Author Topic: Network:  Reduction strength  (Read 6544 times)

wayne

  • Full Member
  • ***
  • Posts: 65
    • View Profile
Network:  Reduction strength
« on: May 07, 2007, 07:01:34 AM »
I searched on "reduction strength" first and came up with 0 hits. I know I can't be the first to ask about this feature.

I've been riding Tucson Mtn Park.  For the record, despite the picture at the top, I will not believe you actually rode up Krein trail to the peak.  That was steeper than anticipated.  I walked the bike up, and then walked it back down the other side.

I'm using a "Gen 2" gps; in simple terms it's a DeLorme PN-20 with an STMicro chipset that's more or less equivalent to SiRF III.  The satellites have general drift, and even with WAAS, I've noticed my tracks over time wind up a maximum of 30 feet (mostly 25 feet) apart on the wider trails.

I ran the Network Analysis on default to reduce my tracks, and it didn't work, in particular I noticed it on two parallel tracks that are 25 feet (9 meters) apart.

I looked at the docs and saw this:
Quote
Reduction strength is a measure (in meters) how how close parallel tracks must be to each other in order for them to be considered the same trail. They are then reduced to one track.

I looked at the default in TopoFusion and saw it was set at a whopping 76 meters.  That should have combined a lot more of my tracks.

Without experimentation, my guess is that may supposed to be 7.6 meters?

Can you please clarify this, or explain what I'm doing wrong?  Thanks, Wayne

Krein

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 1203
  • TopoFusion Author
    • View Profile
    • http://www.topofusion.com/diary
Network:  Reduction strength
« Reply #1 on: May 07, 2007, 01:56:39 PM »
Got your pic by email -- thanks.

It's hard to tell without actually looking at the data (in TopoFusion, where I can zoom in), but there is the requirement that two tracks have to actually cross each other twice in order for them to be reduced.  This way the tracks form a 'face' on the graph.

It's kind of hard to explain why this is, but in some cases, where there isn't a lot of data, you might need to grab a point from one track and pull it so that the track crosses with the other.

Usually this is not a problem, but for some data in can be.

If this isn't the case or you need more explanation, let me know and send me your source data so I can take a look.

For the Krein trail -- I cannot ride the whole thing to the peak, no.  There are 3 sections (one the final turn at the top) that I have never ridden uphill.  I have seen the top switchback cleaned on the downhill, though.

Everything else is rideable, but you have to be superman to string it all together.  I dab and flail around up there, and that's what keeps me going back -- the challenge.

wayne

  • Full Member
  • ***
  • Posts: 65
    • View Profile
Network:  Reduction strength
« Reply #2 on: May 07, 2007, 02:54:50 PM »
I fall too much; I'm very much a poor rider.  I curse the clips at times.

You nailed it on the road--I have two long stretches that run as parallel crescents, where I rode up the road on the right, and rode back down on the other right.  They cross once, up at the trail head.  I will drag two points and make them cross.  You can probably ignore the 2nd picture I just sent where my "ah-ha" moment now seem like a "wow, what a bad guess" moment.

I still don't get the reduction strength--is the default really 76 meters?  Or is that incorrect?

Krein

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 1203
  • TopoFusion Author
    • View Profile
    • http://www.topofusion.com/diary
Network:  Reduction strength
« Reply #3 on: May 07, 2007, 05:50:15 PM »
Yes, it really is 76 meters.  Note, however, that it is the maximum distance that the tracks can be apart -- meaning that at all times they must been within 76 meters or less.

It sounds kind of large, but for general purpose networks, I've found it to be a good starting value, or roughly around there.  I think the default is a little lower than that?

wayne

  • Full Member
  • ***
  • Posts: 65
    • View Profile
Network:  Reduction strength
« Reply #4 on: May 07, 2007, 06:02:26 PM »
Quote (Krein @ May 07 2007,6:50)
Yes, it really is 76 meters.  Note, however, that it is the maximum distance that the tracks can be apart -- meaning that at all times they must been within 76 meters or less.

As in "There is a tube around the tracks; at all points in that tube they must be 76 meters or closer.  The start and finish points can be farther than 76 meters."

Very good information to know.  I'm amazed more of my tracks didn't get pulled together.  I'm amazed how well it works.  I'm going to start breaking tracks up before I network them, and tighten that grouping way down to see if I can get rid of some goofieness; or maybe run it first then play with the tracks and run it again.

From the picture I sent you, two tracks have a long parallel stretch, but then diverge as I go down separate trails.  I have to make certain the two tracks are broken at the point where I separate for the network function to work correctly.

That also explains some of the goofy connections I get.

wayne

  • Full Member
  • ***
  • Posts: 65
    • View Profile
Network:  Reduction strength
« Reply #5 on: August 14, 2007, 04:18:15 PM »
I've gotten more of the hang of it.  I do things like make certain I drag and "twist" ends together to make certain they merge/average.

Please forgive the mis-spellings; I didn't want to go through the work of re-grabbing and re-editing, then rehosting the picture.

What I'm trying to say on Picture 1 is that I drag the track points to make a triangle (which is why the red trails are so perfectly on top of each other), then I lower the track reduction strength until it doesn't eliminate triangles.  I don't mind small triangles getting flattened (this one isn't a large one; I don't mind if it goes away) but the larger ones I like to keep.