Looking to produce slide path maps with GIS

[pbourdon]

^^^ That's what I'm working on doing right now. Only trouble is, I'm in Michigan. My prof is a geomorphologist by trade, but is backing this project 100%. Of course, right now its just a simple project required as part of a class, but part of the requirements of the project is that I write up a technical report of my findings so.......

[highangle]

Build on your work every year. If you play this right, you'll be a rock star and your GIS will be your thesis.


Start small and then scale up, with rigorous keepable work all the way through.

I would start by identifying objective standards of data collection that any observer can use, even a satellite. This is the key to scaling the dataset nationally and globally. Your statistics classes should make much of this clearer as you learn the methods of making valid extrapolations from representative samples of spatial, or any other observational data.
In this way, I think you'll find the Bayesian approaches used in the biological and husbandry sciences quite powerful tools for updating what we know about 'A' based on what we've learned from 'B'.

Don't shy away from something this cutting edge because you're young and untrained. Nathaniel Bowditch was made a Fellow of the American Academy of Arts and Sciences when he was 25. This is something well beyond a PhD or professorship.

As the goals become clearer and you refine your approaches, don't be surprised to find lots of support cropping up from odd places. NASA, for example, has a lot of interest in snow science and GIS...

On the flip side, don't be surprised to find other people viciously competitive, even to the point of being underhanded. If you think people will steal your work and stab you in the back for class standing, wait until you experience the grownup adult world of REAL hacks playing for real money. The higher up you go, the bigger hazard this becomes. Watson and Crick kyped info from 'competitors', Eisenhower and Bradley stole ideas from Patton and took the credit...

Counterintuitively, the best way to 'guard' your work is to make it as public as possible. This is why I support the community being involved in this at some level. There's a lot of specific knowledge here. Maybe the little insight stoke that gives you inspiration to charge a very rewarding line?

[Avmapper]

Hi Guys,

I have been working with GIS and Snow and Avalanche data for 10 years. pbourdon's ideas for Mt Baker were already done about 5 years ago by a couple of other students at Western Washington,in Alaska alot of work has already been done. Also to correct Cookie Monster the USGS DEM's that are available are very poor accuracy approximatly 70-100ft of vertical error. High res lidar is the best you can get but expensive and hard to come by they also need to be shot snow off and then snow on to be effectively used in GIS modeling. Just a heads up on this. Hazard modeling is ok on DEMs and I have done a fair bit as well. But maps tend to be turned all red in the zones most people look at so the detail is poor and we can just say that the general areas poses avalanche risk. This is all still evolving and there is much room for improvement the biggest challange is getting funding for this. I worked on the Colorado State Wide GIS Avalanche Database for 7 years before I saw a nickle. Last year we completed digitizing all the State Washington Highway Avalanche Altas's and produced GIS mapbooks for the corridors and have not gotten anything out of it yet. Currently we have over 4000 historically observed avalanche paths in our GIS database for Alaska, Arizona, Colorado, Idaho, Utah, Wyoming. We are currently working on California and then will do Montana. I have an article in the Feburary issue of the Avalanche Review on the best uses for GIS for snow and avalanche studies.

[gunniride]

Sounds to me like you are in an intro GIS course. Good for you, even if it a second semester so called advanced course, the subject matter discussed here is going to be incredibly work intensive.

My 2 cents is that you want to come away from the course understanding some of what GIS can accomplish. Scaling back a bit, not worrying whether someone has done the exercise previously, may help you. If you take an area that has identified slide paths, say one mtn. pass, digitize them, then do a simple analysis of slope and aspect, you'll walk away from the class with some great intro skills. Find it interesting and more GIS courses may be in your future, where you may dive a little deeper into the previous mentioned ideas.

A project idea I have been throwing around was to build a slide path map that also included aspect/elevation info to correspond with an Avi Rose. I always thought this would be useful for planning objectives and alternatives.
-E

Agreed, this could be a fairly simple, very usable product.

[pbourdon]

Sounds to me like you are in an intro GIS course. Good for you, even if it a second semester so called advanced course, the subject matter discussed here is going to be incredibly work intensive.

It is an "advanced" course of sorts, but on the other hand it is still very basic. There were pre-req requirements for the course requiring that we had all taken the "intro" course or the geomorphology course taught by the same professor that was quite GIS intensive. That being said, the course is very interdisciplinary with majors ranging from geology, and natural resources management to geography and earth science education.

If these maps have indeed been produced by some stu's at WWU, does anyone know where I can find them? Or the accompanying technical papers? For the purposes of this class, I'm not opposed to duplicating these works, not to mention the experience it would offer me.

Avmapper, on the one hand I find it surprising there isn't much funding available for this sort of work but then again most of the administrations since the late 70s and early 80s haven't focused their funding into such arenas much any way. You would think though that this kind of work, especially around ski areas, residential settlements, and highways would be very important and when combined with the analytical potential of GIS very useful in aiding forecasting work.

And if you're not getting compensated for any of this work, how do you get by then? What do you do in the summer?

Thanks for the input.

edit-doing an avy rose seems like it would be fairly easy to do and fairly useful as well. Still, field data are critical for creating the best possible maps.

Avmapper, looking for any help with your work?

[khakis]

I was gonna say you had a pretty solid grasp on the capabilities of GIS for just starting an intro class!
What about building a few hypotheses related to area of continuous high-risk slope(ie acres in the 35-55 range)/aspect/whatever based on your best data available and test the results against documented avy paths. Doing any remote sensing? You could maybe get Imagine to find the young vegetation for you and throw that in the mix. Regardless what you do, based on what you've written so far and the grad-level GIS classes I have taken I can almost promise you that you will have the best project in the class. Don't try to get too crazy with your analysis and spend more time on your presentation and making sure everyone understands your project/analysis. Beyond the basics GIS isn't all that difficult...getting others to understand its capabilities and figuring out how to use it to solve their problems is the hard part.

[CookieMonster]

"Also to correct Cookie Monster the USGS DEM's that are available are very poor accuracy approximatly 70-100ft of vertical error."

Are you referring to the DEM or high-resolution GEOTIFF data? I gather you mean "stock" without error correction?

[F#*k you cat]

I did a hazard analysis for a SWUG (South West ESRI User Group). Just after the Tsunami that hit Indonesia in 2004 I wanted to pose a hazard map for the western coast of the usa. I took raster elevation data from a coastal dataset converted it to pixels and created levels every 10 meters from below sea level to 100 meters above sea level. THis gave me a starting point to procede. I chose to use 30 meters ASL as the cut off for inundation since Banda Ache wave heights were approx 30 meters. I selected elevations from below sea level to 30 meters above sea level and created a hazard shapefile. THere was a bit of editing to show a continous connection to coastal areas, intercoastal waterways, river mouths etc. I then used the US census data for Population (2000) and clipped it to my coastal extent. thus giving me a term coastal population. Then it was easy to run an intersect of coastal pop vs elevation hazard. this provided coastal pop in the Tsunami hazard zone. in 2005 I came up with just over 16 million people. crazy crazy numbers. Yes it can happen here. I would love to see a discovery channel "it could happen tommorrow" on this subject. But who knew 250,000 people could die in Indonesia from the 4th largest ever recordable quake 8.9 to 9.2 depending on source.

I won a cool 1000.00 for the poster and ppt from SWUG.

P-16

My figure broke out the hazard into 4 areas, west coast inset, Pacific NW, NoCAL and So CAL. PM me for the powerpoint and pdf figure.

[highangle]

This may be unimportant for your project. A bit tired. My last post should really have discussed the "relationship" between alpha and beta angles instead of ratio.

An avalanche path has a starting zone, track, and runout. In theory, an avalanche atlas ( a map of slide paths ) should accurately map the starting zone, track, and runout ( along with other parameters such as aspect, elevation of starting zone, and a profile of the path ). Sometimes it's easy to determine the spatial boundaries of the starting zone, track, and runout zone. Sometimes it's very difficult.

If you want to make an avalanche atlas, you're faced with determining the spatial boundaries of the path, including maximum runout for large events that may have a very long return period. ( 1:100. This doesn't literally mean the avalanche returns every hundred years, it means the probability of the maximum event is expressed as 1 chance out of 100 for each year. )

In the past, dynamics models ( after Voellmy ) were used to try and figure out the maximum lower boundary of the path. However most of this work has been discounted. Modern avalanche dynamics models are focused more on granular flows. But forget about that for a moment.

Empirical, statistical methods are more popular these days, especially with the risk management aspect of avalanche hazard mapping. Anyway, the beta angle is the angle between the starting zone and the first ten degree slope in the runout zone. The alpha angle is angle between the furthest known runout distance and the starting zone. Finding the beta angle for any avalanche path is relatively easy. Simply find the first ten degree slope in the runout zone and measure the slope angle from that point ( following a straight line ) to the top of the starting zone.

So how do you determine alpha? It's much more difficult. Often field surveys are used in which vegetation is analysed and possibly the area around the runout zone is examined for damaged vegetation, pseudo-moraines, rocks embedded in trees or rocks found in other unusual places, and other signs of avalanche activity such as flow signatures in soil or on tree trunks. People have been known to take core samples from trees and look for reaction wood that indicates a tree was struck by an avalanche. Simple and complicated. Time consuming as well.

In some mountain ranges there is a relationship between the alpha and beta angles for many of the avalanche paths. I'm not sure that any such number has been determined for the Cascades. But ... If you know the beta angle, it is theoretically simple to use a regression equation to determine the alpha angle. Then you can find the maximum runout distance for large events. With this information in hand, you can create an accurate map of the avalanche path. You should look at Donna Delparte's paper: http://www.ucalgary.ca/asarc/publica...uthor/Delparte and the filename is: DelparteThesis_AvTerrainModelling.pdf. I'm sure you'll find her work quite fascinating.

There are other complications with constructing avalanche atlases. First and foremost, you have to understand terrain and geography fairly well because an avalanche path isn't necessarily simple. There are alpine paths, sub-alpine paths, channelized paths, and open paths. Sometimes avalanche paths have very nice trim lines and sometimes not. Starting zone and track are sometimes relatively easy to determine for sub-alpine paths. It just depends - starting zone geometry can be awfully complicated as well. Alpine avalanche paths often have no visible boundaries and can be very difficult to sort out.

Also keep in mind, avalanches aren't always neat confined by drainages, and dry flowing avalanches in particular often travel wherever the day takes them. Over ridges, over large obstacles, etc. The unpredictable behaviour of dry flowing avalanches sometimes makes the concept of an avalanche path a bit less important. It's very important to remember than an avalanche path only shows you where avalanches have run in the *past*. However, with avalanches, sometimes past behaviour is a good indicator of future behaviour, at least in some paths.

So for fun, you can remember that most avalanches do not run full path but an avalanche atlas should describe the full path.

Try this link for more information on avalanche atlases: http://www.wsdot.wa.gov/Research/Reports/000/008.8.htm

I didn't say this before, but thanks for the great post and great data. It shows a very high level of thought and knowledge on the subject.

I too am interested in the physical science basis for modeling avalanches.
I think a better understanding of the physics of snow on slopes combined with a GIS analysis of snow hydraulics and dynamics can work to objectifying and simplifying avi prediction.

I'm interested in what work has been done quantifying not just the fluid dynamics of a flowing slide, but the statics of slabs and wet slides - what structures can exist, and what it takes to make them break and slide.

Me and everyone else, huh?

[MATH warning!...]

I've read some Mears, and I am aware of the nonlinearity of the Navier-Stokes equations wrt convective accelerations and the frictions of packet (granular) flow that you have touched on, but I wonder has this been approached from a systems analysis pov?

We can look at a slide in 4 dimensional terms of Scope, and Resolution.
Scope is simply spatial boundaries (and boundary conditions) of our system, including time.
Resolution could be defined as the smallest definition between 2 alternative system configurations, ie spatial configs pre and post slide.

Now the problem becomes an exercise in smoothed particle hydrodynamics between 2 equations of states within a spatial geometry (or perhaps matrices thereof). Irrelevant degrees of freedom are integrated out and we come down to vector LaPlacian operators within a defined space.
Now we are talking about the relationship between a system in 2 points of time, a perfect condition for a walking algorithm. Moreover, it can efficiently be applied within a GIS to predictively model "slide space".

[/MATH Warning!...]

I know I sound like I'm talking some crazy esoteric shit, but I'm painting with a broad brush and, as you noted, that is the nature of viscid flow...

I think if we can make a valid physical model of a type of slide, we can apply appropriate statistical techniques to learn much more via GIS.

Thanks for listening.

[Avmapper]

You guys still aren't getting it as far as the USGS DEMs go they are not good enough to do forecasting they can give you an approximate idea f slope but still have between 70-100 ft of error. There have been avalanche deaths in pockets of 30ft in terrian traps that there is no way a model at this scale can show. Therfore mapping and modeling with this can be very misleading. To quote highangle "simplifying avi prediction" you can't simplify this type of procedure. There is too much spatial variablity and unless you have a hi res dem terrian model say less than 2 ft resolution and weather stations say every 5 kilometers your modeling will not be very accurate. pbourdon, as far as how I keep things going for many years I have run Avalanche Mapping as a slightly paid hobby project. I do full time GIS consulting of all types as my main bread and butter. It has only been the last 3 years or so that I have gotten a good amount of paying work for this avalanche stuff the rest of the time it ran on my map sales and sponorships. I run an internship program with CU boulder and have had 10 interns working on this project over 5 years. So the best way to do this full time is get a Masters degree in Snow Hydrology, Meterology or Civil Engineering with a emphsis in GIS and then get a job with a avalanche center that is progressive. The only other work for this is with the DOT's and almost all of them use their maintanence crews to do this type of work and involve very little GIS. Don't mean to be a downer on this but it is a tough business to get into and takes alot of experience of being out there(like ski partroling/winter guiding for 5-10 years or so) to understand variables before you just shove it into the GIS and think you are doing something good. Email me and I'll send you a copyof the work that the guys did on Mt Baker. Also go to my website and read some of the papers that are on there and the most current Avalanche Review has several articles on this topic.

[pbourdon]

Avmapper, thanks for the heads up. And to those of you that think I was considering making a avy model, that wasn't the plan. Cookie pointed out early on that modeling the types of flows that occur during a slide is very tricky. I mainly wanted to produce maps that depicted the slide paths in an area, and then do analysis to see what slopes and aspects are generally the most "active". Avmapper, you make a great point about the vertical error on most DEMS. I was working on a project for this prof on stuff we've been looking at on the Salmon River in Idaho and something he had me do was to check the elev. data off the DEM against the topo's elevation data. Basically, I had to manually calculate the elevation on the topo. Generally, the DEM we were using was no more than about 10ft off from the topo. But I know that can change significantly in steep terrain with a low res DEM. I'll send you an email later today. Thanks again.

[khakis]

...something he had me do was to check the elev. data off the DEM against the topo's elevation data. Basically, I had to manually calculate the elevation on the topo. Generally, the DEM we were using was no more than about 10ft off from the topo.

you don't suppose that the DEM and DRG were created from the same data, or that one was created from the other, do you?

[pbourdon]

^^^^ No, I had never given that any thought....

we were just covering our asses since its so much quicker to just point an click on the DEM. But we also wanted to see what our error was with the DEM...

[highangle]

You guys still aren't getting it as far as the USGS DEMs go they are not good enough to do forecasting they can give you an approximate idea f slope but still have between 70-100 ft of error. There have been avalanche deaths in pockets of 30ft in terrian traps that there is no way a model at this scale can show. Therfore mapping and modeling with this can be very misleading. To quote highangle "simplifying avi prediction" you can't simplify this type of procedure. There is too much spatial variablity and unless you have a hi res dem terrian model say less than 2 ft resolution and weather stations say every 5 kilometers your modeling will not be very accurate. pbourdon, as far as how I keep things going for many years I have run Avalanche Mapping as a slightly paid hobby project. I do full time GIS consulting of all types as my main bread and butter. It has only been the last 3 years or so that I have gotten a good amount of paying work for this avalanche stuff the rest of the time it ran on my map sales and sponorships. I run an internship program with CU boulder and have had 10 interns working on this project over 5 years. So the best way to do this full time is get a Masters degree in Snow Hydrology, Meterology or Civil Engineering with a emphsis in GIS and then get a job with a avalanche center that is progressive. The only other work for this is with the DOT's and almost all of them use their maintanence crews to do this type of work and involve very little GIS. Don't mean to be a downer on this but it is a tough business to get into and takes alot of experience of being out there(like ski partroling/winter guiding for 5-10 years or so) to understand variables before you just shove it into the GIS and think you are doing something good. Email me and I'll send you a copyof the work that the guys did on Mt Baker. Also go to my website and read some of the papers that are on there and the most current Avalanche Review has several articles on this topic.

My apologies for being a little too abstract. My point was that there are perhaps techniques that haven't been fully exploited in our fledgling science.

My reasons for asking about systems analysis as a possible perspective for slide modeling and prediction was not random. SA has shown some promise, as well as some interesting results when applied to water and gravity....





-----
On a more 'practical' note, i think a hell of lot of avi specific map info can be represented with hatching and hatchures themselves. In highest practice, hatchures can convey much about a fall line that contours can't do as simply or clearly.
Subtle shading can also represent many completely different dimensions about an avi zone, and not just spatial ones...




----







I admire your work and think your site is fantastic Doug. A gold mine. Thank you for years of difficult work.

I have to beg your forgiveness for talking to pb like almost nothing had been done in his direction. This might have been a little calculated for my own benefit....

[stalefish3169]

Anybody up in here know of any DEMs for Southeast Alaska w/ better than 30m resolution, and where I can find them?