Reliance on GRIB Files

[Dockhead]

In another thread (here: http://www.cruisersforum.com/forums/...ml#post2662419)

We were discussing what went wrong when highly experienced circumnavigators, only 150 miles (!) away from finishing a 17 year circumnavigation, got caught out in weather so bad that they were rolled and lost their boat.

One of CF's gurus, ka4wja, wondered whether overreliance on GRIBs might have been a contributing factor:

Originally Posted by ka4wja

. . .
---- Another form of complacency is an unfortunate "reliance" (???) on raw computer model data for weather (this is what GRIB charts are)...
I know in our modern times it's common for laypersons to assume if it comes off a computer, it must be accurate...and sometimes as we see good results in short-term use (like a 2 or 3 day forecast, or along tradewind routes where, except for the ITCZ, weather patterns don't change often), it can become "natural" to assume that what you're using is actually accurate/good...and it might be....but it might not be!

GRIB charts are the raw computer model data, and many sailors even exasperate this issue by using data from only one computer model (such as the GFS)...

If you have no other weather info/forecasts available, then sure GRIBs can be great...but, if you have experienced maritime meteorologists using all their years of knowledge (and input from multiple models, and real-time weather data from ships, bouys, and satellite scans, etc.) to draw weather charts for you, and all done for you 4 times a day, every day, 365 days/year, all for free....and transmitted over the radio, for free, from multiple locations, multiple times per day....(US/NOAA WeFax Charts, transmitted by USCG)...why use GRIBs versus a WeFax chart???

I'm hesitant to bring this up...'cuz it might seem critical...so please know this is not my intention, but rather to bring to light the fact that it is possible that these storm conditions could've been avoided (or at least better prepared for), if WeFax charts had been used instead of a GRIB chart?
Hey, maybe it would've made no difference in this situation, but just thinking it might have!

In my opinion, it seems complacent to ignore what most ocean mariners consider to be the gold standard of offshore weather info/forecasts and instead use raw computer model data...
Not sure if others will agree with me here, just voicing my opinion.. . . .

This is a question which has always bothered me and I thought it would be interesting to discuss.

It's really easy to plan passages by using GRIB wind forecasts, and in many cases, the GFS model or whatever model you're using, presented in GRIB, is the best way to incorporate weather assumptions into your planning. You can even put them into a routing program and get a ready-made weather route!

But it always bothered me that we are in some cases maybe assuming too much. I always look at Wefax charts and I ALWAYS read the official maritime forecasts (and recently I have had a big boost with this by figuring out how to get those forecasts easily by Iridium Go! and SailDocs), but I don't think I understand weather enough yet to really do this right. But looking at GRIBs in the context of what those other sources tell you at least gives you some idea of the bigger things going on, which in turn will give you some idea of what might turn out differently, the kind of knowledge which might have made a huge difference in the outcome of the Kelaerin case.

Just to name a banal and extreme example to illustrate the point --

If you are in the path of a developing hurricane, A GRIB wind forecast might show calm weather where you are, on the basis of the fact that the model predicts a certain path for the storm. But we all know that models can't yet predict the paths of hurricanes with great accuracy. So Southerly F3 might be NE'erly F12 just because the storm's path varies by 100 miles by the time it gets to you. The GRIBs don't tell this story, but a Wefax chart does tell this story -- you see the bigger picture and with some knowledge you can have some idea of what might be different.

So in the Kelaerin case, maybe the victims could have seen the gales developing and figured out that there was as risk of severe conditions not shown in the GRIB wind forecasts.

I had this discussion a couple of years ago, and someone I greatly respect, Evans Starzinger, wrote that the models are so good now that you really don't need anything else. But I continue to be skeptical about that, and continue trying to learn more and more.

Of particular relevance to me because I have a difficult passage ahead of me in August -- Greenland to Ireland in the middle of the month, at the ragged edge of the good season to be doing this.

So what do you guys think?

[Snore]

I believe, like navigation, one should not rely on only one source of data.

We have all been caught in weather that was not predicted- even by pros. But if you are using 2 sourc s of info, such as GRIBS and inReach weather, or a weather routing service and inReach— the odds of an “oh-my” moment should be lower. There is just too much at stake to trust in the infallibility of the weatherman

[conachair]

Originally Posted by Dockhead

So what do you guys think?

Gribs tend to mirror closely Wfax I've found, 0 hours/surface analysis should be near identical, same data. ... BUT, you really need to have a display which shows gusts, rain and cape easily to get a better handle on what might happen. Cape is excellent to get some fear that a thunderstorm might get you. Opencpn has to be the best around for display, getting gribs over Wfax is gold dust IMHO , to see if the models have been tweaked by a forecaster.
A quick flick through the other thread didn't come up with any grib/wfax images so maybe they were very similar and the weather just changed its mind.

Overall don't *rely* on any source IMHO, they all should be treated with a fair degree of caution, especially a few days ahead.

[Sedridor]

Dockhead,

IMHO you are right that this is a question worth exploring. All computational models, WRF, GFS, NAM, whatever, suffer from a certain degree of error. This error can come from the data used to initialize the model, as well as approximations used by the model to achieve practical computational runtimes. The point is, every model prediction, from windspeed to temperature, has error bars.

It's crucial to note that these errors compound through time as the model is run forward. What this means is that a model can be relatively accurate at initialization, but as it is run forward, these errors compound and widen the uncertainty around the predictions. It's not to say that a single model can't be right really far out - modern models are quite good, and have been demonstrated to show skill (i.e. demonstrably better than farmer's almanac) over a week out in some cases.

But a grib file doesn't do a good job of depicting this. A grib file just gives you a wind speed, a direction, a temperature, etc for each grid cell and gives no indication of error bars. It is a cozy, albeit misleading overrepresentation of certainty. To combat this problem, meteorologists look at a variety of models, correct model outputs on a statistical basis, compare to satellite photos, and examine how the model predictions are changing through time. Among other things. None of which is particularly practical for cruising, given the amount of data, time, and specialized knowledge required.

I think your recommendation of leaving the analysis to the pros is a wise one. Using a curated product from NOAA or a similar organization will generally yield a higher level of forecast skill than a GRIB file can, because that GRIB file was just a small piece of their analysis.

[Paul L]

For short range forecasts the GRIBs and the WeFax tend to be very close to each other. For longer term forecasts the WeFax aren't available.
Forecasting storms like a named hurricane might be different.

[Kenomac]

There’s also another angle to explore which will contribute to the same outcome on other boats when the owner relies too heavily on weather reporting. I don’t think this was the case for the lost boat, as the owners seemed quite experienced, but two years ago, we did set out to beat a storm while heading south along Corsica which eventually lead to us making our youtube storm video in 40-50 knot wind.

One also needs a boat which can handle more extreme weather conditions, and as much as many on this forum will argue against, the fact is.... not all boats are created equal. The conditions described by the couple who lost their boat 30knot winds producing 20-30ft seas, would not be an issue of concern on many other bluewater type boats.

What I’m trying to say, is you can’t rely 100% on weather forecasts and need to be prepared and have a boat that’s well-suited for the unexpected.

[Paul L]

Originally Posted by Kenomac

There’s also another angle to explore which will contribute to the same outcome on other boats when the owner relies too heavily on weather reporting. I don’t think this was the case for the lost boat, as the owners seemed quite experienced, but two years ago, we did set out to beat a storm while heading south along Corsica which eventually lead to us making our youtube storm video in 40-50 knot wind.

One also needs a boat which can handle more extreme weather conditions, and as much as many on this forum will argue against, the fact is.... not all boats are created equal. The conditions described by the couple who lost their boat 30knot winds producing 20-30ft seas, would not be an issue of concern on many other bluewater type boats.

What I’m trying to say, is you can’t rely 100% on weather forecasts and need to be prepared and have a boat that’s well-suited for the unexpected.

Since seas are far more damaging than wind to an offshore boat, what size breaking seas do you consider dangerous? 20 to 30 foot breaking seas are life threatening to any cruising boat. How they are produced is far less important. If the seas are larger than you'd expect for the existing wind history, then current is probably coming into play making the seas even more dangerous.

[ka4wja]

Dockhead,
Cool idea to discuss this...and since my words were the impetus for this discussion, please allow me to preface this with a clarification or three...
And, not sure I'm a "guru"...but no worries, I'll try to act accordingly...


And, a note to everyone....
If you wish to discuss the specific incident....that's fine...but that is not what I was discussing, and I don't think that was what Dockhead was intending here...
But rather, how much is everyone willing risk on a raw computer model?? And, have we all made everyone aware of these facts and deficiencies??





1- Everyone please know I mean no criticism to anyone, especially those who have lost their boat/possessions (or even worse some who have lost the lives of their family or friends, as I referenced some other recent rescues, too).
My comments are made to help both everyone here, and myself too, to better understand how things can go wrong, and go wrong fairly quickly offshore, in bad weather...and to see if there is anything that can be done in the future to head-off some of these possible issues. (big seas and lots of wind can toss even a well-found sailboat around like it was in a washing machine...and that fact is so often lost on novice offshore sailors....and sometimes forgotten by even those experienced)
2- This is not my usual praise of "HF comms", as you can access both WeFax charts and GRIB's via HF radio, sat comm, wi-fi/cellular (when close to shore), etc....so "how" you get weather isn't part of what I was commenting on, but rather "what" weather info / forecasts some might be relying on.
3- FYI, there is still no real "need" for any real live weather forecasts while en-route...many have sailed across oceans without it, and simply "read the sky", etc...so, nobody should assume that I'm saying that "you must have this, or that", or that "you must do this, or that"...'cuz you don't "have to" do or have anything....
This is just about how some might be desiring of some outside weather info/forecasts, and either not be aware of the differences, or have lulled themselves into a sense of complacency over the years...

~~~~~~~

Now that I got that preface out of the way, allow me to surprise some here...
I agree with Ken!!

Originally Posted by Kenomac

What I’m trying to say, is you can’t rely 100% on weather forecasts and need to be prepared and have a boat that’s well-suited for the unexpected.

This is what I've been saying for years and years (usually in the context of fallacy of "weather windows" for ocean passages)...
I've always said that you should leave on a positive weather forecast, but understand that "you will sail with the weather that you have out there....if you have good weather info / forecasts, you may be able to adjust course along your route to avoid unfavorable weather (or reduce its impact) and/or adjust course to find more favorable weather...but, be prepared for the worst..."


Back on-point of this discussion...
My question / comment was regarding the complacency of experienced offshore sailors (and the ignorance of novices), in regards to weather forecast differences.

I know that I have personally been met with many astonished faces, and even more "unbelieving faces", when informing many of my fellow sailors that GRIB's are the raw computer model data, and are not human forecasts...
(to be clear, many just don't want to believe it...as some have already spent 1000's of $$$$ on PACTOR modems or sat comm gear, just to get GRIB's...and they don't feel comfortable thinking about what weather data they're getting with all that expensive hardware...it's a shame, 'cuz they can still use all of it to get WeFax charts, whether via a WeFax transmission or via saildocs, from PACTOR or sat comm..)

Now, I know the argument:
Many say/write that "GRIB's are just as accurate as WeFax"..
And, while I disagree (my experience is based on looking at computer models on-line versus human-drawn WeFax charts, for the N. Atl, Caribbean, etc.), my comments are not trying to argue this issue at all...

But rather, have some experienced offshore sailors, simply allowed themselves to become complacent and use what is "easiest" for them, and/or what they are most used to / familiar with, rather than what might provide the best forecast, as they haven't seen much differences in short-term forecasts???
(And, on a parallel discussion....have we failed to adequately show the novice offshore sailors that there is a difference between a WeFax chart and a GRIB chart?? Not discussing the accuracy, but the fact that one is drawn by an experienced maritime meteorologists, and one is the raw computer model data...)

Again, although I haven't seen 3day to 5day forecasts to be "close enough", I'm not arguing the "accuracy" of GRIB's...just wondering if this over-promotion of them has lead to complacency?? And, have some forgotten that they're the raw data, without human interaction??

And, yes...I do understand that if there is a difference, without a distinction, some will argue what is the fuss about??
Although I don't see it as a "fuss"...in my opinion, this is about:
a) relying on raw computer model data (rather than a human) for a forecast...
b) having a false sense of security due to collective assumptions that if a computer did it, it must be accurate...

Originally Posted by Dockhead

In another thread (here: http://www.cruisersforum.com/forums/...ml#post2662419)

We were discussing what went wrong when highly experienced circumnavigators, only 150 miles (!) away from finishing a 17 year circumnavigation, got caught out in weather so bad that they were rolled and lost their boat.

One of CF's gurus, ka4wja, wondered whether overreliance on GRIBs might have been a contributing factor:

This is a question which has always bothered me and I thought it would be interesting to discuss.

It's really easy to plan passages by using GRIB wind forecasts, and in many cases, the GFS model or whatever model you're using, presented in GRIB, is the best way to incorporate weather assumptions into your planning. You can even put them into a routing program and get a ready-made weather route!

But it always bothered me that we are in some cases maybe assuming too much. I always look at Wefax charts and I ALWAYS read the official maritime forecasts (and recently I have had a big boost with this by figuring out how to get those forecasts easily by Iridium Go! and SailDocs), but I don't think I understand weather enough yet to really do this right. But looking at GRIBs in the context of what those other sources tell you at least gives you some idea of the bigger things going on, which in turn will give you some idea of what might turn out differently, the kind of knowledge which might have made a huge difference in the outcome of the Kelaerin case.

Just to name a banal and extreme example to illustrate the point --

If you are in the path of a developing hurricane, A GRIB wind forecast might show calm weather where you are, on the basis of the fact that the model predicts a certain path for the storm. But we all know that models can't yet predict the paths of hurricanes with great accuracy. So Southerly F3 might be NE'erly F12 just because the storm's path varies by 100 miles by the time it gets to you. The GRIBs don't tell this story, but a Wefax chart does tell this story -- you see the bigger picture and with some knowledge you can have some idea of what might be different.

So in the Kelaerin case, maybe the victims could have seen the gales developing and figured out that there was as risk of severe conditions not shown in the GRIB wind forecasts.

I had this discussion a couple of years ago, and someone I greatly respect, Evans Starzinger, wrote that the models are so good now that you really don't need anything else. But I continue to be skeptical about that, and continue trying to learn more and more.

Of particular relevance to me because I have a difficult passage ahead of me in August -- Greenland to Ireland in the middle of the month, at the ragged edge of the good season to be doing this.

So what do you guys think?

Well, that's what I think....



And, as for multiple sources of weather info / forecasts....
That's a great idea...but...

Originally Posted by Snore

I believe, like navigation, one should not rely on only one source of data.

We have all been caught in weather that was not predicted- even by pros. But if you are using 2 sourc s of info, such as GRIBS and inReach weather, or a weather routing service and inReach— the odds of an “oh-my” moment should be lower. There is just too much at stake to trust in the infallibility of the weatherman

This is a good idea, and is possible....but not that easy / quick...
A few facts get in the way, a little bit, here..
Sorry, Snore...
Fact #1: The "official" Hi-Seas Forecast (aka "SafetyNet" forecasts) are done by the weather service / met offices of those countries assigned to do this for their METAREAS (under the IMO / GMDSS)...
So, whether you get it via text or image....from HF broadcast or download, or from a sat comm connection....it is the same forecast, prepared by the same guy...
Fact #2: The raw computer model data (from various models??) are available for you to compare...whether downloaded via PACTOR or sat comm...but, these are the models that the above human forecasters use (along with other data) to make their official forecasts...
And, I'm not a good enough meteorologist to second guess the guys who do this for a living...so, while the raw model data is there for you, not sure it will provide much, if you already have the official forecast (whether a WeFax chart of a text forecast)

???Fact #3:???
(not exactly sure here...perhaps someone who is using InReach Weather will chime in here???
But, I'm assuming that InReach uses the computer model data (just as the GRIB's and places like Passage Weather, etc., does)....
So, this doesn't provide you with much of a contrast either...
I could be wrong in this assumption, but in any case, I'm not putting too much emphasis on InReach Weather versus the US NWS/NOAA or the UK/Aus/NZ/SA met offices...




Well, that's what I got...
And, again, I'm not arguing / comparing their relative accuracies....just thinking that some are a bit complacent in their use of the raw model data...

Maybe others will differ....and that's fine with me...(maybe I will even learn something new!)

Fair winds.

John

[Steadman Uhlich]

Originally Posted by Paul L

Since seas are far more damaging than wind to an offshore boat, what size breaking seas do you consider dangerous? 20 to 30 foot breaking seas are life threatening to any cruising boat. How they are produced is far less important. If the seas are larger than you'd expect for the existing wind history, then current is probably coming into play making the seas even more dangerous.

Very good points!

I have seen huge seas coming from distant Pacific storms, at the same time the wind locally (where I was) was not extreme. And when those seas come to a shallower area or against current, they can rise up, breaking. The fetch in the North Pacific can lead to some big ones.

[Dsanduril]

One thing I frequently see with GRIBs is a too-narrow focus. You're going 400 miles and you grab a GRIB to cover that area. When just outside, in the shadows lurks...

If on any length of a passage you really need an ocean-scale picture so you can see what might be gaining on you. You can get that with GRIBs (we tend to use a 1 degree GRIB for local conditions and a more spaced GRIB for ocean level). With WeFAX you're essentially forced to look at ocean level. That's the only size they transmit. You can get similar information from a GRIB, but you have to actually set it up that way and many people don't.

[DumnMad]

Originally Posted by Steadman Uhlich

Very good points!

I have seen huge seas coming from distant Pacific storms, at the same time the wind locally (where I was) was not extreme. And when those seas come to a shallower area or against current, they can rise up, breaking. The fetch in the North Pacific can lead to some big ones.

Huge seas from distant storms are well rounded and not a big deal for small boats. They are also well forcast.
Getting near shallow water while big rollers are around is foolish.
TransTasman sailors are usually conscious of the sea mounts and other shallower areas to avoid when storm waves are around.

[Snore]

@Ka4wja

First thank you for the clear explanation. What I was looking at for longer offshore trips was QTVLM and a Weather router via inReach.

But your point is well made- if both services use the same model they will always agree. Therefore, no safety check. There is also a basic open ocean forecast you can get by querying (and paying) through the inReach.

Clearly, worth some effort to better understand what model each service uses.

THANK YOU!!

[Dockhead]

Originally Posted by Kenomac

There’s also another angle to explore which will contribute to the same outcome on other boats when the owner relies too heavily on weather reporting. I don’t think this was the case for the lost boat, as the owners seemed quite experienced, but two years ago, we did set out to beat a storm while heading south along Corsica which eventually lead to us making our youtube storm video in 40-50 knot wind.

One also needs a boat which can handle more extreme weather conditions, and as much as many on this forum will argue against, the fact is.... not all boats are created equal. The conditions described by the couple who lost their boat 30knot winds producing 20-30ft seas, would not be an issue of concern on many other bluewater type boats.

What I’m trying to say, is you can’t rely 100% on weather forecasts and need to be prepared and have a boat that’s well-suited for the unexpected.

The choice of this or that boat won't always make that much difference. As Paul L said, a 30 foot breaking wave will roll and devastate any cruising boat, and the wind speed is never the issue -- it's the sea state.

Your much-posted video did not show any kind of extreme conditions -- it's always hard to judge wave height from a 2D video, but the horizon is visible in that video, so the seas were not higher than your eye height, and they were not breaking. Those conditions are not threatening even to day sailers. The wind speed is irrelevant.

You are certainly right that you can't be sure about weather forecasts, certainly not long term ones -- something we should all keep in mind -- but the choice of boat is not a panacea. You need tactics, you need proper gear (Jordan Series drogue for me), you need skill, you need stamina, and after all, you actually need some luck. Nothing wrong with a good, stout, seaworthy boat (especially a LARGE seaworthy boat -- size counts a lot), but that's just one part of the equation, which by itself will not get you out of every situation.

My take on the Kelaerin disaster is that the boat had nothing to do with it -- they made a few mistakes which any of us could have made, including possibly using inadequate weather information -- but mostly they were just really unlucky. The boat held up fine -- was not holed, not dismasted, just took a bunch of water down the inadequately secured companionway, and then their pumps clogged, like ALL bilge pumps will do in that situation. Do you have a proper dewatering pump on board? Do you always secure the companionway? Is your washboard latch strong enough to withstanding a direct hit by 5 tonnes of water? We all make mistakes on this level, every day, at least -- I certainly do. There but for the grace of God . . .

[Paul L]

Originally Posted by Snore

@Ka4wja

First thank you for the clear explanation. What I was looking at for longer offshore trips was QTVLM and a Weather router via inReach.

But your point is well made- if both services use the same model they will always agree. Therefore, no safety check. There is also a basic open ocean forecast you can get by querying (and paying) through the inReach.

Clearly, worth some effort to better understand what model each service uses.

THANK YOU!!

Too much info is not helpful. When we are waiting to decide when.to leave we download multiple GRIB models and forecasts. They invariably are close for the first few days. If they diverge significantly in the later days then I have little trust in either forecast.
Offshore I get only one model.downloaded and make my decisions based on it and what I can.see out the port. Having conflicting info while trying to make decisions in trying offshore conditions is not helpful. Take the base info and, as captain, make your decision and move on. No matter what you see forecast you will still deal with whatever weather you get. The only weather conditions you absolutely get to pick is the conditions when you start the passage.

I was on a passage once with a confusing weather scenario after the first 3 days. One of the boats on the same passage had info from the 3 different shoreside weather routers. Each had a different waypoint for where they should head to minimize issues with an.oncoming weather system. Two of the waypoints meant adding many miles and hours of motoring away from the rhumbline. Rather than getting comfort from the additional expert info from shoreside routers, the skipper ended up with dealing with a lot of unneeded stress, lost sleep and running low on fuel.

[DumnMad]

Originally Posted by Dockhead

The choice of this or that boat won't always make that much difference. As Paul L said, a 30 foot breaking wave will roll and devastate any cruising boat, and the wind speed is never the issue -- it's the sea state.

What do you mean here by 30ft breaking ??
Kelaerin appeared to be in deep water.
In 60kts and open ocean I've seen 30ft waves but only the top 10ft or so appeared to be breaking.

Compare this with a shoreline where you see quite small ocean waves rise up to a breaking height of over 10ft. e.g. Google Greymouth bar.