Magicseaweed Logo
Profile Page  ADD SPOT   Profile Page  PROFILE   Log In  LOG IN

Ten Day Forecasts and Forecast Probability

By Ben Freeston - Published on 5th July 2010 | Viewed 212886 times.

Related Content
News Image
You're showing 6.1 feet at 15 seconds as the dominant swell. But when I look at the components of that swell this doesn't make sense.
News Image
With our latest update your forecast has changed. Read here to find out how.
News Image
Explaining how our 'star ratings' work and what you can, and can't expect from them.
News Image
We've added a new feature so you can now upload almost any kind of news article to Magicseaweed.com and reach out directly to over 500,000 surfers worldwide. Whether it's ...

Probabilities




Every surfer who's used any computer generated surf forecast regularly in the past ten years or so has learnt one fairly obvious fact: forecasts can change and the further you look into the future the more likely this is to happen.

You know the story, you spot a new swell right at the end of the seven day forecast and cross your fingers, as it draws closer the forecast updates and it either sticks around, changes, or sometimes even disappears completely. You know this and we know this, but magicseaweed is the first and only free surf forecast service to now try to put some science to this phenomena.

Against almost every three hourly forecast you'll now see a new number, the 'probability'. This gives you a great idea of how likely the forecast is to change. You'll notice that for almost all locations this is as high as 100% for up to three or four days from now. This shouldn't come as too much of a surprise, if you compare your MSW forecast to a local wavebuoy you'll find that it's really very very accurate over this time range.

What you'll also notice is that towards the end of the forecast (particularly now we've extended the range to a full ten days) is this becomes generally less certain. But remember not all swells are created equal, that local wind swell might be showing only a probability of 60-70% for the weekend, but a nice long range groundswell has probably already been created by a storm somewhere, so you'll find this sort of swell can be up to 100% certain seven or more days from now - this is information that simply wasn't available to the surfer before.

Another phenomena you'll spot is a swell showing at 90-100%, but the first day of the new swell looking a lot less certain. This means the forecast is absolutely confident the swell will arrive, it's just a bit less sure as to exactly when, these little issues will iron themselves out once the swell gets to within three or four days from now, but if you're making plans make it it's really helpful to know when things are most likely to change.

Bear in mind that this is still a forecast, it's a complex series of mathematical process designed to predict the future and that's a hard thing to do. It's worth repeating that - we DON'T have a crystal ball, we can't actually compare a future forecast with the future, a forecast of 100% certainty simply means that the computer is so confident of the situation that no amount of pushing will get it to forsee a different future - it DOESN'T and CAN'T mean that there's an actual 100% certainty of the weather responding exactly as predicted. So even a 100% certainty doesn't mean that tomorrow's forecast, with new information, can't change. As with any forecast this data is a guide to help you plan a session, not a cast iron guarantee.

Nonetheless this new tool should really change the way you're able to make plans around our data.

This probability only applies to incoming swell, we'll add probabilities for wind and weather shortly.

Ten Day Forecasts




We've extended most of our detailed long range forecasts to a full ten days. The first seven days are exactly the same data you've been used to using. The extra three days come from a slightly different forecast method, and to run a forecast this far out we've compromised with the level of coverage - you'll see that beaches in the same area share the same data and that the forecast is run on this regional level. The size predicted may vary slightly by comparisom to the data for your local spot (Although ocean swell isn't produced locally and regions are selected so that all spots will be affected by similar swells).

Nonetheless our testing has shown this forecast a reliable and accurate indicator of the arrival of both easier to predict long range groundswells and even spotting short period wind swell up to 240 hours in the future. Coupled with the new 'probability' rating you're better placed than ever to start making plans to travel up to ten full days in advance. Magicseaweed is proud to be the only free surf forecast service offering this level of detail and length of forecast and with the new probabilities showing occasions where the forecast is hard to rely on we hope that we're giving you this information in the most honest way possible...

Multiple Swell Warning




We've come a long way in the last decade understanding swell and period and getting used to accurate long range forecasting. But our ongoing work here at MSW it to figure out how to best present complex ocean conditions as simply as possible and sometimes that simplicity can show less than the whole truth.

In many locations it's very ususal to have more than one definded swell running at the same time, your MSW forecast shows you EITHER the most powerful swell likely to make waves at the beach OR a predicted combined conditions if there is more than one swell that looks like it'll make ridable waves.

However there are issues. We've noticed this season in Indonesia that there have been regular south swells running in conjunction with the usual south-westerly swell. We show the combined effect of this incoming swell, but on the beach in Bali, for example, this more southerly swell hasn't made it to the reef and the forecast looks like it's overcalling, a quick check of the full breakdown shows the real situation.

In other locations this measurement of the effect of combo swells is much more useful - so to give you a warning when this is the case we've added this symbol everytime a combination if swells is running. Click the 'Full Swell Breakdown' tab at the top of your forecast and you can instantly see the exact mix of swell likely to make surf.

N.B. Our star ratings DO take this into account, giving you a rating for the best looking individual swell NOT the combination.
Rate this article: 1 2 3 4 5
This article has been given an average rating of 4.3 from 324 votes.
Comments
 
Archived Comments
atalaiaboy532 days ago -1 Rating Rate Positive Rate Negative
I still do not understand how the arrows indicating the direction of the wind?
soutie529 days ago +7 Rating Rate Positive Rate Negative
well its not too complicated. the arrows indicate the wind direction.
soutie529 days ago +8 Rating Rate Positive Rate Negative
what you have to do is look outside to see which way the wind is blowing, and then calibrate you screen by turning it so the arrow for the forecast for "now" is pointing in the same direction as the wind is blowing to.
rabatter568 days ago -23 Rating Rate Positive Rate Negative
Comment Voted Below Threshold
Bear Kookilicious570 days ago -5 Rating Rate Positive Rate Negative
I also like the way forecasts made up of multiple swells are being flagged in the summary forecast. This is definitely a useful improvement.
rabatter562 days ago -20 Rating Rate Positive Rate Negative
Comment Voted Below Threshold
Ben Freeston474 days ago +7 Rating Rate Positive Rate Negative
We're clearly not advertising the holiday park on that page or anywhere else. Who'd pay for an advert that you can't see (secret adverts for 'secret spots'???) and who'd argue it was there when it clearly wasn't. The fact that there's an enormous holiday park on the beach, with it's own RNLI lifeguard station, it's own surf school and it's own surf hire business does give you a pretty good insight into how we decided it probably wasn't a secret...
Bear Kookilicious571 days ago +6 Rating Rate Positive Rate Negative
I think this measure of confidence in a forecast is a great idea and very helpful.
rabatter574 days ago -22 Rating Rate Positive Rate Negative
Comment Voted Below Threshold
plappers575 days ago -3 Rating Rate Positive Rate Negative
Thinking about this more - and assuming you are using the ensemble to calculate this - would you not be better off creating a a percentile (lets say the most likely 20-30 percent) of the ensemble members and then showing what that band is showing - that way you will miss out the massive jumps from one timestep to the next.
plappers575 days ago -2 Rating Rate Positive Rate Negative
Hi, nice to see the latest developemts here, Interesting to say the least - I presume you are using the ensemble to make your prediction in the longer range and thats fine but i have noticed that the per centages can vary greatly from one 6 hour window to the next - this doesn't make meteorological sense in my opinion - as the confidence between different time steps should not change a great deal from 1 forecast timestep to the next. Yes it can and will change from one forecast timeissue to the next but to see a 20-35% confidence differential from one 6 hour time stamp to the next seems like you have missed something out here. I stress i'm not dissing the forecasting here but jsut offering constructive commentary.
Ben Freeston575 days ago Unrated Rate Positive Rate Negative
Looking at the data it can make sense, the example I gave above I've seen several times during validation, a similar swell showing in all members but the arrival time varying enough to move it into different timesteps. In this scenario it does seem valid to alert the surfer that there's uncertainty at the forecast time of arrival even if the overall picture is one of high confidence of the of the swell hitting the coast. Obviously the balance we need to find with a new set of data is to create information that's easily accessible to every current user, in time we then create product that allows anyone with greater interest to drill down and get a clearer picture of exactly what's going on.
Kieran100575 days ago -2 Rating Rate Positive Rate Negative
Perhaps on the main forecast page you could assign the probablities for each time step on a scale of 1-5 like the star system making it simpler for the casual user. Giving precise probabilites in terms of percentage is misleading as the current ensemble methods used in WW3 are relatively crude. A 5 star likelihood would mean that the swell is most likely to occur, whereas a 1 star would obviously mean that it is highly unlikely. To account for the simplification, a more detailed page could be introduced with a graph showing the 19 different ensemble members over the 10 day period with the individual swells plotted. This would make it easier to see the divergence in the ensemble over time. The likelihood of individual swells being forecast rather than the SWH is going to be most useful. So people understand what they are looking at, a bit about the use of the differnt wind fields in the WW3 ensemble would also help. The current forcast is great so keep up the good work.
Ben Freeston575 days ago +1 Rating Rate Positive Rate Negative
Cheers Kieran, I think people have an intuitive grasp of probabilities to some extent - to add another index becomes very complex from a layout perspective. I think this makes some sense, particularly as weather forecasters generally start to use similar language, or if it doesn't it at least leaves the rest of the information relatively uncluttered. In our testing the product has been extremely useful, but certainly it's just a taste of what can be done with the data...
Ben Freeston575 days ago Unrated Rate Positive Rate Negative
Regarding accuracy I think precision, particularly long range, implies accuracy anyway (7 days out swell heights to .1 of a foot etc.) and is an ongoing flaw in everyones presentation of model derived spot data - the fact that there's an explicit percentage, often low towards the end of the period is a fairly strong step towards addressing this.
 
 About MSW  Our Shop Friends  Legal
Home
Site Map
Advertise
Support
Surf Shop
Surfboards
Xcel Wetsuits
Leashes
Tailpads
Go Pro Camera HD
Surf DVDs
Boardbags
Roofracks
Surf Books
Surfwax
Ding Repair
FCS Fins
Longboard Fins
Posters and Prints
Balance Boards
Mambo Clothing
Learn to Surf
All content remains copyright of Metcentral Ltd unless stated otherwise, we'd kindly ask that you don't reproduce it in any form without our permission.

Total Time 0.1184