I another thread I asked about the Sierra Cascades (ACA) route. Contributions pointed to the fact that this is a very difficult course.

What tools do you use to (objectively) assess a course difficulty? At this point I know about climByBike. Great on a per-hill basis, but not so much to, let say, compare the Sierra Cascade to the Pacific Coast, or to fine-tune a course to avoid killer-hills.

At this point I intend to map on Google Earth, examine the elevation profile and zero-in on what may look like places we'd rather avoid. The problem is that a 100km-long 5% hill looks like a monstrous obstacle but in fact is much less problematic that a series of short climbs at 15%+.

hmmm.... I suppose that my question is something like:

1. what constitute a practically unsurmountable obstacle?
2. what is the most convenient way to make sure that there is no such road-block on a (lengthy) planned course

---------
UPDATE
---------

I stumbled across what seems to be the best calculator for my purpose. It relies on wind, gradient and several other measures to compute a flat course equivalent. It highlights the relative importance of headwinds and maximum gradient in making a course difficult. For example, the Sherman pass (H99) climbs 1627 meters over a distance of 24.5 kms. According to the FLA calculator, these 24.5 kms are equivalent to 87 kms on a flat, windless course. The climbing effect is equivalent to headwinds blowing at 50 kms...

If the same climb were of a lower gradient (say, spread over a 80k distance), the flat windless equivalent would have been 125k. With 50k headwinds, the FCE becomes a mind boggling 365k.

As good as this calculator is, it is still not trivial to solve the basic problem of carving up a route in FCEs of a given length. (If someone is interested in giving a shot at it, let me know.) But it is pretty good as it is.

-----------
UPDATE 2
-----------

Computing the flat course equivalent (FCE) may be interesting, but (1) it would take forever to do this with actual data (2) two climbing courses may have an identical FCE, yet one course could be impossible to ride if it rises too abruptly, beyond the gradient that a loaded tourer can negotiate.

An interesting calculator computes the speed at which a rider will progress given power (expressed in watts), weight and gradient among other things. An average rider i said to be capable of sustaining outputs in the 100-200 watts range. It is also possible to use the calculator to infer wattage. 100-200 + bursts at 300 fit my experience quite well.

Armed with this, we find that at 200 watts, a rider weighing 75kg, riding a 13kg touring bike loaded with 20kg of supplies will travel at 27 kph on a flat course, and at 5.9 kph if the gradient is 10%. 5.9 kph is the lowest ratio that my drivetrain can deliver (26x34) at the sluggish cadence of 64. If the gradient becomes 15%, achievable speed drops to 4.1 kph -- meaning either dismounting or increasing the effort to an unsustainable level. These limits fit very well with my experience and allowed me to look at the impact of (bike=supplies) weight (negligible) and headwinds (also negligible) on course practicability.

Bottom line for this first part of the "solution" is that the maximum long-run gradient that I can practically climb is 10% max, with short sections of 15%.

Then it becomes a fairly simple matter of inspecting a course for sustained high and maximum gradients. The best if not only way to do it right is to use data triplets (lat-long-elevation) from GPX files and compute point-to-point gradients. Routing softwares aggregate these date coarsely and therefore provide "optimist" gradients.

The easiest way to generate the GPX is by using gMaps to create the route (most routing sites use Google's API anyway) in conjunction with gpsVisualizer. Copy and paste Gmaps' URL and choose "add DEM elevation data". Convert and plug the resulting data in a spreadsheet (which will require some text processing). Use this formula to compute distances, divide elevation differences by point-to-point distance and you'll obtain the point-to-point gradient.

A typical course will contain thousands of gradient, making it much easier to anticipate killer hills. Charts may look like (elevation in blue and gradient in red):



------------
UPDATE 3
-----------

I now think that I have what was looking for -- I can generate (conservative) time estimates for a course, calculated by taking the average gradient at each kilometer.

1. I used this calculator to generate speed estimates at various gradients (from -10% to +10%) for a loaded touring bike on which the rider applies 100W of force. Estimates run from a low of 2.8kph at +10% to a high of close to 80kph, that I capped at 50.

2. I generate the course profile (gMaps or Strava) + GPS visualizer and plug the lat-long-ele data into a spreadsheet, which computes average speed and expected riding time at each km.

Estimates are quite instructive. For exemple, the Wawona-> Lee Vining segment of the Sierra Cascades is forecast at close to 16h even though it is less than 160km long. Those I did for segments I am familiar with are reasonable. If anyone wants more detail on this recipe, let me know.