Predictwind routing

[lustyd]

Just planning a cross channel trip and using Predictwind and realised it doesn't use tidal data at all in the planning algorithm. It seems 100% based on true predicted wind. I thought this weird so did a bit of playing around with a Solent based example where more tide is present and I can more easily see what's going on. A trip from Hurst to just past Bramble bank, straight line with departure planning set for a westerly flow and easterly flow gives the exact same results. I would expect the tide to make an enormous difference to this trip, having watched many boats make zero progress over the years when fighting tide even with good wind.

For a channel crossing, which I realise is still a bit of a small trip for the tool, this means they are getting the apparent wind wrong for most of the journey, and therefore boat speed is wrong for most of the journey, and therefore the planning is completely useless aside from the nice animation of a boat and weather over time through the trip.

I'm not expecting the tool to be perfect, but this seems like an important thing for them not to take into account!

 

[lustyd]

Never mind, a bit more playing and it turns out this is a pro level feature but is available

 

[Martin_J]

It doesn't answer your question directly but have you tried the Sailgrib WR app?

Sailgrib WR does include tidal streams, has a good number of polars (if you don't have your own) and has a number of different grib download possibilities.

It's not easy to find the limitations for the free version but I think it's just that grib downloads are limited to two days ahead in the free version.

 

[Martin_J]

At least Predictwind do make it clear which features each version includes...

Pricing - PredictWind

 

[franksingleton]

I saw the track of a PW routing some while ago from the Solent down to the south, Canaries or Madeira, I forget which. They went through the Alderney Race at night, south of Guernsey, tacked along the Brittany coast (it was a catamaran), through the Chenal du Four at night against the tide. Had they waited a day, they might well have laid the Chenal du Four. Whether they could or not, it did seem a very odd route at the time.

 

[Sandy]

I'm not expecting the tool to be perfect, but this seems like an important thing for them not to take into account!

They are based in and based in either Australia or New Zealand. It's a weather tool not a tide tool. We are blessed with some great tides off this small island the rest of the world don't really enjoy the thrill of getting tidal gates wrong or even SOG of 1.5 knots on the 0300 to 0600 watch as you punch a foul tide in La Manche.

 

[lustyd]

It’s a routing and planning tool which uses boat speed to predict where you’ll be and when based on the wind. Given that the wind could vary by 12 knots due to tide that’s a rather important stat.
As said above though, they do do tides after all, it’s just very expensive

 

[franksingleton]

Do they take TSS rules into account and ship dodging generally?

 

[Martin_J]

Do they take TSS rules into account and ship dodging generally?

Tides and TSS mid channel as well as the Isles of Scilly TSS zones fully taken into account in Sailgrib WR.

 

[Martin_J]

It’s a routing and planning tool which uses boat speed to predict where you’ll be and when based on the wind. Given that the wind could vary by 12 knots due to tide that’s a rather important stat.
As said above though, they do do tides after all, it’s just very expensive

As mentioned earlier.. try Sailgrib WR Free version..

I think even with the restricted number of days in the free version grib downland, you should get an idea of what is like for a cross channel routing calculation.



Although with the paid for version, you'll be able to work out crossing times for the week ahead as well...

 

[Bristolfashion]

It's a hellishly difficult thing to automate - a huge range of variables, all of which can change. I sailed Pierowall, Orkney to S.W. Shetland yesterday - crazy currents, wind all over the place. I had been tempted to give up trying a CTS, but, in the end, was less than 3nm on the safe side of my destination waypoint, which I called a win. Almost nothing was as predicted, the slight discrepancy from plan was not getting pushed E. as far as expected on the final approach due to stronger winds & weaker currents.

Still, arriving bang in the middle of the night with adequate light to enter & anchor was quite something - and, at 2am when we had our arrival cuppa, it was distinctly dawn!

I have tried the Savvy Navvy version, but didn't go beyond the trial period.

 

[lustyd]

try Sailgrib WR Free version

Unfortunately I don't have any Android devices (and won't be getting any!) so can't even try that one.

 

[franksingleton]

It's a hellishly difficult thing to automate - a huge range of variables, all of which can change. I sailed Pierowall, Orkney to S.W. Shetland yesterday - crazy currents, wind all over the place. I had been tempted to give up trying a CTS, but, in the end, was less than 3nm on the safe side of my destination waypoint, which I called a win. Almost nothing was as predicted, the slight discrepancy from plan was not getting pushed E. as far as expected on the final approach due to stronger winds & weaker currents.

Still, arriving bang in the middle of the night with adequate light to enter & anchor was quite something - and, at 2am when we had our arrival cuppa, it was distinctly dawn!

I have tried the Savvy Navvy version, but didn't go beyond the trial period.

The first sentence is why I am sceptical about weather routing, particularly over long passages.

 

[TLouth7]

It's a hellishly difficult thing to automate - a huge range of variables, all of which can change.

Clearly it is well within the capabilities of modern software (I am avoiding calling it trivial) to find the optimum route given a single forecast, a decent tidal atlas and a set of polars for a given vessel.

But do any of the current routing software offerings provide sensitivity analysis, or more helpfully suggest routes that might be less than optimal given information at the time, but which are less sensitive to potential future changes in the forecast?

 

[DFL1010]

The first sentence is why I am sceptical about weather routing, particularly over long passages.

I'm not sure I'd be quite so dismissive. Over the decades the tech's been in use, it's shown itself more useful than the alternative, in the right hands.

 

[DFL1010]

Clearly it is well within the capabilities of modern software (I am avoiding calling it trivial) to find the optimum route given a single forecast, a decent tidal atlas and a set of polars for a given vessel.

But do any of the current routing software offerings provide sensitivity analysis, or more helpfully suggest routes that might be less than optimal given information at the time, but which are less sensitive to potential future changes in the forecast?

In short, yes. A common beginner's mistake is to take the output of a model as gospel without considering the assumptions underlying it.

One method of creating what is roughly analogous a confidence interval is running a reverse isochrone analysis and considering the envelope that it creates. What it effectively creates is an area that you can theoretically arrive at in the same amount of time. The thinner this band, the more uncertainty that's in the (routing) model.

And, of course, you can step through the gribs themselves to see why it's suggesting what it's suggesting, and take manual intervention when appropriate - e.g. is it sending you off on a wild goose chase to hook into a low in 6 days' time, the existence of which, let alone the location, is up for debate?


If you want to, you can download either the OpenCPN plugin, or qtvlm and have a play. Some polars are included, and others are relatively easily sourced online. And weather forecasts in grib form are very easily sourced.

Qtvlm's in-depth documentation is located here, and may be of some use.

 

[franksingleton]

Clearly it is well within the capabilities of modern software (I am avoiding calling it trivial) to find the optimum route given a single forecast, a decent tidal atlas and a set of polars for a given vessel.

But do any of the current routing software offerings provide sensitivity analysis, or more helpfully suggest routes that might be less than optimal given information at the time, but which are less sensitive to potential future changes in the forecast?

PredictWind, I believe, offer 3 or 4 routes based on different models. It is a kind of ensemble. However, ensembles run by NOAA have 30 members, those run by ECMWF have 50 members. A 3 or 4 member ensemble would seem too small to be useful. Of course, it may be that, over a route, random arrows average out and that the small sample is more use than a small ensemble for a short period forecast.-

 

[franksingleton]

I'm not sure I'd be quite so dismissive. Over the decades the tech's been in use, it's shown itself more useful than the alternative, in the right hands.

It probably depends on the route, length of passage and speed of vessel. On a long passage, what may seem to be a sensible route on the first few days might put you on a course that put you in danger ten days later. A storm might develop so quickly that a 5 knot yacht just cannot keep out of its way. I have quoted before a winning RTW navigator who said that he would not route his own, slow 40 footer.
On cross Channel passages, the difficulty in maintaining a steady speed, ship dodging, TSS and, sometimes, tidal constraints for port entry must be complications that make routing difficult. Forecasting might be good enough for routing over passages around, say 4 t0 8 days. But, even here, it only takes a small error in the overall pattern to give a significant error in the wind.
We sailors get annoyed when a forecast F4 becomes a F6, for example. I sometimes quote a W F4 up the English Channel will occur with, say, a pressure of 1004 hPa over Southampton and 1008 over St Malo. An increase of 1 hPa over St Malo would result in F5. An increase of 2 hPa would give a F6. Those are changes of 0.1 and 0.2 % in pressure. It is the old problem in physics of measuring or predicting differences in large quantities.
I am sure routing will continue to be used and that benefit will be seen. It is far from being a safeguard against trouble. I am trying to highlight that.

 

[Roberto]

It’s a routing and planning tool which uses boat speed to predict where you’ll be and when based on the wind. Given that the wind could vary by 12 knots due to tide that’s a rather important stat.
As said above though, they do do tides after all, it’s just very expensive

Have a look at "Qtvlm" for example, the PC version is free with full functionalities including routing with currents, different sea states, etc.
Grib for tidal currents can be downloaded here for example
Index of /MyOcean
they are big files so only available when one has a land internet connection.
Ocean current gribs are available through rtofs, oscar, etc etc.
IMHO weather routing for boats having speeds in the order of 5-6-7-8 knots is very often a waste of time (vs common sense, or having some sailing experience) we tried to modify wind grib values by a few knots and +-5° direction and likewise tiny variations in boat polars, all consistent with real life errors forecast/verification , then re-run the routings and the boat was sent in very different directions; it was so evident we did not even try to modify the time variable (same conditions but with 3-6-12 hour delays). Not sensitive in very standard situations such a big veer of wind for example, or no specific wind dynamics, when one does not really need a routing toolto know what to do.
I have not delved into it more than that, but possibly higher speed boats have a relatively reduced influence of errors. Also, they (Imoca et al) seem to spend so much time routing and routing that it is difficult to imagine such thing done in a normal boat, except for personal intellectual curiosity

 

[franksingleton]

Have a look at "Qtvlm" for example, the PC version is free with full functionalities including routing with currents, different sea states, etc.
Grib for tidal currents can be downloaded here for example
Index of /MyOcean
they are big files so only available when one has a land internet connection.
Ocean current gribs are available through rtofs, oscar, etc etc.
IMHO weather routing for boats having speeds in the order of 5-6-7-8 knots is very often a waste of time (vs common sense, or having some sailing experience) we tried to modify wind grib values by a few knots and +-5° direction and likewise tiny variations in boat polars, all consistent with real life errors forecast/verification , then re-run the routings and the boat was sent in very different directions; it was so evident we did not even try to modify the time variable (same conditions but with 3-6-12 hour delays). Not sensitive in very standard situations such a big veer of wind for example, or no specific wind dynamics, when one does not really need a routing toolto know what to do.
I have not delved into it more than that, but possibly higher speed boats have a relatively reduc. ed influence of errors. Also, they (Imoca et al) seem to spend so much time routing and routing that it is difficult to imagine such thing done in a normal boat, except for personal intellectual curiosity

Thank you, Roberto, for confirming my views on weather routing. These are a result of my long ago experience as a meteorologist working with routers, my knowledge of Weather models and as a sailor.