MTB Project Logo

automated conditions


Original Post
Preston Jordan · · Albany, CA · Joined Jan 2018 · Points: 613

Hi-

I just bought my first smart phone (my flip phone buttons were starting to fail and my family has been on me for years to get a smart phone so they can more readily track me and call the cleanup crew if I stack too badly).

I started riding in 1983.  For decades I have envisioned an automated trail conditions system. Experiencing the capabilities of a smart phone, I think the time has arrived to offer the concept. After looking around at platforms, I picked MTB project for rolling out this suggestion. I have been an REI member for over thirty years, and I like that the MTB project does not have a pushy social networking or competitive aspect, and is not for profit.

There are a number approaches to automating trail conditions. To start with, the simplest would be for the MTB project mapping system to ingest the total precipitation data for the past 1, 7, 14, 30, and 60 days from NOAA (https://water.weather.gov/precip/download.php) and map these onto trails in some manner. Cumulative precipitation over various previous length periods is highly correlated to trail conditions. I don't really care if it is raining at the moment or might rain in the near future when I am riding, which is the focus of all the weather apps I have browsed. Riding in whatever conditions come is part of the fun for me.

Because cumulative past precipitation alone is helpful but not sufficient, a next step after mapping cumulative precipitation over previous periods onto trails might be implementing a model based on a number of inputs using an science-engineering approach. This could utilize existing or derived data layers such as soil type, geology, slope magnitude, slope direction, topographic convexity/concavity, vegetation type, and trail type. For example, a trail I ride improves after a little rain and becomes unrideable (and stays that way for quite awhile) after a lot of rain. As an engineering geologist, I could help research approaches and develop the methodology if that is of interest. Years ago I encountered a paper that indicated there is a whole field called "military trafficability" that would provide one starting point.

A next step after that could be implementing a neural network that trains on the accumulated condition reports by MTB Project users as well as the output of the scientific-engineering model. Or the model output could be used to perform quality control on the condition reports, and the resulting "cleaned" condition report set be used to train the neural network.

The results of all this could be adapted to support the Hiking Project and Trail Run Project as well. One goal of the work would be to increase trail sustainability by providing people the information to use the trails where they will do the least damage. Using the example trail above, I suspect most people stay away from that trail after any amount of rain whereas they could actually use that trail into the season a bit, and avoid other trails that are more sensitive to smaller amounts of rain over a few days but more rapidly become rideable again a few days after more substantial storms.

If any of this is of interest, happy to talk to the MTB project staff further. I hope others on the forums will contribute their thoughts as well.

Thank you.

Preston

Nick Wilder · · Boulder, CO · Joined Jan 2005 · Points: 9,763

Thanks Preston - we've recently been talking about the basic idea (precipitation history) by coincidence.  Not sure if/when we'll do this, but it's on our list to discuss in the next month or so. 

Preston Jordan · · Albany, CA · Joined Jan 2018 · Points: 613

Great. Thank you. I think this feature will become more important as we further lose climate stability.

If you want help developing a subsequent science-engineering approach, happy to do so. And/or on the subsequent machine learning approach, my son .might be willing to advise as he is becoming a professional in that field.

Guideline #1: Don't be a jerk.

MTB Project Feedback & Suggestions
Post a Reply to "automated conditions"

Log In to Reply