Results 26 to 50 of 57
Thread: online slope mapping tool
-
08-05-2013, 04:20 PM #26
This site accesses the averaged temperature and precipitation for Western States for Western Weather stations, but no wind information:
http://www.wrcc.dri.edu/htmlfiles/res.sno.html
This is where the money is-water so the information is there. How the wind blows and the way it lays in is less of a concern to Agriculture.
Look forward to checking out Canada.
-
08-05-2013, 05:38 PM #27Registered User
- Join Date
- Feb 2012
- Location
- Truckee, CA
- Posts
- 64
Thanks. I really want to take a crack at integrating SNOTEL data for the western states this fall, which gives me snowpack information but mostly no wind. It looks like the COOP data referenced by that link may also be able to serve as another source for snowpack data, but I'll have to play around with it more.
-
08-06-2013, 02:53 PM #28Gel-powered Tech bindings
- Join Date
- Dec 2004
- Location
- Amherst, Mass.
- Posts
- 4,684
Whoah, that should have clicked for me when I was reading your blog -- of course, such a great on-line mapping tool would be drawing on those MIT mad skillz!
Great to see you putting your talents to good use there, and with backcountry skiing, and with the SAR volunteering.
Too bad though about the limited market for such a great mapping tool. On the one hand, all this technology greatly enhances detailed route planning, but on the other hand, it also encourages very simplistic approaches along with atrophying old-fashioned navigational skills.
(I'll write more via email...)Mo' skimo here: NE Rando Race Series
-
08-06-2013, 07:54 PM #29Banned
- Join Date
- Dec 2009
- Posts
- 7,167
why would you need more info than what the avy forecast center for the chic chocs puts out. in one neat and tidy paragraph/page you get recent weather and snow data (all you really need for a maritime snowpack) with all the pertinent info for planning a trip there. just like pretty much any other region in the us/Canada where daily/weekly forecasts are readily available. pretty simple. and then choosing terrain is just that. avy terrain is avy terrain and non avy terrain is not. not hard.
rog
-
08-07-2013, 05:25 AM #30
Your right about the chic chocs, I knew that. Brain fart
-
08-07-2013, 06:28 AM #31Banned
- Join Date
- Dec 2009
- Posts
- 7,167
you goin? I haven't been since 06' but have about 40 days of touring under my belt up there from 98' till then. such an amazing place with some wicked good "how the fuck is my body gonna contort to link turns on that pitch?!" kind of skiing. land of the roll over into the abyss skiing
rog
-
08-07-2013, 01:01 PM #32
In order to model radiation flux, you'd need to use ray tracing. Not trivial
Sent from my Paranoid Android using TGR forums.
-
08-07-2013, 01:09 PM #33
Modeling wind is exceedingly non-trivial, even if you try to use a very simple model for "predominantly lee slopes". Have you ever looked at weather station data?
Here is a scratch-the-surface discussion:
http://avalanchesafety.blogspot.com/...t-in-wind.html
Sent from my Paranoid Android using TGR forums.
-
08-07-2013, 01:54 PM #34Registered User
- Join Date
- Feb 2012
- Location
- Truckee, CA
- Posts
- 64
Please tell me if I'm overlooking something:
* When developing the "view from here" feature, I wrote code to compute the dot product between an arbitrary vector and a slope's normal vector. In that case it was using aspect/angle data to do relief shading, but I have the data and the code can be repurposed.
* I had to develop ray tracing code for the viewshed layer, since that's all about showing what areas are visible from a given point.
So given an angle and altitude for the sun, I can determine which areas are visible v. obstructed and how off the the slope's normal vector is from the sun's rays. To show overall shading for a day, I'll average a number of samples with the sun in different positions. At this point I have no idea how many samples I can use before running into performance issues, or how many I need to do in order to see reasonably accurate results. 1 per hour?
I haven't dug into all the details so I don't yet know if solar warming is linear or nonlinear relative to the dot product or how the sun's intensity changes with solar altitude. The end result won't be perfect, but I think I should be able to make something that's reasonably accurate for recreational use.
-
08-07-2013, 03:29 PM #35
No, you certainly haven't dug into the details.
Where do I begin? You'll need to learn *a lot* more if you think computing raw diffuse lighting and/or viewshed determination is the same as raytracing! I also think you'll want to be a bit more careful before you rather boldly state that you can use a nakedly inadequate diffuse lighting algorithm to model radiation flux in a fashion that is "reasonably accurate for recreational users". I'll pose a few questions...
1. What happens when light strikes a surface? Does it just vanish? If it doesn't vanish, what happens? If it doesn't vanish, and perhaps is reflected and strikes surfaces nearby, do you think that certain geographic features could tend to focus these reflections and significantly increase local warming, while perhaps other geographic features might tend to diffuse the reflected light. Furthermore, is it important to consider the albedo of the snow, which varies across space and time? Oh, and what about shadows? Computing the diffuse term does not account for shadows, which is perhaps an important detail. Would you consider any local flora in your computations? What about clouds? Do they make a difference?
The naive diffuse lighting algorithm you suggest is in no way capable of providing a reasonably accurate model of solar warming. It's not the right approach if you want to solve the problem.
Sent from my Paranoid Android using TGR forums.
-
08-07-2013, 05:01 PM #36Banned
- Join Date
- Dec 2009
- Posts
- 7,167
well there ya have it, next!
rog
-
08-07-2013, 06:10 PM #37Registered User
- Join Date
- Feb 2012
- Location
- Truckee, CA
- Posts
- 64
I haven't made any claims, bold or otherwise, because I haven't built anything. I'm throwing ideas out and looking for feedback. And I never stated that I was trying to perfectly model solar warming or radiation flux - your term, not mine - merely the amount of sunlight a slope gets.
Obviously you have a lot of knowledge about avalanche modeling. But, and please don't take this the wrong way, a lot of your posts seem to follow the theme of "modeling is hard" without addressing the point at which they're good enough to still be useful. Take your post about wind loading above. It's hard to figure out exactly which aspects are getting loaded to what extent in which locations, but after a windy night the avy center doesn't say "shit guys, beats us." It's (hopefully) understood that just because they predict wind loading on NW-SW slopes doesn't mean that the wind couldn't have been blowing in other directions in isolated spots.
So yes, Avy Gulch on Shasta has a solar oven effect going on, and no, I'm not going to model that. As stated earlier, I'm not going to incorporate vegetation. And I don't see what relevance clouds have. I'm not saying "turn this layer on and you'll magically see which slopes have warmed and stabilized and which ones haven't." I don't know about you, but I can't look at a topo map and figure out how many hours a day a slope gets sun for, just like I can't look at a map and tell a 20 degree slope from a 35 degree one without measuring. I'd like to build an easy-to-understand tool which will help answer that question; there's still a lot that humans will need to fill in for themselves.
The slope shading layer I built out isn't perfect, and when I shopped it around to various avy & SAR people before making it live, the reactions ranged from "awesome" to "this is incredibly dangerous and going to get someone killed". There's not much consensus over the right balance between providing people with tools and making sure those tools are not over-trusted / overused.
If your overall point is that I shouldn't move forward with this because it will make people inaccurately overconfident about slope stability, then I think that judgement should at least be reserved until you see how it's presented. If reflected light from surrounding terrain is really so important that the amount of time a slope spends in direct sun has little correlation with how quickly it stops being soft powder or slushes up on a spring day, then can you point me to any sources on that? It doesn't match my personal observations.
-
08-08-2013, 01:23 AM #38
Being capable of GIS programming in some fashion does not a scientist make.
It's not acceptable to make something for recreational use, where arguably the standards should be the very highest, without a deep and abiding respect for the science required to do an excellent job of it. You may be a mediocre programmer, or a good programmer, or even a great programmer, but "good enough for recreational use" is really laughable, and honestly a bit detestable in the way it reveals your willingness to perform experiments on people ( while not even realizing that you are doing so ). Recreational users aren't your guinea pigs, and most of the content of your response to my original remarks indicates complete and utter ignorance on your part with respect to the most BASIC tenets of radiometry. That's my very direct way of saying that you should really think long and hard about your credentials before you make something and feed it to the public.
The bar here is awfully high if you want to do REALLY GOOD science that delivers USEFUL, ROBUST results. On the other hand, you do have a right to the utterly incompetent belief that you can solve an incredibly difficult problem with a single dot product. Buyer beware I suppose.
Merely the amount of sunlight a slope gets?
What is it with you GIS programmers?
A diffuse reflection model by itself simply does not even pretend to approximate, to any degree of accuracy whatsoever, the amount of sunlight a slope gets. There are other details: since snow has a high albedo, reflected sunlight is an important factor, as are shadows. For example: are you aware that a diffuse reflection model may indicate exposure to sunlight for surfaces that are actually in shadow? In simple terms, diffuse reflection will be quite absolutely dead wrong a lot of the time. Doesn't this seem like an important detail? Sorry if I'm getting scientific about it, but if you want results that are "reasonably accurate", then you'll just have to consider the physics in very rigorous fashion. If you can't handle the physics, you will produce pseudo-science.
A lot of my posts seem to follow the theme of "modeling is hard" without addressing the point at which they're good enough to be useful?
Are you trying to make me laugh?
I'm terribly sorry; we wouldn't want things like science and reality to intrude, now would we? In my opinion, when I provide information about avalanches, as a public service, it's my sacred duty to make damn sure that I provide information that is absolutely accurate. I don't speculate because that's the same as making something up from thin air. The hard science is as follows: wind modeling is incredibly difficult, and the science is not yet good enough to make small scale models that are suitable for recreational use. There you have it: that's the science; that's the plain truth. You'll note that my posts, that post in particular, says that local observations remain the gold standard in determining how wind has affected the snow cover. There's a good reason for this, and there's also good reason for why I spend a fair amount of time trying to illustrate the pitfalls in using what amounts to rank speculation about wind deposited snow during the process of backcountry avalanche forecasting.
While the public avalanche forecast might not literally say "shit guys, beats us", very often the public avalanche bulletin is heavily disclaimed, and may even include information about the forecaster's confidence. Whether or not you LIKE the idea that it's not yet possible to accurately model wind flow and snow deposition beyond the synoptic scale, the fact is that it simply is not possible to model wind flow and snow deposition at small scales.
Would you rather just make it up as you go along? I only ask because presenting a diffuse lighting model as something that is capable of "modeling the amount of sunlight a particular slope receives" constitutes making it up as you go along. Furthermore, you might not know enough about radiometry to even be expected to know the serious limitations of diffuse reflection models relative to a complete model of global illumination, but that in no way exempts you from being responsible for working within your competencies and/or extending them as needed.
I understand that you can't look at a topo map and figure out how many hours a day that a particular slope receives sunlight, but the tool you're planning to write certainly won't be able to come up with that information either. Further, considering only diffuse reflection will produce results that are useless, misleading, and utterly worthless.
Yes, but slope angle shading can be done with a fairly high degree of accuracy. I mean, who are we kidding, assuming you have the surface normal, you can get the slope of a particular surface with a single subtraction.
No, my point is that modeling wind flow, snow deposition, and global illumination are exceedingly non-trivial tasks. There still isn't much science for accurately modeling these phenomena at small scales, even with the billions that have been invested in numerical weather simulation ( which is traditionally the field where this research is conducted ).
***
My source btw: The Avalanche Handbook. You might note the text on page 39: Radiation balance at snow surface. Incoming, absorbed, and reflected solar radiation is short wave ( visible and ultraviolet ). The importance of terrain shape relative to solar radiation is also discussed later in the book.
***
One more thing:
Where did I say there was little correlation? I said indirect illumination was very important as well.
I'll repeat myself for your edification: snow has a high albedo, which means that it reflects most of the light that strikes its surface. Does that light just magically vanish? Of course not. Direct sunlight is of course very important, but there are many places where reflections become strong enough to significantly amplify first order effects.
This is something that someone with even a rudimentary understanding of physics understands.
-
08-08-2013, 05:27 AM #39Banned
- Join Date
- Dec 2009
- Posts
- 7,167
bated breath...............
rog
-
08-08-2013, 11:54 AM #40Climb the mountains and get their good tidings. Natures peace will flow into you as sunshine flows into trees. The winds will blow their own freshness into you, and the storms their energy, while cares will drop away from you like the leaves of Autumn. - John Muir
"How long can it last? For fuck sake this isn't heroin -
suck it up princess" - XXX on getting off mj
“This is infinity here,” he said. “It could be infinity. We don’t really don’t know. But it could be. It has to be something — but it could be infinity, right?” - Trump, on the vastness of space, man
-
08-08-2013, 12:10 PM #41Banned
- Join Date
- Dec 2009
- Posts
- 7,167
squeeze me? popcorn? this is one long ass intermission. waiting for cookie mans reply. maybe jonathan can chime in.......
rog
-
08-08-2013, 02:52 PM #42Registered User
- Join Date
- Feb 2012
- Location
- Truckee, CA
- Posts
- 64
I thought it would be it clear that I'm explicitly not trying to do science. Because yes, not only would the end product would be horribly inaccurate, people would be left guessing at where and why it's wrong. I'm trying to make it easier for people to graphically understand the answers to some very simple questions, with as little interpretation and modeling from me as possible.
Go back and read the post where I describe what I'm trying to do. You keep throwing the term diffuse reflection model around, but it's your word, not mine. The primary reason I want to build this is to figure out when slopes are in the sun and when they're in the shade. Most of the work is going to be spent tracing outward at the angle/altitude of the sun to see if surrounding terrain blocks it, and my naive understanding of physics is that it's pretty obvious when you're in the shade. It would be nice to incorporate the sun's strength and whether light received is direct or glancing, but if that aspect is truly a horrible idea, I will definitely reconsider.
Oh come on. You very smugly implied that I was a blathering idiot for daring to think that the amount of direct sunlight a slope gets could, on its own, be a useful piece of information. The bottom line is that people make decisions based on direct illumination all the time, because it's easy for us observe and understand. Overall it seems to serve us well.
I'm looking to take information that people learn through observation and act on every day, and make it available without standing in the field and watching a peak to see when the sun strikes it. Are you really, seriously, claiming that knowing whether a slope spends all day in the sun or in the shade is "useless, misleading, and utterly worthless"? Or that the same is true for knowing whether it receives 4 hours of direct midday sun or 4 hours of glancing light right before sunset?
I'm going to build some kind of sunlight shading layer, because it's useful for all kinds of activities in all kinds of seasons to know when the sun will come up from behind a ridge or disappear for the day or how many hours of sunlight your campsite will get. Until I build a prototype and start testing it, I don't know exactly what form it will take or what information (e.g. sun's intensity) will get incorporated or left out. I do know that it's not going to be, or be presented as, ski or snow specific.
If you have advice, including on what to leave out, I would like to hear it. I'm still trying to sort out what it does and doesn't make sense to show, but from my perspective it feels like you're more interested in beating me over the head with the idea that the issue is so complex I shouldn't even touch it. I'm sorry, but I simply don't accept that the only two options are build an awesome scientific model or don't bother telling people when the sun goes behind a ridge.
Edit: I really do want to make it clear that I'm trying to have a discussion and looking for information to help me make a better product, including finding out what I don't know. Based on this conversation, I will be doing some more research before altering the shading based on the sun's angle to the slope, since that was never the focal point. Unfortunately for Rog I don't see the point in drawing this out, so if you want to keep dumping on me then have at, but I'm not going to argue back and forth over it.Last edited by mattyj; 08-08-2013 at 03:44 PM.
-
08-08-2013, 09:48 PM #43Banned
- Join Date
- Dec 2009
- Posts
- 7,167
DAMMIT! DON'T STOP NOW, YER ON A ROLL!!!
rog
-
08-08-2013, 11:43 PM #44
Seems that some adjustment for time of year would be useful. Even if only by week. What you are doing reminds me of a photography tool I read about that gives where & when the sun and moon will rise to help plan specific shots. I think it's called the photographers epheremesis or something similar.
-
08-09-2013, 01:51 AM #45
So, in summary, you're trying to create a physical model of light propagation without doing science? Are you dense? Do you know how ridiculous that sounds?
Maybe I should just repeat myself. You can't even make a ballpark computation about how much light a surface receives without considering the problem from the perspective of global illumination. What you outline in your initial response, the post in which you discuss dot products and so forth, won't cover shadows or reflections, which are a critical component of the picture. Diffuse reflection is not sufficient, and will result in serious, intractable errors in your "model".
Is that so hard to understand? You can use Google, right?
The term diffuse reflection is not my term either; it's from radiometry, and it's a very common term in computer graphics. You can use Google, right?
You cannot figure out which slopes are in the sun and which slopes are in the shade without considering global illumination. Radiometry, which is science, is an absolute requirement.
No, I didn't imply that you were a blathering idiot. I believe that I was quite direct; it's not my fault if you blather.
I'll repeat myself. Diffuse reflection does not include shadows, so you won't get the results you want. A radiometric perspective, which again, is science, is an absolute requirement.
How will you know how much light a surface receives if you don't include shadows? Diffuse reflectivity ( point on surface, vector to light ) does not include shadows. You think it includes shadows because you can use diffuse shading to determine whether or not a surface faces the light source. But knowing when a surface faces the light source does not account for shadows. Shadowing occurs when a surface faces a light source, but is occluded by another object located between the lit surface and the light source. If you hold your hand below a light, you'll notice a shadow. You'll also notice that the surface still faces the light source. So yes, diffuse reflection is utterly worthless for all the purposes you describe.
How will you know if the sun comes up from behind a ridge if you don't include shadows? I've already explained many times now that diffuse reflectivity does not include shadows.
If you don't build a reasonably scientific model, you won't be able to tell people when the sun goes behind a ridge. I've been telling you that diffuse reflections are not sufficient for several posts now.
You're going to feel, and quite deservedly so, like a complete and utter idiot when you start Googling and find out that I'm absolutely correct and that my advice has been quite good.
I'll be waiting for that post.
-
08-09-2013, 01:55 AM #46
-
08-09-2013, 04:14 AM #47Banned
- Join Date
- Dec 2009
- Posts
- 7,167
-
08-09-2013, 12:00 PM #48Registered User
- Join Date
- Feb 2012
- Location
- Truckee, CA
- Posts
- 64
I can't tell if you're intentionally being dense in order to provoke me or if you really don't understand what I'm trying to do. It feels like the former, which is too bad since your other posts at least gave me some things to think about and research. I 100% agree that doing a relief shading with the sun to the south instead of the northwest provides nothing useful. I do have questions about how/if to incorporate that into a viewshed analysis showing which points have line of sight to the sun.
When I have a working prototype in the fall I'll post it up here, and at least then you'll have something concrete to rip apart.
-
08-09-2013, 12:16 PM #49Registered User
- Join Date
- Feb 2012
- Location
- Truckee, CA
- Posts
- 64
Photography is one of the uses I'm hoping to cover. This will be a dynamic layer similar to the DEM Shading and Viewshed Analysis layers currently on caltopo. Having someone enter an arbitrary date and time and showing which areas are lit at that point is easy, computation wise, so it will definitely be included. Averaging a number of dates or times to show how the sun's reach changes at 5pm over the course of a month, or how much sunlight a face gets over the course of a day, requires a lot more resources. The viewshed layer (which I'll essentially be repurposing for this) is very CPU-intensive and one of the things that already led me to upgrade my server. I'm not yet sure how many viewshed computations I can average out while still having the tiles load in a reasonable time.
-
08-09-2013, 03:21 PM #50
Here's something I did several years ago for modeling solar warming. The current implementation is much more advanced. Anyway, the snow scene and the water scene render in real-time, including shadows and reflections.
***
The snow scene uses standard radiometric techniques to approximate the global illumination of the scene. This includes ambient, diffuse, and specular lighting, along with shadows, and a few other goodies. Unfortunately, there's just no way to extract snowpack temperatures from the rendering because there isn't much enough science to work at this scale. There is also the problem of validating any such model. We did use the SWARM model ( Google it ), which is rigorous, but the results were still not tenable from a scientific perspective.
The second snow image ( with black background ) shows the remapped warming values. As it turned out, reconstruction of snow temperature values, even using SWARM, was very unscientific. So it was scrapped.
The water images show light propagation modeling at a very small scale. It's an excellent model, which uses volume rendering techniques for the rocks, ray-tracing for the reflections, and highly accurate light propagation down to very small scales ( pebbles ). It certainly could be applied to larger scales. Once again, there still isn't enough science to actually compute snow temperature values.
***
Just in case you thought I was being dense.
I'm TRYING to tell you how to make something that will work, because I've actually built software that does what you want to do.
Original Links:
http://farm4.staticflickr.com/3800/9...6f0395f7_o.jpg
http://farm3.staticflickr.com/2845/9...a864fe85_o.jpg
http://farm8.staticflickr.com/7315/9...c2e843a7_o.jpg
http://farm3.staticflickr.com/2826/9...fed2ef7b_o.jpg
Bookmarks