Truth in snow reporting

[mondeo]

what kind of ruler is that? I'm pretty sure most rulers are divided into 1/8th of an inch not 1/10th.

Machinist's scales use 10ths

The study's most shocking conclusions do not depend on a comparison between resort-reported snow totals and government-reported snow totals. Rather, they observed significant increases in snow reporting on weekend mornings, a pattern that was not replicated in government reporting (because actual snowfall obviously does not correlate with the day of the week). How do you explain a statistically significant surge in reported snowfall on weekends? Does Ullr commute to the MRV every weekend?

It's not statistically significant.Their R^2s are crap, and most of their p-values aren't that high either. Look at the plot of mean snowfall with error bars, weekday vs. weekend. There's only one point where the weekday number falls outside the error for the weekend.

Oh, and it's not a good measure, as precipitation does depend on the day of the week. Due to the heavy automobile traffic and power generation during the week in the West, precipitation in the East actually is higher over the weekend than the average weekday because of higher levels of cloud-forming particulates. Probably helps Saturdays and hurts Mondays in Colorado, too.

 

[ski_resort_observer]

Don't get me wrong I think using this technology can be very useful. I first heard about the skier/rider mobile phone reports last spring. Here are just some of the questions that were discussed about this. Will they all carry rulers? Are they going to paint a scale on their $800 boots( that would be cool)? Will they measure both sides of the trail per elevation? Will people actually measure or just take an educated guess? More times than not, snowfall amounts can vary, on a windy night, quite alot. I look forward to how this whole thing works out.

All these reasons is are also why many times what is measured by the resorts's snow stakes do not match up with what is seen on any particular trail. If you get first chair and get to a resort snow stake and what your seeing is way off than what the resort is reporting then I think you have a valid beef and you should talk to someone in the lodge, if it's really that important to you. Mistakes happen plus sometimes mother nature has a way of changing things pretty fast.

This never got noticed much until the last few years with the huge spike in tree/glade
skiing but when it's really windy, new snow and sometimes snowmaking snow will blow off an open slope and end up putting down some nice powder in the woods upwind. The slope/trail be bullet proof but the woods near the trail can be sweet. Works best when the woods already have a decent base, of course.

 

[ta&idaho]

It's not statistically significant.Their R^2s are crap, and most of their p-values aren't that high either. Look at the plot of mean snowfall with error bars, weekday vs. weekend. There's only one point where the weekday number falls outside the error for the weekend.

This doesn't make sense to me. The relatively low R-Squared values don't negate the statistical significance, they just indicate there is unexplained variance.

Oh, and it's not a good measure, as precipitation does depend on the day of the week. Due to the heavy automobile traffic and power generation during the week in the West, precipitation in the East actually is higher over the weekend than the average weekday because of higher levels of cloud-forming particulates. Probably helps Saturdays and hurts Mondays in Colorado, too.

They acknowledge this possibility, see footnote 15, and control for it by including government-reported snowfall data in most of the regressions.

Quibble all you want, but its a pretty interesting study. I don't find the possibility that ski resorts exxagerate all that shocking, but the degree to which ski resorts appear to exxagerate strategically really surprised me.

 

[riverc0il]

The study's most shocking conclusions do not depend on a comparison between resort-reported snow totals and government-reported snow totals. Rather, they observed significant increases in snow reporting on weekend mornings, a pattern that was not replicated in government reporting (because actual snowfall obviously does not correlate with the day of the week). How do you explain a statistically significant surge in reported snowfall on weekends? Does Ullr commute to the MRV every weekend?

Exactly. This was a good control because they compare apples to apples in regards to changes in weekend versus week snow reporting on average. Great control and this is definitely the appropriate comparison. They did not compare NWS to the mountains but rather weekend variance to average from the mountain to a reputable control. Not perfect but it is no way shape or form as SRO paints the picture.

 

[ski_resort_observer]

Exactly. Not perfect but it is no way shape or form as SRO paints the picture.

Hey, I didn't paint the picture it is what has been reported on from a couple of diferent news sources. WCAX reported on it last night. the VSAA is saying it. If your saying they got it wrong then just what did these guys use as a comparison to the resorts numbers. Isn't this baseline data important before they can make any credible conclusions?

 

[AdironRider]

Im with SRO on this one. This study, IMO, is completely bogus unless they were out there compiling data at the same time the resort took their measurements in the same place. 30 MPh sustained winds will rip a foot of pow off an icy groomer in a couple minutes.

Luckily here in Jackson the snow reports are all automated. Everything is measured by computer in real time, including new snow, wind speed, snowpack, temp, the whole nine yards. Its pretty interesting to watch storm totals ebb and flow based on the conditions at the time.

In SRO's example, taking data from MRV and comparing it to a NWS station 25 miles away is complete BS. A crappy control is a crappy control, and really doesnt prove a damn thing.

Besides anyone who's skied on a pow day or two knows that sometimes when they report 8", its 18" in some spots, and maybe 2" in others. I dont think Ive ever been to a powder day where new snow totals didnt vary all over the mountain.

 

[ski_resort_observer]

Im with SRO on this one. This study, IMO, is completely bogus unless they were out there compiling data at the same time the resort took their measurements in the same place. 30 MPh sustained winds will rip a foot of pow off an icy groomer in a couple minutes.

Luckily here in Jackson the snow reports are all automated. Everything is measured by computer in real time, including new snow, wind speed, snowpack, temp, the whole nine yards. Its pretty interesting to watch storm totals ebb and flow based on the conditions at the time.

In SRO's example, taking data from MRV and comparing it to a NWS station 25 miles away is complete BS. A crappy control is a crappy control, and really doesnt prove a damn thing.

Besides anyone who's skied on a pow day or two knows that sometimes when they report 8", its 18" in some spots, and maybe 2" in others. I dont think Ive ever been to a powder day where new snow totals didnt vary all over the mountain.

Thanks for the update on JH's way of doing things, wonder if my old friend Jake Elkins is still running the snow/avalanche reports. I know he has moved as his office used to be inside in the old tram building right next to the monster bullwheel. I think I even have a pic of some of the equipment he used in my AZ gallery from a full page newspaper story I took the pics for on how western resorts handle within boundary avalanche predictions.

Anyways, I think this was around 1990 and JH was ahead of the curve even then. They had live camera's at each snow stake on Rendezvous so all he had to do was look at the monitor to see each snow stake.

I think it's understandable as places out west have to worry about avalanches as a major safety issue in additional to just the total snowfall.

 

[AdironRider]

Not sure if hes still there or not, but you can check out the whole system at jhavalanche.org.

 

[snoseek]

This

http://www.snowbird.com/snowcam.html

Is truth.

Doh, I just checked their snow report and the YTD number is only 20 inches!

 

[polski]

Hey, I didn't paint the picture it is what has been reported on from a couple of diferent news sources. WCAX reported on it last night. the VSAA is saying it. If your saying they got it wrong then just what did these guys use as a comparison to the resorts numbers. Isn't this baseline data important before they can make any credible conclusions?

Im with SRO on this one. This study, IMO, is completely bogus unless they were out there compiling data at the same time the resort took their measurements in the same place. 30 MPh sustained winds will rip a foot of pow off an icy groomer in a couple minutes.

I'd respectfully suggest reading the report itself -- even just the intro -- rather than relying on reports about it by a news station (though I watched a stream of the WCAX report last night and didn't notice they'd made this mistake) or ski area association (interested party).

As I said in my OP and ta&idaho and Riv tried to explain too, the study's key findings were NOT based on direct comparison of resort obs and contemporaneous NWS/non-mountain obs. The researchers compared 1) ski resort reports on weekends vs ski resort reports on weekdays, with 2) nearby government weather station observations on weekends vs government obs on weekdays. They found the resorts reported 23% more snowfall on weekends than the resorts did on weekdays (p = 0.013, i.e. 99.7% confidence that the difference wasn't because of chance). In contrast, there was no statistically significant difference between govt obs on weekends and govt obs on weekdays.

Here and elsewhere I've also seen the claim that the study inappropriately compared snowfall totals reported by resorts with contemporaneous NWS reports from 25 miles away. First, see the preceding paragraph. Second, read the report at p. 6-7. The researchers used actual observations from 1) up to 20 government weather stations within 25 miles* plus 2) the HIGHEST snowfall estimate from the 25 nearest data points -- which would be no farther than around 2.5 miles away -- from the Snow Data Assimilation System (SNODAS), an NWS model used mainly in hydrography.

* one detail in the report I do question as I look over this again: It says the NWS obs come from stations "at elevations within 500 feet of the resort summit." I'd have to think in some if not many cases there would be no NWS stations that fit that bill.

 

[polski]

The study seems incomplete. To get the full picture, they should have determined if exaggerated snowfall reporting actually makes casual skiers more likely to pick one resort over another. Such as, "Honey, Killington got 10 inches of snow, but Sugarbush only only got 8. Let's go to Killington!"

I agree that's an important variable but the authors noted it's pretty much impossible to get skier visit data, especially on a day-by-day (or weekend vs weekday) basis, as most ski resorts are privately owned.

It also would be interesting to know how surface condition reports (e.g. one resort's claim of pp vs another's forthright acknowledgment of lsgr) affect destination decisions, but again there's the problem of no skier visit data, and it also would be daunting logistically to try to validate the condition reports. Although I'd happily volunteer for that data collection!

 

[polski]

I disagree that social media, wifi, and live reporting will make resorts more honest. [lot of real good points snipped]

Plus, Joe Average skier who is the target market of ski resorts is much less likely to be involved in social networking that provides a wide range of opinion on how good the snow reporting is. These folks don't even buy their ticket and/or get to the mountain until the new snow is tracked out any ways. Joe Average skier is not a good powder skier and is just fine not getting first tracks. Simply put... the customers that provide the ski resorts with the most profit either don't care or can't be bothered about the accuracy.[snip]

A few thoughts:

- the Dartmouth study found a significant drop in exaggeration after that new iPhone app was rolled out in January. But they'd also found there tends to be less exaggeration during holiday periods and spring anyway. Without plowing back into the study, I wonder if there may be a confound here. Time will tell.

- I think the iPhone may be more of a "disruptive technology" vis a vis resort reporting accuracy than earlier forms of online/social media. You're right about the important differences between Joe Average and powder pigs but if Joe Average starts posting from his iPhone on the lift that "no way Mt. X got as much snow as they claimed" or (something any skier would be qualified to report accurately) "they said pp but bring your ice skates - it's boilerplate," well, I could see resorts being scared straight by that.

- I think there are some instances where message boards et al have already made a difference. For example, this is purely anecdotal but I'm under the impression that in the past few years Jay has made a serious effort to shed the image of the "Jay Inch," whether or not that was warranted in the first place. They're certainly among the best mountains around here in being transparent about potential wind holds, from what I've seen.

 

[riverc0il]

I stand by my statement. Unless I have misunderstood the research and their methodology, it is a sound comparison. What they did NOT do was compare the mountains to a NWS station. They did NOT do that as a comparison. They compared relative differences within the same sampling compared to other differences in a different sampling. Thus, they compared the daily differences on certain days of the week rather than the exact snow totals. They are saying that ski areas report x% higher snow fall totals compared to other days of the week compared to NWS reporting which does not have a significant different of snow falling on weekend days compared to midweek days.

Put more simply, the calendar is an arbitrary man made system for telling days of the week. While it is fact that the Earth revolves around its axis 365.25 times in one full cycle around the sun, it is not "fact" that there are 7 days in a week. Weather should, over a long enough time, average out so that no one day has statistically higher chance of having above average snow fall than any other day (assuming like days on the calendar are compared year to year). Therefore if NWS records suggest that snowfall is averaged out with equal chance of big snows on say Wednesday as Saturday but the ski area data suggests that Friday and Saturday have significantly higher chance of big snow fall than say Wednesday, something is off. That is what is being compared as I understand it.... not the snow fall totals themselves. Perhaps I (or the news media?) misunderstood the study but that is what I understand it to have for a conclusion.

 

[riverc0il]

A few thoughts:

- the Dartmouth study found a significant drop in exaggeration after that new iPhone app was rolled out in January. But they'd also found there tends to be less exaggeration during holiday periods and spring anyway. Without plowing back into the study, I wonder if there may be a confound here. Time will tell.

- I think the iPhone may be more of a "disruptive technology" vis a vis resort reporting accuracy than earlier forms of online/social media. You're right about the important differences between Joe Average and powder pigs but if Joe Average starts posting from his iPhone on the lift that "no way Mt. X got as much snow as they claimed" or (something any skier would be qualified to report accurately) "they said pp but bring your ice skates - it's boilerplate," well, I could see resorts being scared straight by that.

- I think there are some instances where message boards et al have already made a difference. For example, this is purely anecdotal but I'm under the impression that in the past few years Jay has made a serious effort to shed the image of the "Jay Inch," whether or not that was warranted in the first place. They're certainly among the best mountains around here in being transparent about potential wind holds, from what I've seen.

I hope they are not extrapolating a result from one season worth of data regarding implications and effects of an iPhone app. :blink: While that is questionable as it is, I can not imagine that they were able to control other variables that could have caused areas to report more accurately or perhaps it was just a fluke

Regarding someone punching into their Twitter to "bring your ice skates" we are already seeing those posts on internet forums and blogs the same day. Those types of reports are only going to help folks making a late decision on where to go... by default, those people are not going some where due to powder because by the time they make their decision and drive to the mountain, the powder is gone.

Another factor to consider is that it is standard acceptance in the industry to believe that snow reports are at best questionable and at worst out right lies. Is a Twitter post really going to tell people something they did not already know?

But then we get into Joe Average skier who may not know that reports are dubious at best. But Joe Average skier booked three months in advanced and is skiing regardless and can't ski powder so really does not care.

The only time these totals really matter is when the snow has stopped falling before bed time and the ski areas update at 10p or at 6a. Otherwise, who is really looking at reports and making decisions based on actual snow reported? Any one that cares about powder is usually making their decision based on the weather forecast. It is a rare day that I wake up on a powder day and think "I have no idea where to go and I need snow reports!" Occasionally, I will wake up and have a choice to make between two areas and may do so based on snow phone reports after I am on the highway. But a difference of a few inches is not going to sway my decision.

Any ways, bottom line is that this is a really interesting report but I think it is more likely that public perception, internet forums, and non-instant online communication is much more likely to contribute to honesty than iPhone Apps and live reporting. The backlash comes from trip reports that smoke ski areas. I think this happened at Jay, they have gotten a LOT better (though I think the perception is going to be hard to completely defeat). We all have seen things change due to things said on here at AZ. The more people that get behind an image problem issue, the more likely change will occur. Therefore, I would suggest such change would result form the masses and not an individual live blogging.

 

[polski]

well, what the iPhone apps do is make it easier for many individuals to do live reporting. Granted that isn't going to make much difference that same day but I'm thinking (hoping?) over time it helps grow the size of the "masses" who, one way or another, are willing to call out any ski area that exaggerates snowfall/conditions, thus encouraging better behavior.

Yes, the conclusion on the iPhone "shock" comes from one season -- actually, little more than half a season. I think that app was rolled out on Jan 9 of this year. I'd say that isn't one of the paper's stronger findings, even if it was statistically significant.

 

[billski]

skireport.com

Snow report exaggeration reportedly fell sharply since January, when Skireport.com rolled out an iPhone app for skiers to post their own reports. "But first-hand reports spike only at resorts with adequate coverage from AT&T's data network," Zinman and Zitzewitz wrote, "and these covered resorts experience a disproportionate post-launch drop in exaggeration."

Lots to chew over here.

Well, I just spent some time reviewing skireport.com input from iPhones into various reports, ranging from Stowe to MRG to WaWa to Vail and Breckenridge, Magic.

Bottom line is that these "reports" are loaded with trash, garbage and agendas. A few are constructive, but many like to bash the area they hate, pump up who they love and provide opinions about areas they have never been to. There is no accountability. In my spot-check of various areas I've recently been to, the majority of corresponding comments are off the mark. What good is a report of 6 feet in Tahoe on the Stowe thread other than to gloat and antagonize? There is no way I would trust anything posted there. In it's present form, skireport.com is veritably useless.

 

[ta&idaho]

Well, I just spent some time reviewing skireport.com input from iPhones into various reports, ranging from Stowe to MRG to WaWa to Vail and Breckenridge, Magic.

Bottom line is that these "reports" are loaded with trash, garbage and agendas. A few are constructive, but many like to bash the area they hate, pump up who they love and provide opinions about areas they have never been to. There is no accountability. In my spot-check of various areas I've recently been to, the majority of corresponding comments are off the mark. What good is a report of 6 feet in Tahoe on the Stowe thread other than to gloat and antagonize? There is no way I would trust anything posted there. In it's present form, skireport.com is veritably useless.

Sounds a lot like most user-created internet content, including forums like . . . :wink:

 

[billski]

Sounds a lot like most user-created internet content, including forums like . . . :wink:

Agreed. So, the argument that forums and skireport.com drive better resort reporting behavior is very weak.

 

[ta&idaho]

Agreed. So, the argument that forums and skireport.com drive better resort reporting behavior is very weak.

Let me get this straight--you're saying that user-created internet content is unreliable, but in assessing whether the introduction of this iphone application influenced ski area reporting, we should trust user-created internet content (your anecdotal comments) rather than a published report from two Dartmouth economists analyzing actual data? Hmmm...

Since we're pontificating, I would argue that the quality of the user-created content does not have to be universally high for it to provide useful information (and discipline ski are reporting). This website, for example, has tons of useless information (including many of my own posts), but it also contains nuggets of incredibly useful information and *seems* to have influenced at least some ski resort management decisions.

 

[billski]

Let me get this straight--you're saying that user-created internet content is unreliable, but in assessing whether the introduction of this iphone application influenced ski area reporting, we should trust user-created internet content (your anecdotal comments) rather than a published report from two Dartmouth economists analyzing actual data? Hmmm...

Since we're pontificating, I would argue that the quality of the user-created content does not have to be universally high for it to provide useful information (and discipline ski are reporting). This website, for example, has tons of useless information (including many of my own posts), but it also contains nuggets of incredibly useful information and *seems* to have influenced at least some ski resort management decisions.

Not at all. Go back and re-read what I wrote:
"the argument that forums and skireport.com drive better resort reporting behavior is very weak."

On the flip side, it is much easier to drop incendiary remarks and run untraceably on the iPhone app than on a forum which requires you to register. You can still hide behind the mask if you care to, but you now have a history of remarks by which I judge your credibility. Consequentially, I am much more willing to take serious the remarks of someone with an credible history.