they never raced a climb which would score 1600 points which would require a new categorie.

 

[Krzysztof_O]

I also use L*G^2 (original G*height difference) to get a rough idea how difficult a climb is. My impression is that one could use simpler bounds:

HC: 800 points
1C: 400 points
2C: 200 points
3C: 100 points
4C: 50 points

Of course that bounds are not that accurate as yours, but you have the nice interpretation that one categorie higher means that the climb needs to be twice as hard. By the way, i think they never raced a climb which would score 1600 points which would require a new categorie.

At the TdF they never raced 1600 pts climb but at Giro it could be a different story. Finestre is about 1550 but if you add any altitude bonus it could exceed 1600. And if climbs subsections are calculated separately scores could be higher for irregular climbs. In top 10 hardest GT climbs there were a few 1600+ climbs (16+ in my calculation) but with above additional assumptions. But there will always be some HC freaks like Andore-Arcalis. And some climbs (like Colombiere) should be HC.

 

[Danskebjerge]

Maximum percentages have little influence

There is no evidence that the maximum slope has a significant influence on the way the climbs of the Tour de France are categorized. If that was the case, there would more climbs deviating from the expected intervals.

This may have something to do with the nature of the French climbs. Few climbs are both long and extraordinarily steep.

 

[Krzysztof_O]

I prepared ranking of all HC and 1 cat. climbs and some of 2 cat. climbs of the last 10 years (2005-2014) of Le Tour. I sorted them according to my formula, which takes into account slope variations (climb division into sections) as well as col altitude. In parantheses numbers based on Linkinito's formula are given. In these rankings the hardest and easiest climbs of each cathegory are presented. Some interesting freaks are seen in these ranks.

HC:

Mt Ventoux 14.1 (1218)
Agnel 14.1 (946)
Croix de Fer 13.4 (1066)
Bonette 13.2 (1057)
Madeleine 13 (1186)
...
Annecy-Semnoz 8.5 (773)
Galibier (from Briancon) 8.4 (601)
Pla d'Adet 8.1 (703)
Iseran (from Val d'Isere) 7 (540)
Andore-Arcalis 7 (534)

According to this ranking hardest TdF climbs of last decade are Mont Ventoux and...italian Agnel. According to Linkinito's formula it's Mont Ventoux and Madeleine. I think there are no doubt that mentioned climbs are kings of HC ascents. My formula favours Agnel because of its altitude and very hard final section while Linkinito's formula favours Madeleine because its long and steep all the time. More interesting are weakest HC climbs. I have no idea why shortened Iseran (15km@6%) and Andore-Arcalis (10.6km@7.1%) are HC - they are neither steep, nor long. Lots of 1 cat climbs are harder and some of them are much harder. Well, maybe legendary Iseran speaks for itself but Arcalis? Tour organizers probably pay much attention to climb altitude. Maybe mountaintop finishes are also favoured somehow. But not always.


1 cat:

Chourchevel 10.6 (799)
Roselend 9.9 (724)
Colombiere 9.3 (754)
La Toussuire 8.9 (670)
Mente 8.4 (770)
...
Aspin (from St Marie de Campan) 4 (286)
Chevreres 3.8 (316)
Epine 3.2 (325)
Croix 3.2 (313)
Firstplan 2.4 (236)

More freaks here. Giant mountaintop finish in Chourchevel (22km@6.2%) is only cat 1. Why? No idea. The same situation with its neighbour - Cormet de Roselend (20km@6%). Although they are not too steep both climbs contain 12-km long sections of 7% average grade. Colombiere is also very hard and it simply must be HC. La Toussuire and Mente are on a HC/1 boarder which is around 8-9 points in my calculations. Some 1 cat climbs are completely undeserved. What can you say about Firstplan (8.4km@5.3%) in 2009 or Epine (6.1km@7.3%)? And now the best fact: Port de Pailheres (15km@8%) one of the hardest climbs in the Pyrenees was cat 1 in 2003! It's score is 11.2 (960). Pailheres and Firstplan in the same cathegory. Tour organisers must have climbed Pailheres by bike and decided to change its status to HC

2 cat:

Grand Ballon 4.8 (284)
Port 4.5 (360)
Montgenevre 4.3 (300)
Portet d'Aspet 4.1 (405)
Lamoura 4 (350)

Here are some contenders to 1 cathegory. The 1/2 cat. boarder is around 4-4.5 points in my calculation. For example Grand Ballon (22km@3.7%), Port (17km@4.6%) or Lamoura (14km@5%) are long, irregular climbs. And Portet d'Aspet is short and very steep (4.3@9.7%). 1/2 cat. boarder is less clear than HC/1 and is violated more often by some weak 1 cat. climbs. Here is a difficulty chart of all HC and 1st cat. climbs and most of 2nd cat. climbs. It's chronogically from left (2005) to right (2014). Blue squares are HC, red diamonds are 1 cat and yellow triangles are 2 cat:

We can seen that over the last 10 years climbs difficulty is more or less the same at the TdF.

 

[Linkinito]

Nice analysis - However you did a mistake, Firstplan never was a Cat. 1 climb.

Iseran was surely HC thanks to its very high altitude (2770 m), even incomplete. However, even if it tops at 2240 m, even if it is a MTF, Arcalis doesn't deserve an HC categorization (even when starting from Andorre-la-Vieille which is 28 km @ 4.5 %). Especially when considering that La Toussuire is Cat. 1.

Also Arcalis was HC in 1997, but Courchevel, being twice as long and nearly as steep, is 1C the same year. Also in 1998 Semnoz from Annecy (17 km @ 7 %) was 1C too.

 

[Krzysztof_O]

Over the internet I found this image

But maybe I am wrong. I also attached a chart to my previous post.

 

[Linkinito]

The official map of the stage says otherwise, so it might be a WIP profile from ASO and there was a mistake.

Also, Télégraphe in 2003 was Cat. 2

 

[Krzysztof_O]

2003 2 cat. Telegraphe's score is 7 (560). Similar difficulty to HC Andore-Arcalis and Iseran Another huge deviation (like Pailheres) in 2003. Tour climb cathegories before 2005 also deserve interest: especially considering climb trends mentioned by Danskebjerge.

 

[Netserk]

Glandon S also went from cat 1 to HC like Pailhères.

 

[gregrowlerson]

Good thread.

I love the Agnel climb. It's almost an angel

On the subject of Arcalis, it gets a lot of hate around here, and a lot to do with that dull 2009 stage I reckon. The climb itself though is solid. The 10km at about 7% isn't really HC (especially as Verbier from the same tour was Cat 1 and possibly harder), but doesn't it go gradually uphill for 20kms before that too (though it's uncategorised)? The '97 stage produced major time gaps so the steady climb is good enough to make for good racing if you put enough climbs before it and don't have a seriously strong headwind.

 

[rghysens]

Good thread.

I love the Agnel climb. It's almost an angel

On the subject of Arcalis, it gets a lot of hate around here, and a lot to do with that dull 2009 stage I reckon. The climb itself though is solid. The 10km at about 7% isn't really HC (especially as Verbier from the same tour was Cat 1 and possibly harder), but doesn't it go gradually uphill for 20kms before that too (though it's uncategorised)? The '97 stage produced major time gaps so the steady climb is good enough to make for good racing if you put enough climbs before it and don't have a seriously strong headwind.

But the 1997 stage was 257km long and had 6 climbs before Arcalis, good for 80km of climbing with more than 4000m elevation gain.

 

[rghysens]

Glandon S also went from cat 1 to HC like Pailhères.

When I first followed the tour (1993), Aspin east and portillon east were 2nd category (last couple of times they were used, it was 1st cat), Pla d'Adet was 1st (now HC), Télégraphe wasn't categorized (just a part of Galibier), Izoard north 1st (now HC),... So that shows there's some inflation in the assesment of the difficulty of climbs the last years.

In the case of the east side of Port de Pailhères, the HC is justified. It's quite long, steep and high enough.

 

[Libertine Seguros]

Good thread.

I love the Agnel climb. It's almost an angel

On the subject of Arcalis, it gets a lot of hate around here, and a lot to do with that dull 2009 stage I reckon. The climb itself though is solid. The 10km at about 7% isn't really HC (especially as Verbier from the same tour was Cat 1 and possibly harder), but doesn't it go gradually uphill for 20kms before that too (though it's uncategorised)? The '97 stage produced major time gaps so the steady climb is good enough to make for good racing if you put enough climbs before it and don't have a seriously strong headwind.

Arcalis has this horrible aura about it and because of all the false flat, unless they come through the Coll d'Ordino everyone's leaving everything until that last bit which isn't really hard enough to break everything up unless you have got a stupendously hard stage like in 1997. It works pretty well as an MTT though, like in the Volta a Catalunya back in the early 2000s.

As for MTFs, Andorra has many better options, as I discussed here. My favourites (assuming Coll de la Gallina is definitely out due to the unpaved part of the south side) being Arinsal:

Els Cortals d'Encamp:

And the double-climb of Llac d'Engolasters:

You could have Envalira-Els Cortals, Envalira-Ordino-Arinsal, Envalira-Ordino-Engolasters, or more quality options when coming from the south (eg in La Vuelta or like in 2009) - La Rabassa-La Comella-Ordino-Arinsal, La Rabassa-Ordino West-Els Cortals, La Rabassa-Engolasters via La Comella... when they get around to paving the whole of La Gallina south and Beixalis East, we will have the scope for some amazing races in Andorra. I'm begging for La Rabassa-La Gallina south-Llac d'Engolasters. Truth be told, given it's only 3km of sterrato, it would be possible now as the full descent would be paved...

 

[Netserk]

When I first followed the tour (1993), Aspin east and portillon east were 2nd category (last couple of times they were used, it was 1st cat), Pla d'Adet was 1st (now HC), Télégraphe wasn't categorized (just a part of Galibier), Izoard north 1st (now HC),... So that shows there's some inflation in the assesment of the difficulty of climbs the last years.

In the case of the east side of Port de Pailhères, the HC is justified. It's quite long, steep and high enough.

You mean Aspin west, no? Or was the east side really categorized 2nd? :O