EpicSki › The Barking Bear Forums › On the Snow (Skiing Forums) › General Skiing Discussion › How interested are you in the weather?
New Posts  All Forums:Forum Nav:

How interested are you in the weather?

post #1 of 25
Thread Starter 
I'm starting a new business. It's a venture capital backed high tech play that will deliver highly accurate (to within 100 meters) pin-point weather forecasts.

Examples:

o You're driving down the road in a heavy rain you could check this device and it will tell you "it's going to stop raining in 34 minutes."

o It's 7AM and you don't know whether to bring your foul weather gear or not to the slopes. You check our device and it tells you it's going to snow at 9:30.

o You're planning your daughter's birthday party. It's raining. You check our device and you find out that it will stop raining in 92 minutes.

How is this different from current weather forecasts? Accuracy! If we say it'll rain, it will. If we tell you it'll stop raining at 9:34 am, it will. And, it's for YOUR location. We'll be able to pinpoint this to GPS accuracy.

Current weather forecasts are highly generalized. If they say "there's a 50% chance of rain" what does that mean? That there's a 100% chance that it'll rain in 50% of the area; or does it mean that there's a 50% chance that ANY location within the area will get ANY amount or rain.

Our value proposition: Personal weather information for YOUR location... anytime, anyplace.

Our initial implementation will work via cellphone. You'll press *63 and a voice recognition system will take your location to the best you're able (once cellphones have the ability to locate themselves this point will become moot. This'll happen within two years). Of course it'll also work via the web, but the killer app is giving you the info while out skiing or boating.

I have bought the rights to this from MIT who was funded by the FAA to give pinpoint forecasts to airplanes in flights.

BTW, I know EpicSki is non-commercial so I got permission from AC before posting this.

Thanks for helping out.

Bob

[ August 14, 2002, 09:33 AM: Message edited by: WVSkier ]
post #2 of 25
Nice idea. I could see lots of commercial uses for this, but only occasional use by any one individual.

What is the research history behind this product? Can you cite references to the development history in refereed scientific journals?

My initial reaction is concern about the accuracy of the predictions.

For example, do you have credible studies which tell you the probability of error of the method as a function of how far out in time the prediction is needed? What level of accuracy do you think your customers will need & consequently, just how far out in time will you claim your product will be accurate to this level? Will different customers need different levels of accuracy?

How well does it handle distances away from meteorological monitoring stations? Does it factor in microenvironments, local altitude, heat islands, etc?

Related to my scientific concerns about accuracy is the marketing issue of whether people in your potential subscriber base are currently so doubtful about weather prediction accuracy, that you will have a difficult time marketing the product.

Just some thoughts. Good luck with it!

Tom / PM

PS - BTW, my question asking for a literature reference was not just rhetorical, I'm personally curious about this, so if you have any info on the research and research funding history, I'd love to see it.
post #3 of 25
This sounds pretty cool. Would it be possible to get it to also let me know what the potential for rain on the flood plain of a canyon that I am about to hike is?

I do a lot of slot canyon hiking in the summer time, and flash flood danger is a major concern. Not the way I want to die, so if this is a possible solution, I would be all about it!
post #4 of 25
My concern is also with accuracy. Since the weather on the news can be totally inaccurate making even generalized predictions, how are you going to make this both more accurate and more specific?
post #5 of 25
you may find yourself getting a lot of complaints from customers if your predictions on the exact time the weather is going to change are wrong. im all for it if it can be done though.
good luck man

"who maeee"
post #6 of 25
I will express my skepticism here. Unbelievable amounts of computer power, scientific research and resources go into our current inaccurate, area-generalized weather forecasting. I have a hard time believing somebody can muster the resources (both technically and human) to pull this off. If they could the big pay off would be in agriculture where they would be willing to pay huge amounts of money for that accurate of data.
post #7 of 25
So the surprise will be gone? Sounds a wee bit anal. Would it be culled from NWS or NOAA? About as crazy as I get is to check www.crh.noaa.gov/let/marine/ Then to observations and the Harrison-Dever crib(water pumping station). Halve the single inquiry price and you might get some biz, otherwise you would have to hook folks on longer term contracts, because they would probably realize they did not use it enough to warrant the monthly. I guess in the end I like looking at the skies and trying to recognize what might happen. And I suppose that is why we've been the only sailors out on the Chicago shore in a full blown lightening storm.
post #8 of 25
[ August 14, 2002, 11:53 AM: Message edited by: John J ]
post #9 of 25
Thread Starter 
Thank you all for the input.

I can't talk too much about the technology because the company is in "stealth mode" but the algorithms were developed at MIT as a result of FAA grants in the hundreds of millions of dollars.

We've done some preliminary tests and at one hour out we have 99%+ accuracy. That is, if you looked at the "actual" map against the "predicted" map you'd see only a couple of pixels difference from where we said it was going to rain to where it actually rained, or vice versa.

The commercial uses are immense. For example, farmers, truckers, boaters, aviators, etc. But, the crux of my questions here were regarding potential consumer uses. Our thought is that if it became cheap enough that it'd be come a no-brainer.

Regarding value-added stuff like predicting canyon flooding and the like, that is in the plan, but it wouldn't be for few years down the road.

Bob
post #10 of 25
Charging even a small fee for something consumers are used to getting for free is generally not a good idea. Sure, it is more accurate weather report, but it's still just a weather report. Also, be aware that for most folks, this service is useless. You should always bring foul weather gear to the mtns if there is any doubt. Most people plan events much earlier than a few hours. People are going to keep driving to their destinations no matter what the weather (look how many get stuck in flooded intersections). Also, it could be a LONG time before the masses would trust the accuracy enough to actually change their behavior based on your forecast.
The skiing and boating apps are feasible, except cell coverage is VERY spotty, even in metro areas.
Not something that you would want to depend on.
Great commercial apps, if you could get a piece of that.
post #11 of 25
WVSkier, check this site out for some competition.
I used them for last season, and was pleased enough with the results to keep my registration set for both the base and the top of my local hill. Good enough to decide what layers to wear for the day.

http://www.my-cast.com
post #12 of 25
I will add my skeptic's voice to that of Rio. Even ignoring the unliklihood of the kind of accuracy you posit (99% that you can predict when the rain will stop to the minute??), if government funding from the FAA is anything like the government funding I am familiar with from the NIH, it can't be simply used to fund a "stealth" start-up company, and not have the data from the project published for review by the wider scientific community. The figures are also a little suspect- hundreds of millions of grant dollars to MIT? and then the technology is just sold to some investors? This sounds fishy to me.
post #13 of 25
Quote:
Originally posted by WVSkier:
...I can't talk too much about the technology because the company is in "stealth mode" but the algorithms were developed at MIT as a result of FAA grants in the hundreds of millions of dollars.
... We've done some preliminary tests and at one hour out we have 99%+ accuracy...
In preparation for what I am going to say, let me give you a bit of background about myself:

I gave my first scientific talk in 1965 and have been in the R&D business continuously ever since. I have sat on numerous review committees (Navy, other DoD incl. DARPA, DoT, DoE, NSF, FBI, NIH, etc.). Every 2 weeks for 4 years I sat on the patent review board of one of the largest government labs. The proposals I review are commonly for ideas exactly like this - ones with potentially huge commercial benefits.

Now, here's the part you probably won't really like to hear: To be blunt, the two statements of yours that I quoted give me the whammies.

If someone came before me looking for funding and said those things, red flags would be going off right and left in my mind. I have sat in review numerous times of companies claiming to base their product on millions of dollars of government research and are now "in stealth mode". 99 times out of 100 it means that they are spewing smoke, and simply haven't done the basic science, done the hard work, set up the infrastructure, whatever to back up their claims. If federal agencies funded MIT to do the basic work, and then this company licensed the technology from MIT, all of the basic research documentation will be in the public record. This company should be able to point you to it immediately, and you should be able to point us to it immediately.

With respect to your second comment that you have done some preliminary tests showing about 99% accuracy 1 hour in the future, I'm sorry, but I remain very skeptical of the utility of your tests unless the "we" you refer to have specific, Ph.D. level scientific expertise in meteorology and statistics. Were you led through the demos by this company? Did they select the data? Was the data randomized yet representative? How large of a data set? etc. etc.

My seat-of-the-pants estimate is that it would take at least a couple of man-years of effort by Ph.D. level people to validate the basic methodology you appear to be relying on. Are your validation efforts even within an order of magnitude of that?

There is another aspect to these sorts of enterprises, namely, given scientific validity, what is the rate of successful commercialization of a given technology. The government lab that I mentioned above had a patent portfolio in excess of 3500 patents. They were all great ideas. However, of them, only 1% were "productive", ie, actually producing licensing revenues for my lab. This sort of low commercial success rate at high tech enterprises is common.

Don't bet the ranch unless you and numerous advisers with differing fields of expertise (meteorology, statistics, business, telecommunications, etc.) all give you the thumbs up on this technology.

Again, I certainly wish you the best of luck, but again I urge you to be very, very careful in this undertaking.

Tom / PM
post #14 of 25
Thread Starter 
Tom,

It would give me the heebie-jeebies too. But, this is not the forum for us to disclose all this.

We have a lot of what you talk about. We are just not at liberty to post it here. We have a professionally written business plan with detailed analysis of the technology, the methodology, and the markets.

Our company is headed by one of the most pre-eminent meteorologists in the country as Chairman and a very seasoned CEO. My own background is meteorology and wind engineering and high tech since 1975.

Quote:
My seat-of-the-pants estimate is that it would take at least a couple of man-years of effort by Ph.D. level people to validate the basic methodology you appear to be relying on. Are your validation efforts even within an order of magnitude of that?
The answer to that question is "Yes".

This is a VERY serious endeavor with some very highly qualified people. The purpose of this poll was simply to get some feedback and input from a potential market.

But, thank you for the heads up. This kind of input is invaluable.

Bob
post #15 of 25
Permit me to suggest a few of MY thoughts on where to go with this...
Link it to the navigation system in a car to provide current (and forecasted weather) along a route, i.e. it is dry here, but 50 miles down the road I will be travelling it is currently raining, but by the time I get there (based on current speed) it will have stopped, etc. In the same way that traffic management control can advise of alternat routes avoiding current traffic jams, perhaps the system could advise on routes which aren't fog-bound, etc.
Now that would be interesting!

S

P.S. If you do decide to go down this route, half of my share for coming up with the idea goes to me, and half to AC to help run the forum.
post #16 of 25
Thread Starter 
[ August 15, 2002, 05:38 AM: Message edited by: WVSkier ]
post #17 of 25
Thread Starter 
TomK,

We're quite familiar with My-Cast. They are our major competition. The way we'd compete with them is simply through accuracy. They say:
Quote:
My-Cast has created a network of high-resolution weather models, that calculate forecasts eight times a day.
Our forecasts are updates every 10-minutes and to GPS level precision.

Interesting, though, they charge $10/month, much more than anyone in here would pay.

Bob
post #18 of 25
Quote:
Originally posted by WVSkier:
...We have a lot of what you talk about. We are just not at liberty to post it here. We have a professionally written business plan with detailed analysis of the technology, the methodology, and the markets. ... Our company is headed by one of the most pre-eminent meteorologists in the country as Chairman and a very seasoned CEO. My own background is meteorology and wind engineering and high tech since 1975. ... The answer to that question is "Yes". ...
Bob -

!!!WAY TO GO!!! ... ... !!!EXCELLENT!!!

I had really hoped you would be able to reply the way you did. In my experience, I am eternally surprised by the huge number of people who start companies who would not have been able to answer these questions the way you did. Almost all of these companies are headed by smart people, but new to a particular field. Such people are usually completely unaware of the huge amount of work that has already gone on in that field, what resources are available, what validation really means, who they should have on staff, who they should be talking to, etc. It sounds like you are in the opposite situation (but still be careful - grin).

The best of luck,

Tom / PM
post #19 of 25
Personally I am a look at the sky to see if it's raining or not kinda guy but from looking at the poll of the survey it seems many here would subscribe at a certain price so using this micro study it seems it would be worth it to try. When I do dabble in weather forcasts I just want to know if there's a good dump coming and can see that looking at a doppler or the current fronts approaching well enough in adance to throw my powder skiis in the car and not just by what someone tells me is coming. The only down side I see is the years and years of poor accuracy at predicting the weather that many people aren't going to believe you. And most of all never never admit that a government agency was involved with funding, thats a sure way for people to believe it doesn't work. That does bring up the issue of the people (ME AND YOU) paying for the research and developing of this supposed technology so that WV can buy the rights from MIT to sell it to us ... oh well I digress. My Cast seems to be the typical web site weather forcast site very interactive to me that being the selling point. The have a nine hour forcast which so does every other weather forcaster so I guess I am being told that I can now find out that at 3:45 today it going to rain in arvada. At 4:45 it's going to stop or is it going to have moved to brighton ... Colorado is going to be tough lol GOOD LUCK

[ August 15, 2002, 12:26 PM: Message edited by: mercuri22 ]
post #20 of 25
It better be 100% accurate. Lots of farmers would gladly subscribe, for example, to time their hay cutting. But if they get 200 acres of hay down and it rains, they'll sue your butt off.
post #21 of 25
Thread Starter 
Accuracy and precision are the key selling points. No one will ever be 100% accurate.

My-Cast has a nice offerering but they are expensive $8/mo for the PowerApps feature on the Nextel phone and $10/mon for their service.

You are right about the farmers. If they put down a ton of fertilizer and they get a gully washer, they're really hurting.

[edit] Truckers are a good potential market too. Time to put on those chains!

Bob

[ August 19, 2002, 02:59 PM: Message edited by: WVSkier ]
post #22 of 25
Thread Starter 
Just bouncing this to the top, hopefully, to get a few more responses...
post #23 of 25
WVSkier
I think you have market potential in the construction industry. Broad market where timeliness and accuracy are critical.
What does your crystal ball say about snow in the East this year?
Add road departments(snow removal)to your list too.
Bill
post #24 of 25
Okay, bear with me.

The grid resolution you are proposing (100 m) seems a tad unreasonable both in terms of the effort and information required for the forecast modeling. Given the state of the weather observing network (maybe 50 km between stations on average?) the amount of interpolation required for an accurate current condition of a fixed point, the size of a football field, seems unrealistic. Especially if you are then going to tell me the future state of the same point, to the same degree of certainty.

This brings me to your statistics. A 99% accuracy, out to one hour, is not beyond current forecasting capability. It's called "nowcasting". It's just not terribly useful for the majority of consumers of weather information, hence it's not measured. For example, you can predict 7 times out of 10, that the current condition will be very similar to the one that will exist an hour from now. No models, no measurements, just a pair of eyes, ears, and a nose. Add a Doppler radar image, high-res satellite picture, a few isobaric charts, and a pencil, and you can pretty much nail it every time. Out to an hour...

Now show me a trend of 6 to 8 hours and you've got something. Specialized forecasts for events, like the Olympics, are done this way; With their own dedicated team, an onsite set of measurements, and access to all available upstream data. Let's not forget the knowledge and experience required to take into account, local effects (mountains, bodies of water, etc.). I've yet to see a model incorporate geographic anomalies with any degree of certainty of any significant scale.

Are you proposing you would model the entire continent at this 100 metre resolution? Given the "looseness" of the data collection network and the relatively crude scales of available radar and satellite imagery, I doubt it's possible. Pulling and pushing information out of thin air, so to speak [img]smile.gif[/img] .

If you could make it work, then your biggest and best client would be the World Meteorological Organization. Heck you might even get a plaque out of it .

Come to think of it, what accountability and liability are you willing to undertake for "errors"? Given a litigious client base that pays for your service, and the potential outcomes ($$), walk and talk softly my friend...

[ August 20, 2002, 01:07 PM: Message edited by: Warren ]
post #25 of 25
Thread Starter 
Quote:
This brings me to your statistics. A 99% accuracy, out to one hour, is not beyond current forecasting capability. It's called "nowcasting". It's just not terribly useful for the majority of consumers of weather information, hence it's not measured. For example, you can predict 7 times out of 10, that the current condition will be very similar to the one that will exist an hour from now. No models, no measurements, just a pair of eyes, ears, and a nose. Add a Doppler radar image, high-res satellite picture, a few isobaric charts, and a pencil, and you can pretty much nail it every time. Out to an hour...
You are 100% right. It is doable, IF you do it by hand. We intend to do it with computers and radar. Forecasts will be updated at 10 minute intervals during steady (nice) weather and at 5-minute intervals during "weather events". Heck, you wouldn't worry about it if the weather is nice.

We will be doing extensive tests for validation soon.

Bob
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: General Skiing Discussion
EpicSki › The Barking Bear Forums › On the Snow (Skiing Forums) › General Skiing Discussion › How interested are you in the weather?