This is a continuation of a previous article (→here).
I'll write about cases where SRAM's Karoo recognizes a route as a climb but then withdraws that recognition.
The content includes place names and routes that will be completely unclear to anyone not from the local area,
so please bear with me.
↑This is my mental map of the Rokko area.
On the map above, I run along National Route 2 from the Osaka direction
to around Shukugawa,
then turn right at the Kaguramachi intersection,
and the prefectural road 82 on the map has a temple called Jūrinji (鷲林寺)
along the way, which also gave its name to the surrounding area.
I call this the "Jūrinji Route."
After turning left at the Kōjuubashi intersection,
it's a single road all the way to Ikkenchaya, which is the summit within the paved section.
However, the route with general consensus for measuring Rokko Mountain times
is the "Sakasegawa Station Route" from Sakasegawa Station to Kōjuubashi to Ikkenchaya.
When climbing from Jūrinji, there's no place where multiple people can gather as a starting point,
and the road to Kōjuubashi is narrow with heavy traffic,
so overtaking timing is determined by traffic volume.
Plus, it's quite a steep climb, and the average gradient to Kōjuubashi
is far steeper than the Sakasegawa Route.
The reason I don't climb from the Sakasegawa Route is simply that I don't know how to get to Sakasegawa Station.
I've climbed from the Sakasegawa Route maybe five times in my life,
but each time I ended up climbing back up after coming down from Kōjuubashi.
↑Near the Kōjuubashi on the Sakasegawa Route, I've drawn a green mesh overlay.
This is because the road passes through in a way that divides a golf course,
and they've installed netting over the road
so golf balls don't hit passing vehicles.
A little further down from there is a bakery called "Pannell" (パンネル) with its main shop in Takarazuka,
and its Sakasegawa branch is in the first floor of a condominium along the road.
I'd stop by saying things like "Let's go! Pannell!" in a Haman Karn-like way,
but when I mentioned this to a customer familiar with Rokko,
they told me that Pannell had closed long ago.
. . . After checking, it turns out it closed in October 2010 (Heisei 22).
That's how long I haven't been climbing the Sakasegawa Route.
When returning from Sakasegawa Station, I would either climb to Kōjuubashi
and descend via the Jūrinji Route,
or head east somewhat haphazardly
and when I recognized a familiar spot on National Route 176 (Inaroku),
head south toward Jūsan,
which was a low-reproducibility, roundabout return route.
After descending the Jūrinji Route and turning at Kaguramachi
to head toward Osaka, around the Kawaragi intersection
(on the opposite lane for the return trip) there's a Chinese restaurant called "Yelaixiang" (夜来香)
where I often ate plate udon, but it also closed in October 2020.
The Jūrinji Route, which had no consensus whatsoever,
I set as my starting point at a railroad crossing of the Hankyu Kōyō Line
after turning at Kaguramachi, going under the elevated JR Kobe Line at Sakura Shukugawa Station,
and going under the elevated Hankyu Kobe Line at the Wakamatsucho intersection
(red crossing in the diagram above).
There's a Seven-Eleven at the intersection just after Wakamatsucho,
where I could resupply, but checking now shows it was already closed
as of November 2009, the limit of what Google Maps can track,
and now there's an internal medicine clinic there.
Even for people living in the Shukugawa or Nishinomiya area,
wouldn't they also climb the more commonly recognized Sakasegawa Station Route
rather than the local Jūrinji Route when talking about climb times with someone?
So on June 28th, starting early in the morning at 3:31 AM,
I went from Nomulab to Sakasegawa Station following "100% following Karoo's navigation" as the rule,
then climbed to Kōjuubashi and descended via Jūrinji to return.
The start had large raindrops falling continuously,
becoming light rain around sunrise, but the road surface remained wet throughout.
↑This is the Hammerhead Dashboard log for that route.
The purple circle before the Yodogawa is from accidentally pressing the lap button
on the way back.
99% of my Karoo lap records are from accidental taps.
The mental map from the beginning again.
According to Karoo's navigation,
I turned north at the Mukoōhashi West End intersection on Route 2
and rode on Prefectural Road 114.
It was truly a road I'd never ridden before.
This road runs along the Mukō River, but it's a narrow road with just one lane each way,
and despite the early hour with little traffic, the probability that a vehicle is a truck is extremely high.
Since the road is narrow, trucks split the centerline to give me a wide berth when passing,
and in doing so, without any ill intent,
I got hit with intense spray from their tires several times.
Later, a different person from the one who said "Pannell's gone" told me,
"It's better to go down to the riverbed and run along the riverbed as much as possible to get to Sakasegawa Station."
Apart from being narrow, this road also passes under several elevated structures,
and each time the road dips like the disinfectant pool you wade through up to your waist
before a school pool class
(though apparently they don't do that anymore;
newly built schools don't even have pools to put them in).
While not quite a flooded section, quite large puddles had formed.
I ran on Route 114 right up to just before Sakasegawa Station,
and where I thought I'd turn left,
at a certain junction, I was instructed to turn left in a wide arc
away from the Mukō River.
The dotted line I added to the diagram above shows that.
↑In the Karoo log, it looks like this.
The white circle in the gray frame in the image is the pin I placed for Sakasegawa Station.
This residential area route would normally require caution about moms on bikes or scooters darting out,
a road with poor visibility rather than just narrow,
but since the rule was "100% follow Karoo's navigation,"
I rode exactly as instructed.
Climbing from Sakasegawa Station to Kōjuubashi.
The image above is right after starting near the station,
where the road is rutted by car tracks
and the resulting puddle splashes quite a bit,
so I carefully rode in the narrow gap between the left tire rut and the white line.
The Sakasegawa around here,
except right after rain, on a clear day
the water doesn't flow like a waterfall this heavily.
Immediately after starting from Sakasegawa Station,
Karoo automatically generated a folder for "Climb 1"
from the starting point to just before the elevation of Rokko Mountain starts descending
(meaning up to around Ikkenchaya)
and it appeared from the bottom of the screen.
The Sakasegawa is a tributary of the Mukō River.
At the start of the climb from Sakasegawa Station,
it flows along the left side of the road,
but after running for a while, the road curves right as it moves away from the river,
and at a bridge called Hakase River Bridge a short ways ahead,
the river passes under the road,
so now it appears on the right side.
The image above is the last section of the Sakasegawa visible from the road,
with a mysterious monument on the left side,
and when I made a U-turn on the opposite lane to photograph this,
the Climb 1 measurement ended.
And just a trivial thing,
the sight of this as a stepped waterfall with flowing water
is not something you see on clear days except right after rain.
We've learned that Karoo's climb recognition behavior will end the climb immediately
not just by leaving the route and going somewhere else,
but even just by making a U-turn.
This climb is "complete," not erased as if it never happened.
Then after climbing again, Karoo automatically generated Climb 2
to the top of Rokko around the golf course netting area,
but since I was already before Kōjuubashi,
as soon as I turned left (Jūrinji Route) instead of going straight (to Rokko),
Climb 2 was also marked complete.
The Kōjuubashi intersection.
When climbing Rokko from the Jūrinji route,
it's like coming from the lower left in the image
and turning left onto the road in the upper left,
so with my "riding on the sidewalk" performance,
I essentially don't hit the traffic signal.
Right before this ride, I'd set my front disc rotor to 140mm,
and I thought the steep descent of Jūrinji would be useful
for bedding-in (break-in),
but descending under the conditions of a new rotor plus rain,
even a disc brake can experience
the situation where "you can decelerate but can't stop immediately."
Release the brake just a little (don't grip the lever hard)
and you're instantly doing 50 km/h, yeah!
Let me look at the story up to here on the Hammerhead Dashboard log.
↑The white circle in the upper right is Sakasegawa Station,
and the white circle on the left is Ikkenchaya on Rokko Mountain.
Below is the elevation graph of the route I rode,
and the highest point of this is Kōjuubashi.
The blue bar below that is Karoo's climb records,
the purple bar below that is lap time, where I accidentally tapped on the way back.
Despite being a mistake, taking a lap time meant that everything up to that point became Lap 1,
and everything after became Lap 2, but if I hadn't taken any lap times at all,
the purple bar itself wouldn't be displayed.
Clicking on the Climb 1 section highlighted the portion of the route corresponding to Climb 1
in blue instead of black.
From here, let me look at the start and end points of Climb 1 in detail.
↑The start point.
I bought a canned coffee from a vending machine at the station
and took a commemorative photo before the first bridge
before starting, and that trace remains,
but aside from that, Climb 1 starts from a little way after riding off.
↑The end point.
I can see the trace of the U-turn I made to photograph the stepped waterfall at the bridge,
and Climb 1 ends exactly at the U-turn point.
After Climb 1 is complete, riding a bit more,
Karoo undeterred again generates Climb 2
from that point to the top of Rokko,
but I turned at Kōjuubashi so quickly
that Climb 2 was also marked complete at that turn.
When you hover your mouse over the elevation graph,
all the data recorded at that point appears.
Also, where that location was appears as a yellow dot on the route line of the map.
The gradient being -12% and the speed being 36.3 km/h is actually
where I was braking more than necessary to slow down.
I wanted to get the disc rotor bedding done in a single descent.
With a 42T front ring and a rear cog not of 10T or 11T
but 14T means I wasn't pedaling on the descent.
When the automatically generated Climb 1 folder
appeared from the bottom of the screen with a ridiculously loud beep,
I was taking photos of the river and the start was heavy rain,
so I manually stopped Karoo's measurement to remove
some of the rain gear I was completely equipped with
to travel lighter.
I didn't ride at all during that time.
It wasn't stopped for many minutes, but it seems
there was some barometric pressure fluctuation at the same location,
as the elevation changed like a cliff face.
Looking at the gradient recorded every second in this section,
between 0% and 0%,
it changed as: 0% → 23.6% → 30.4% → 39.3% → 47.1% → 53.5% → 58.5% → 54.2% → 0%
and this maximum value of 58.5% was recorded as the maximum gradient for that day.
It doesn't always happen, but on days when I remember manually stopping Karoo's measurement,
the maximum gradient data sometimes shows ridiculous numbers.
Previously, I wrote about how Polar's heart rate monitor displays
the gradient during a ride by calculating it from recently recorded altitude data,
so it shows the gradient of the road you were riding a few seconds ago with a lag.
In contrast, Karoo apparently displays the gradient of your current location
from pre-loaded map data, with no lag like Polar.
But then this kind of gradient display bug shouldn't occur.
One more thing I've figured out: for example, if you pick up a GPS signal with both
Polar and Karoo on the 5th floor of an apartment building,
then move to the 1st floor with barely any GPS position change (elevator or stairs),
Polar's elevation drops by about 15m, but Karoo's altitude
stays at the 5th floor reading unchanged.
However, right after arriving on the 1st floor, it sometimes displays
an absurdly negative gradient.
So does that mean Karoo gets current elevation from map data
but gradient readings from the barometer?
That's not quite right either, as I've confirmed.
When you go up a sloped pedestrian overpass
that bicycles can climb, not stairs,
Karoo's elevation increases by about 6m,
and when you descend the overpass, it decreases by about 6m.
This means the elevation data also refers to the barometer.
In this case, you've moved tens of meters,
so GPS-wise you're clearly moving.
If GPS data includes the exact position of the pedestrian overpass rather than ground level,
then elevation would increase by about 6m only while directly under the overpass,
but I don't think that's actually happening.
↑This is my commute route.
I turn left at the north dead end of Abiko-suji
(the initial left turn is actually at a T-junction;
the road going north is for vehicles only on an elevated expressway),
do a two-step right turn at the Abeno Harukas intersection,
and head north on Tanimachisuji . . .
at that gentle uphill near Tennoji Station,
which has nothing special for gradient,
is always recognized as Climb 1.
According to the "actual ride" results in this log,
Climb 1 has an average gradient of 1.3%
and a maximum gradient of 5.1%.
This matches my sense of the gradient.
↑This is an actual image of this Climb 1's automatically generated folder.
Since I couldn't photograph on the road, I stopped on the sidewalk to take it.
According to map data, the latter half of Climb 1
is 11.9% followed by an 8.0% slope
(it varies by about 0.1% each time),
and in the remaining 0.1km (100m), you'd gain 14m.
But no such gradient actually exists.
When you finish riding through this Climb 1 section,
it's recorded as Climb 1 "complete."
However, looking at the log afterwards, with this Climb 1 showing
an average gradient of 1.3% and a maximum of 5.1%,
the elevation data saved in the ride log is barometer-based,
but in map data, this hill supposedly has a maximum gradient in the 11% range,
so it seems the recognition as a climb works
based on map data while the actual gradient is from the barometer.
This Climb 1 is announced with a ridiculously loud beep
before the hill, but sometimes doesn't appear on Sundays.
That would mean Karoo's judgment on whether to recognize an uphill section
has a "day of the week" element . . . which doesn't make sense.
Yet it's a fact that on Sundays it's more common for Climb 1 not to appear.
This is related to how on Sundays, there's no line of taxis
waiting for passengers in the middle of the hill like they're parked,
so I can ride in the part of the road nearest the sidewalk.
There's something like a collision detection box from a computer game
set in the middle of the road at what Karoo decides is the start of the uphill section,
and if your GPS position doesn't contact this, Climb 1 won't be announced—
I think that's how it works.
The "fact" that "the same hill is only not treated as a hill on Sundays"
seems like a setup for a false cause fallacy,
and I personally find it interesting.
In America, GM received a complaint from a customer:
"My new Pontiac doesn't stall when I go to an ice cream shop and buy chocolate or strawberry ice cream,
but the engine won't start when I buy vanilla ice cream,"
and GM engineers actually investigated and resolved it.
Of course, the Pontiac isn't changing its attitude based on the ice cream flavor.
There was once a one-off modified Pontiac Firebird Trans Am
with artificial intelligence, able to talk and going 300 mph top speed,
but this is unrelated.
If you want to know more, search "ice cream stalling."
↑This is the last image from the first part of the article.
When descending Kazahuki Pass toward the Osaka side, just past the rest stop,
Karoo announces a climb with an absurd gradient.
This happens with 100% reproducibility every time.
Of course, the day of the week doesn't matter.
This is at the 104km mark when I rode 141.7km on July 28th,
almost the same location.
Sure enough, the usual climb appeared.
A climb of 0.4km distance and 33m elevation,
but most of the uphill consists of 15.0% and 11.0% gradient sections.
When you look at this on the map
the climb content was entering the old road straddling the opposing lane.
The current Kazahuki Pass road has been rerouted toward a new tunnel,
and parts of the old road are visible beside the new road
with accessible sections, but
since nowhere goes, normally nobody rides this road.
If you ignore the right turn intersection about 200m ahead
and continue on this road, this climb becomes "complete"
rather than "erased."
Two images back is just before that intersection, about 200m away from here.
After seeing this, I wondered if the Tennoji climb somehow actually connects
to a road with grades in the 11% range, so I checked again,
but that climb has the actual riding route and
completing it marks it as "complete,"
so the situation is different.
The continuation is getting long, so
I'll put it in the next article.
I'll write about cases where SRAM's Karoo recognizes a route as a climb but then withdraws that recognition.
The content includes place names and routes that will be completely unclear to anyone not from the local area,
so please bear with me.
↑This is my mental map of the Rokko area.
On the map above, I run along National Route 2 from the Osaka direction
to around Shukugawa,
then turn right at the Kaguramachi intersection,
and the prefectural road 82 on the map has a temple called Jūrinji (鷲林寺)
along the way, which also gave its name to the surrounding area.
I call this the "Jūrinji Route."
After turning left at the Kōjuubashi intersection,
it's a single road all the way to Ikkenchaya, which is the summit within the paved section.
However, the route with general consensus for measuring Rokko Mountain times
is the "Sakasegawa Station Route" from Sakasegawa Station to Kōjuubashi to Ikkenchaya.
When climbing from Jūrinji, there's no place where multiple people can gather as a starting point,
and the road to Kōjuubashi is narrow with heavy traffic,
so overtaking timing is determined by traffic volume.
Plus, it's quite a steep climb, and the average gradient to Kōjuubashi
is far steeper than the Sakasegawa Route.
The reason I don't climb from the Sakasegawa Route is simply that I don't know how to get to Sakasegawa Station.
I've climbed from the Sakasegawa Route maybe five times in my life,
but each time I ended up climbing back up after coming down from Kōjuubashi.
↑Near the Kōjuubashi on the Sakasegawa Route, I've drawn a green mesh overlay.
This is because the road passes through in a way that divides a golf course,
and they've installed netting over the road
so golf balls don't hit passing vehicles.
A little further down from there is a bakery called "Pannell" (パンネル) with its main shop in Takarazuka,
and its Sakasegawa branch is in the first floor of a condominium along the road.
I'd stop by saying things like "Let's go! Pannell!" in a Haman Karn-like way,
but when I mentioned this to a customer familiar with Rokko,
they told me that Pannell had closed long ago.
. . . After checking, it turns out it closed in October 2010 (Heisei 22).
That's how long I haven't been climbing the Sakasegawa Route.
When returning from Sakasegawa Station, I would either climb to Kōjuubashi
and descend via the Jūrinji Route,
or head east somewhat haphazardly
and when I recognized a familiar spot on National Route 176 (Inaroku),
head south toward Jūsan,
which was a low-reproducibility, roundabout return route.
After descending the Jūrinji Route and turning at Kaguramachi
to head toward Osaka, around the Kawaragi intersection
(on the opposite lane for the return trip) there's a Chinese restaurant called "Yelaixiang" (夜来香)
where I often ate plate udon, but it also closed in October 2020.
The Jūrinji Route, which had no consensus whatsoever,
I set as my starting point at a railroad crossing of the Hankyu Kōyō Line
after turning at Kaguramachi, going under the elevated JR Kobe Line at Sakura Shukugawa Station,
and going under the elevated Hankyu Kobe Line at the Wakamatsucho intersection
(red crossing in the diagram above).
There's a Seven-Eleven at the intersection just after Wakamatsucho,
where I could resupply, but checking now shows it was already closed
as of November 2009, the limit of what Google Maps can track,
and now there's an internal medicine clinic there.
Even for people living in the Shukugawa or Nishinomiya area,
wouldn't they also climb the more commonly recognized Sakasegawa Station Route
rather than the local Jūrinji Route when talking about climb times with someone?
So on June 28th, starting early in the morning at 3:31 AM,
I went from Nomulab to Sakasegawa Station following "100% following Karoo's navigation" as the rule,
then climbed to Kōjuubashi and descended via Jūrinji to return.
The start had large raindrops falling continuously,
becoming light rain around sunrise, but the road surface remained wet throughout.
↑This is the Hammerhead Dashboard log for that route.
The purple circle before the Yodogawa is from accidentally pressing the lap button
on the way back.
99% of my Karoo lap records are from accidental taps.
The mental map from the beginning again.
According to Karoo's navigation,
I turned north at the Mukoōhashi West End intersection on Route 2
and rode on Prefectural Road 114.
It was truly a road I'd never ridden before.
This road runs along the Mukō River, but it's a narrow road with just one lane each way,
and despite the early hour with little traffic, the probability that a vehicle is a truck is extremely high.
Since the road is narrow, trucks split the centerline to give me a wide berth when passing,
and in doing so, without any ill intent,
I got hit with intense spray from their tires several times.
Later, a different person from the one who said "Pannell's gone" told me,
"It's better to go down to the riverbed and run along the riverbed as much as possible to get to Sakasegawa Station."
Apart from being narrow, this road also passes under several elevated structures,
and each time the road dips like the disinfectant pool you wade through up to your waist
before a school pool class
(though apparently they don't do that anymore;
newly built schools don't even have pools to put them in).
While not quite a flooded section, quite large puddles had formed.
I ran on Route 114 right up to just before Sakasegawa Station,
and where I thought I'd turn left,
at a certain junction, I was instructed to turn left in a wide arc
away from the Mukō River.
The dotted line I added to the diagram above shows that.
↑In the Karoo log, it looks like this.
The white circle in the gray frame in the image is the pin I placed for Sakasegawa Station.
This residential area route would normally require caution about moms on bikes or scooters darting out,
a road with poor visibility rather than just narrow,
but since the rule was "100% follow Karoo's navigation,"
I rode exactly as instructed.
Climbing from Sakasegawa Station to Kōjuubashi.
The image above is right after starting near the station,
where the road is rutted by car tracks
and the resulting puddle splashes quite a bit,
so I carefully rode in the narrow gap between the left tire rut and the white line.
The Sakasegawa around here,
except right after rain, on a clear day
the water doesn't flow like a waterfall this heavily.
Immediately after starting from Sakasegawa Station,
Karoo automatically generated a folder for "Climb 1"
from the starting point to just before the elevation of Rokko Mountain starts descending
(meaning up to around Ikkenchaya)
and it appeared from the bottom of the screen.
The Sakasegawa is a tributary of the Mukō River.
At the start of the climb from Sakasegawa Station,
it flows along the left side of the road,
but after running for a while, the road curves right as it moves away from the river,
and at a bridge called Hakase River Bridge a short ways ahead,
the river passes under the road,
so now it appears on the right side.
The image above is the last section of the Sakasegawa visible from the road,
with a mysterious monument on the left side,
and when I made a U-turn on the opposite lane to photograph this,
the Climb 1 measurement ended.
And just a trivial thing,
the sight of this as a stepped waterfall with flowing water
is not something you see on clear days except right after rain.
We've learned that Karoo's climb recognition behavior will end the climb immediately
not just by leaving the route and going somewhere else,
but even just by making a U-turn.
This climb is "complete," not erased as if it never happened.
Then after climbing again, Karoo automatically generated Climb 2
to the top of Rokko around the golf course netting area,
but since I was already before Kōjuubashi,
as soon as I turned left (Jūrinji Route) instead of going straight (to Rokko),
Climb 2 was also marked complete.
The Kōjuubashi intersection.
When climbing Rokko from the Jūrinji route,
it's like coming from the lower left in the image
and turning left onto the road in the upper left,
so with my "riding on the sidewalk" performance,
I essentially don't hit the traffic signal.
Right before this ride, I'd set my front disc rotor to 140mm,
and I thought the steep descent of Jūrinji would be useful
for bedding-in (break-in),
but descending under the conditions of a new rotor plus rain,
even a disc brake can experience
the situation where "you can decelerate but can't stop immediately."
Release the brake just a little (don't grip the lever hard)
and you're instantly doing 50 km/h, yeah!
Let me look at the story up to here on the Hammerhead Dashboard log.
↑The white circle in the upper right is Sakasegawa Station,
and the white circle on the left is Ikkenchaya on Rokko Mountain.
Below is the elevation graph of the route I rode,
and the highest point of this is Kōjuubashi.
The blue bar below that is Karoo's climb records,
the purple bar below that is lap time, where I accidentally tapped on the way back.
Despite being a mistake, taking a lap time meant that everything up to that point became Lap 1,
and everything after became Lap 2, but if I hadn't taken any lap times at all,
the purple bar itself wouldn't be displayed.
Clicking on the Climb 1 section highlighted the portion of the route corresponding to Climb 1
in blue instead of black.
From here, let me look at the start and end points of Climb 1 in detail.
↑The start point.
I bought a canned coffee from a vending machine at the station
and took a commemorative photo before the first bridge
before starting, and that trace remains,
but aside from that, Climb 1 starts from a little way after riding off.
↑The end point.
I can see the trace of the U-turn I made to photograph the stepped waterfall at the bridge,
and Climb 1 ends exactly at the U-turn point.
After Climb 1 is complete, riding a bit more,
Karoo undeterred again generates Climb 2
from that point to the top of Rokko,
but I turned at Kōjuubashi so quickly
that Climb 2 was also marked complete at that turn.
When you hover your mouse over the elevation graph,
all the data recorded at that point appears.
Also, where that location was appears as a yellow dot on the route line of the map.
The gradient being -12% and the speed being 36.3 km/h is actually
where I was braking more than necessary to slow down.
I wanted to get the disc rotor bedding done in a single descent.
With a 42T front ring and a rear cog not of 10T or 11T
but 14T means I wasn't pedaling on the descent.
When the automatically generated Climb 1 folder
appeared from the bottom of the screen with a ridiculously loud beep,
I was taking photos of the river and the start was heavy rain,
so I manually stopped Karoo's measurement to remove
some of the rain gear I was completely equipped with
to travel lighter.
I didn't ride at all during that time.
It wasn't stopped for many minutes, but it seems
there was some barometric pressure fluctuation at the same location,
as the elevation changed like a cliff face.
Looking at the gradient recorded every second in this section,
between 0% and 0%,
it changed as: 0% → 23.6% → 30.4% → 39.3% → 47.1% → 53.5% → 58.5% → 54.2% → 0%
and this maximum value of 58.5% was recorded as the maximum gradient for that day.
It doesn't always happen, but on days when I remember manually stopping Karoo's measurement,
the maximum gradient data sometimes shows ridiculous numbers.
Previously, I wrote about how Polar's heart rate monitor displays
the gradient during a ride by calculating it from recently recorded altitude data,
so it shows the gradient of the road you were riding a few seconds ago with a lag.
In contrast, Karoo apparently displays the gradient of your current location
from pre-loaded map data, with no lag like Polar.
But then this kind of gradient display bug shouldn't occur.
One more thing I've figured out: for example, if you pick up a GPS signal with both
Polar and Karoo on the 5th floor of an apartment building,
then move to the 1st floor with barely any GPS position change (elevator or stairs),
Polar's elevation drops by about 15m, but Karoo's altitude
stays at the 5th floor reading unchanged.
However, right after arriving on the 1st floor, it sometimes displays
an absurdly negative gradient.
So does that mean Karoo gets current elevation from map data
but gradient readings from the barometer?
That's not quite right either, as I've confirmed.
When you go up a sloped pedestrian overpass
that bicycles can climb, not stairs,
Karoo's elevation increases by about 6m,
and when you descend the overpass, it decreases by about 6m.
This means the elevation data also refers to the barometer.
In this case, you've moved tens of meters,
so GPS-wise you're clearly moving.
If GPS data includes the exact position of the pedestrian overpass rather than ground level,
then elevation would increase by about 6m only while directly under the overpass,
but I don't think that's actually happening.
↑This is my commute route.
I turn left at the north dead end of Abiko-suji
(the initial left turn is actually at a T-junction;
the road going north is for vehicles only on an elevated expressway),
do a two-step right turn at the Abeno Harukas intersection,
and head north on Tanimachisuji . . .
at that gentle uphill near Tennoji Station,
which has nothing special for gradient,
is always recognized as Climb 1.
According to the "actual ride" results in this log,
Climb 1 has an average gradient of 1.3%
and a maximum gradient of 5.1%.
This matches my sense of the gradient.
↑This is an actual image of this Climb 1's automatically generated folder.
Since I couldn't photograph on the road, I stopped on the sidewalk to take it.
According to map data, the latter half of Climb 1
is 11.9% followed by an 8.0% slope
(it varies by about 0.1% each time),
and in the remaining 0.1km (100m), you'd gain 14m.
But no such gradient actually exists.
When you finish riding through this Climb 1 section,
it's recorded as Climb 1 "complete."
However, looking at the log afterwards, with this Climb 1 showing
an average gradient of 1.3% and a maximum of 5.1%,
the elevation data saved in the ride log is barometer-based,
but in map data, this hill supposedly has a maximum gradient in the 11% range,
so it seems the recognition as a climb works
based on map data while the actual gradient is from the barometer.
This Climb 1 is announced with a ridiculously loud beep
before the hill, but sometimes doesn't appear on Sundays.
That would mean Karoo's judgment on whether to recognize an uphill section
has a "day of the week" element . . . which doesn't make sense.
Yet it's a fact that on Sundays it's more common for Climb 1 not to appear.
This is related to how on Sundays, there's no line of taxis
waiting for passengers in the middle of the hill like they're parked,
so I can ride in the part of the road nearest the sidewalk.
There's something like a collision detection box from a computer game
set in the middle of the road at what Karoo decides is the start of the uphill section,
and if your GPS position doesn't contact this, Climb 1 won't be announced—
I think that's how it works.
The "fact" that "the same hill is only not treated as a hill on Sundays"
seems like a setup for a false cause fallacy,
and I personally find it interesting.
In America, GM received a complaint from a customer:
"My new Pontiac doesn't stall when I go to an ice cream shop and buy chocolate or strawberry ice cream,
but the engine won't start when I buy vanilla ice cream,"
and GM engineers actually investigated and resolved it.
Of course, the Pontiac isn't changing its attitude based on the ice cream flavor.
There was once a one-off modified Pontiac Firebird Trans Am
with artificial intelligence, able to talk and going 300 mph top speed,
but this is unrelated.
If you want to know more, search "ice cream stalling."
↑This is the last image from the first part of the article.
When descending Kazahuki Pass toward the Osaka side, just past the rest stop,
Karoo announces a climb with an absurd gradient.
This happens with 100% reproducibility every time.
Of course, the day of the week doesn't matter.
This is at the 104km mark when I rode 141.7km on July 28th,
almost the same location.
Sure enough, the usual climb appeared.
A climb of 0.4km distance and 33m elevation,
but most of the uphill consists of 15.0% and 11.0% gradient sections.
When you look at this on the map
the climb content was entering the old road straddling the opposing lane.
The current Kazahuki Pass road has been rerouted toward a new tunnel,
and parts of the old road are visible beside the new road
with accessible sections, but
since nowhere goes, normally nobody rides this road.
If you ignore the right turn intersection about 200m ahead
and continue on this road, this climb becomes "complete"
rather than "erased."
Two images back is just before that intersection, about 200m away from here.
After seeing this, I wondered if the Tennoji climb somehow actually connects
to a road with grades in the 11% range, so I checked again,
but that climb has the actual riding route and
completing it marks it as "complete,"
so the situation is different.
The continuation is getting long, so
I'll put it in the next article.