I was just reading up on numerical weather prediction systems and have a couple of questions. It goes on to say to currently the most powerful super computers can predict with somewhat confidence up too 6 days in advance. Now, the way i think about it is that 15 years ago the most powerful phones could only produce snake 1 and a sms inbox capacity of around 40. Fast forward to the present and look what we can do with phones now! I would simply assume though, different technology, different advancing rates.
So then, hypothetically, in 100 years could weather predictions be made im X amount of weeks/months/years.
Also. Each day that passes is another statistic/equation (?) of weather predictions therefore increasing the unpredictability of nature?
Great question that has a long, complex answer.
To begin with - it depends on what you're trying to forecast. The more granular the information, the more processing power a computer needs to produce the same information.
And it's also about timeliness. Theoretically, a 95% forecast may be able to be generated for scenario X, but it may take two months for the supercomputer to produce. In order to be useful, we need the data to be crunched in a specific time frame (ie 2 hours) and this therefore reduces the level of detail that the forecast can achieve. Obviously as computers become more powerful this ratio will tighten up.
But more importantly - all forecasts require input data to begin with. And until we have a better grid of surface observations (wind, pressure, temp but also wave parameters) to (1) verify the forecast data and (2) provide initialisation into the forecast models, the models will always be at the risk of beginning with incorrect assumptions and errors - which propagate and magnify over time. This is a much bigger issue IMO.
That being said, it depends on what kind of detail you need for your forecast. Specific to surfing, are you interested in (1) broad seasonal trends (ie "will this summer will be bigger/smaller than usual"), or medium range forecasts (ie "we are expecting a 4-6ft south swell next Thursday") or would you prefer some kind of macro-level forecasting (ie "the south swell on Monday will peak around 9:00-9:15am, with an average of four waves per set, decreasing to three waves per set by 1pm and then two waves per set by 4pm, with set wave frequency easing from forty per hour to fifteen per hour during this time).
Obviously the last example of over the top, but it's the kind of high resolution outcome one would expect inline with other numerical forecast developments.
Wont there always be a limit on how far ahead you can forecast and the accuracy as i would have expected if the weather systems do not know what its going to do then how can it be predicted?
Sure there are indications and factors on how things will pan out and you can come up with percentages and likelihood's, but end of the day won't there always be unknowns if you know what i mean.
Of course there will always be limits. I'm just trying to show that there are different kinds forecasts, each of which has a different kind of application.
For example, day to day forecasting is probably less valuable to someone who's heading to Indo for the season and will be able to surf every day. They are more interested in whether the broader season is like to be better or worse than average (and in that statement is a wide range of variables that we can discuss at further length).
But an urban surfer with occasional days off probably doesn't care about the seasonal trend and instead need to know what days are going to have good waves.
Actually the Indo day to day forecast have become more useful in recent years, in the past in remote areas internet access was impossible but now with smart phones and better mobile coverage, i can now access not only swell data but wind data.
Maybe for someone on a charter boat or resort they don't need this, but to me combined with local knowledge its become an amazing tool especially the wind data for areas like Mentawai's and Telos, because i was injured this year and still had a month to kill in this region, i was checking it all the time and comparing it to the conditions and i was surprised at how good the wind predictions have become....
Although dare i say I've got to say Surfline have taken it to the next level with there High resolution wind charts of the Mentawai's and Telos and other areas.
Would love to see the same here :D
Macro level forecasting? Is that possible. Wowzers. I gues it would be handy but also ruin the whole experience of just keeping a eye on the surf to maximise best conditions. Could almost decipher macro level forecast via medium forecasting i.e with the 6hr periodic forecast avaiable here, the peak of the swell event would almost be the most consistent sets too? Im sure its not the simple..
The thing that got me is that i thought that with every day, every hour, every second is another unique part of a ever changing eqaution. I guess the idea is that computers go off similarities of previous weather patterens?
Gahhhhhhh soo much curiosity, soo little brain power..
Yeah all weather model forecasts go off the basic fluid dynamic (Navier-Stokes) equations, with various tweaks and changes made over time to increase the accuracy. As Ben said if we had more weather stations covering the world giving is high resolution data in a timely matter the forecasts would become even more accurate.
The weather model data is then ingested into the wave model, which has it's own set of different equations, and so on.
Ok. Just had a browse of navier stoke eqautions and thats way beyond me. Would it be right to assume that technology or perhaps more available use of current technology will out grow the rate of every day eqautions of nature for longer/precise forecasts.
BTW, this isnt a dig at what present forecasts are capable of, but basically as time goes on will forecasts become more accurate across the board of those different types of forecasts Ben previously spoke of?