Using routing software for long offshore passages

[Tommo]

Just wondering how people use routing software on long passages. Having played with a few trial packages, they use GRIBS and can only really cope with routing for a handful of days ahead. If the passage is going to take longer than this, they simply try to route to a point along the rhumb line. Clearly this is not very clever!

So if going on a long passage, do you first plan a "strategic" route making use of historic wind data, then use the GRIB planners just to make short-term "tactical" route along the "strategic" route? Or is there any clever (and no doubt very expensive!) routing software that can do this automatically?

[rgleason]

There are several you can use but

Opencpn + Grib_pi + Climatology_pi (similar to pilot charts but in a format for routing) + Weather_Routing_pi

offers routing based on 30 year NOAA averages, and uses the current grib files for the first part of the routing and Climatology for the longer extended part.

Climatology_pi as average winds for each month. Pretty reliably shows the trade routes but does not differentiate for La Nina, El nino and other year specific conditions.

[Paul Elliott]

Expedition (expensive software) can route you using your gribfiles, and take the last available conditions and hold these for days / weeks, routing you as if the conditions remained constant. You can get gribs that forecast out beyond a week, although the accuracy is questionable beyond perhaps five days.

This "assume conditions don't change" method isn't too bad when conditions are settled. I doubt that you can do much better using historical pilot chart data.

It's the weather. It all depends...

[rgleason]

Paul

This "assume conditions don't change" method isn't too bad when conditions are settled. I doubt that you can do much better using historical pilot chart data. It's the weather. It all depends...

Actually on some routes I think the historical pilot chart data will be helpful. However I am interested in the Expedition technique

..using your gribfiles, and take the last available conditions and hold these for days / weeks, routing you as if the conditions remained constant.

How is the extent of the "last available conditions" defined or selected?
I would like to ask Sean if he could implement this technique in the weather_routing_pi plugin.

[conachair]

I played around a bit with something similar ages ago, might have been a maxsea add on.
Gribs were too inaccurate to be of any use more than a few days out.
These days offshore it's keep the boat happy and with hopefully you'll be heading roughly in the right direction, then just enjoy it

[Roy M]

I've been watching several websites over the last few months that provide a lot of interesting information, leading to a greater appreciation of what's going on up there. These sites are centered on my area of interest: the Eastern Pacific. You can select your own areas to study. Here they are:

Animation of Jet Stream Analyses for Eastern Pacific/Western North America - this one is an animation of the jet stream which can be programmed for watching what's been happening for the last few days. Storms are strongly influenced by the courses of the jet stream.

earth :: a global map of wind, weather, and ocean conditions - this was my first favorite and shows the surface wind patterns as an animation for the last three hours.

https://www.windyty.com/?32.795,-116.963,2 - this is my latest new favorite. You can play with it to see different elevations. 10 Km gives a pretty close simulation of what the jet stream is doing and helps corroborate the predictions. It's especially cool because, by manipulating the slide bar at the bottom, you can get an approximation of what will be happening in the next ten days or so. If you click on the sidebar to see clouds, while staying at the 10 Km altitude it appears to track the fronts as they move through.

I'm not a meteorologist, but I can appreciate the value of this info.

Ocean Prediction Center - Pacific - and finally, a traditional weather map to help me understand what information it contains and relate it to the other sources.

Too bad I can't get this stuff a thousand miles out in the Pacific. Not yet, that is.

[Paul Elliott]

Originally Posted by rgleason

Paul


Actually on some routes I think the historical pilot chart data will be helpful. However I am interested in the Expedition technique



How is the extent of the "last available conditions" defined or selected?
I would like to ask Sean if he could implement this technique in the weather_routing_pi plugin.

As a wx routing option, Expedition will use the gribs you have loaded up, and if comes to the end of the grib data before the passage you're trying to route is finished it can just take the last available (furthest in the future) data and continue routing as if that data/weather remains static.

Obviously any kind of wx routing beyond perhaps a week, or using pilot chart data, is going to be approximate at best. If you sail into a stray high-pressure wind hole, or a ridge moves out further than you had anticipated, you can spend a couple of days going nowhere with the sails hanging limp. Fortunately, once we actually get out there we can update our WX info and have at least a few days of useful forecast in front of us. This gives us time to make adjustments in our route.

I consider long-term route predictions to be a useful estimate of likely course / speed, but hardly something to rely on. And of course, computer-aided wx routing requires some care in configuring and interpreting. I've seen wildly different routes result from very minor tweaking of my polars, or slight changes to the grib data. It's necessary to choose your route with this in mind, often going for the route that has the "least bad" possibilities when conditions change (as they will).

[zboss]

I had the same question.. what are the best practices around using such sites?

Do you leave your rhumb line in place and then recalculate or do you reset the starting position of the rhumb line at your current position and recalculate?

How often do you recalculate? Daily?

Do you take the average of the next nth positions and make your course to the average or do you follow it perfectly?

[seandepagnier]

for opencpn weather routing plugin, it can use the last available data when the grib runs out, or it can use historic prevailing wind data with various options for how to interpret it.

I would like to provide historic grib data for this purpose so you can make routes for the last 10 seasons for your intended route and compare their results based on actual conditions.

An other thing I am contemplating is a different type of weather router. Most weather routing is mostly focused on minimizing sailing time to reach a destination. Instead we might not care much about this and instead optimize all sorts of other parameters, like sea comfort, avoiding thunderstorms and shipping traffic, avoiding strong currents or reaching several different locations.

[Paul Elliott]

Originally Posted by zboss

I had the same question.. what are the best practices around using such sites?

Do you leave your rhumb line in place and then recalculate or do you reset the starting position of the rhumb line at your current position and recalculate?

How often do you recalculate? Daily?

Do you take the average of the next nth positions and make your course to the average or do you follow it perfectly?

I use Expedition WX routing when racing to Hawaii and cruising back to the mainland. Before I describe my technique, I should tell you this:

• I'm not a hard-core racer.
• The polars I use for my boat are estimated from previous passages, using the recorded wind speed / angle and boatspeed. I don't consider sea state, the particular sails I may have had up, or how hard we were trying at the time.
• I adjust the polars underway to try and align them to the performance we are actually getting.
• If I don't like a suggested route I will make up my own.

So what I do is download GRIBs twice a day. Sometimes when conditions justify it I get a high-resolution (1/2 degree, six hour steps) GRIB that covers the area just ahead of me. I always get a 7-day grib with a 1 or 2 degree grid, with the time steps getting larger as the forecast moves out. I figure that there's no reason to download lots of high-resolution / low-accuracy data. I'm looking for the big picture in the extended forecast.

I also download a 24-hour synoptic chart, mainly to spot early-stage tropical depressions. If there are hurricanes / storms brewing I will get more synoptic charts.

I re-run my WX routing whenever I get a new GRIB, starting from my current position. No matter what my initial route had been, my job is to get from where I am to my destination.

If I think that the computer-generated route is taking me too close to potential problems (such as a high-pressure ridge that may decide to extend beyond its forecasted limits), I will modify my course to give me some extra clearance. I usually place a waypoint that forces the route to do this for me, otherwise I just set a course to steer and re-evaluate my route at the next opportunity. If a route puts me in or near strong wind or big seas I may to the same type of route modification. Whenever possible I leave myself a "bugout" option should conditions get too extreme.

I generally don't worry about the finer details of the route, but instead sail to the major inflection points. My polars and sailing technique aren't refined enough to benefit from one or five-degree adjustments. The real pros do follow their routes more closely, but they have the skills and tools to take advantage of a 1/10 kt speed improvement.

For all that, I would probably do just as well if I only looked at the daily synoptic charts!

[svTOTEM]

Paul's practical approach covers the point well. On Totem, our methods are similar.

We just finished crossing the Indian Ocean. Conditions were more challenging than the Pacific Ocean, not because of wind or sea state (Pacific was much windier and generally lumpier for us), but because currents were on average stronger and meandering. Based on that here are few additional things to think about – some are obvious but perhaps worth mentioning.


-As a sailmaker with full-on racing background, I approach passage planning and routing underway as a dynamic process much like doing tactics while racing. My goal is to get the boat to each point as efficiently as possible. The only difference between racing and cruising tactics is an additional comfort element. This isn’t a statement of constantly trimming sails or pushing boat speed. It’s about saving a day or more from a passage by going 40 miles out of the way to avoid foul current that cuts speed by 50% - and a million or scenarios.


-Some passages are easy: point and go without risk of apposing currents or adverse winds. Other passages not so much. This seams an obvious point, but I see many passages planned only for going A to B without hitting anything and leaving in reasonably good weather – then hear bitching about how long it took because of bad current (even though it was avoidable).


-Tools: GRIBS, synoptic weather, routing software, understanding VMG, understanding polars, knowing the boat well, self proclaimed weather gurus, etc. are all tools for the mental algorithm that is routing. Understand the limitations of each tool.


-Preplanning: in the tricky passage from Madagascar to South Africa, my starting strategy changed wildly, several times. The aggregate of conditions changed the pros/cons of each general route. I don’t pick a route based on what Mr. Salty did the year before or because that’s the direction others are going or because of historic weather data (except for big picture routing with regards to cyclones/gales/etc).


-Underway: On simple passages I update wind and sea state forecasts once a day from at least 2 sources (usually GFS GRIB and synoptic). As things get more challenging I use wind, current, and sea state from multiple GRIB models along, synoptic weather, and PredictWind routing software. Most importantly, is to compare forecasted and actual weather – and if required adjust course as necessary for better VMG and comfort or to setup for better location for coming weather change.


-The shortest distance is often not the most efficient path. Draw a rhumbline for reference, but don’t worry about straying from it. Variables change, so adapt. Make the most of present conditions to setup for what is likely to be over the next days.


-Weather wins: everyone gets caught in unpleasant weather. Blaming inaccurate forecasts and GRIB files or routing software or self proclaimed weather gurus is easy. They’re all predictions! And there is more to the 25 knots or 3 meter waves the GRIB shows Understand the numbers). Yes, they’re completely wrong sometimes and under/overstated sometimes, and less accurate the further into the future you get. And they're good to really good most of the time; but if you expect perfection then stay at the dock or stick to observation and barometer (both great tools, but very limited).


-Routing software: can be a great tool, especially if you don’t have good practical experience mashing variables into a course that is faster, safer, and more fun than a just-don’t-hit-anything strategy. Routing software puts wind and current info AND boat polars (theoretical speed for any wind direction/velocity) into an algorithm to arrive at a route based on your choice of fastest or most comfortable path. It’s not a road map though! This is 3 different predictions used to create a best path – subject to accuracy of each. For most cruisers, the polars that the user enters are far less accurate than a 3 day GRIB forecast (but it’s easier to blame the GRIBS). We’ve done very well with PredictWind routing software – though as with all routing software it takes some effort and understanding.


Passage tools are tremendously helpful and better than ever and still imperfect. The best passages come from a crew’s ability to use a range of tools, use them well, form their own conclusion, and adjust along the way – imperfect as it may be.

[rgleason]

Sv Totem How often do you get gribs, via what means, and what size and resolution? Do you use stepping gribs? Or are you using predictwind on globalsat?

[svTOTEM]

rgleason - easy passages once a day, of entire area looking for changes. On tougher passages 1 to 2 times a day depending on challenges. I generally use the 4 different wind GRIB models by PredictWind via Iridium Go! at resolution of 1/2 degree. Current GRIBs I get via SSB/Airmail with resolution of 1/4 mile (lower res I find misses to much). I use fixed area GRIBs, not moving, for bigger picture perspective. PredictWind routing software produces a route that projects your moving location (which is why accurate polars are so important) with wind/sea/current conditions overlaid for that time - so essentially a moving GRIB.


Iridium Go! is new to us in the last year - hardware is nice, software is buggy and unrefined, but getting better -
http://www.sailingtotem.com/2015/02/iridium-go-predictwind-for-totems-indian-ocean-debut.html

[estarzinger]

Not mentioned so far . . . . For me perhaps the single most important skill to develop, in order to use grubs and routing well, is the ability to assess the forecast confidence (that is how likely is it that this forecast actually occurs). If the forecast has high confidence then you can do quite clever routing, going off the GC line, going close to features . . . . But if the forecast has low confidence then you do not want gamble on any particular weather feature (occurring or occurring where forecast).

Forecast confidence varies highly. Sometimes the next couple days are quite certain, while other times it is a complete crap shoot. There are several "signs" of confidence - #1 if it gets "current conditions" correct that is a good start , #2 if the forecast has been the same for the last several GRIB runs that is nice, #3 if the forecast is the same between several different models that is nice, #4 and finally there are several different types of patterns that just naturally lead to uncertainty and low confidence - like a very large weak low pressure pattern without well defined gradients, you (and the computer models) just can't know what and where things will develop out of this sort of background pattern.

[barnakiel]

@ Tommo,

You asked a simple question that does not have a quick and simple answer. I will try to be brief. You can always PM me if I you need more detailed view.

Basically, route planning is done in two stages:

- pre-trip, on-land - based on stats and data available at the start,

- underway (either land based or onboard) - based on data available at the point.

Pre-trip planing is done based on charts, pilots and publications (e.g. the excellent BA NP 136) and is concerned mostly with finding the general route and any alternatives.

Once this data is collected and processed and some time ahead of the departure, the router will start monitoring wx data sources (often NOAA graphical products and various grib processors) for general idea of how patterns are developing in this specific year/season. Later, nearly at the start of the trip, the router will likely feed some of the data into wx routing modules of selected software. This software will return suggestions for track depending on whether one looks for comfort, fast passage or any other criterion (e.g. very light conditions, no icing, no fog, etc.).

And so we arrive at the departure point and on the departure day the latest software generated routes will be compared against the existing conditions and the one that fits best may be followed at the early stage of the trip (24, maybe 48 hours).

Now once on the water, further routing will be done either by onboard navigator, an onshore router, or both (pro events mostly, but also big maxi charter boats, serious ascents, i.a.). The weapons of choice are graphic data from various agencies, grib files processed by routening software and, last but not least, local observations. The frequency and style of work will depend largely on how important some factors are (time, comfort, fuel consumption, etc.) and to similar extent on how skilled the crew are and how strong the technology onboard is. How strong to bias for the computer calculated decisions will too vary greatly depending on how much discrepancy/divergence is found, how well the models performed in past situations and how competitive the event is.

The process looks similar for sailing and for other sports. It always makes huge sense for the involved (be it a climber, a kayaker, an ultramarathoner or a sailor) to understand the wx as much as possible, not only because one can get cut off from their router (human or technological) completely, suddenly and permanently, but also so that the athlete/adventurer and the router (be it a person or a track-me-wx App) 'talk the same language'. All technology can be great, but we will always be limited by this exact technology as well as by the human factor.

OK. So this is the framework of how some people do it. Depending on experience, skill and actual needs your method may contain more or fewer building blocks. And departing 'as is when is' happens too. ;-)

I can imagine that you like to play with toys, so here are some:

pre-passage: e.g. Visual Passage Planer
on-passage: e.g. Adrena

There are plenty of alternatives. If you actually want to perform routening tasks, you will likely look at all alternatives and chose the one that best fits your personal preferences for workflow and interface.

Happy routing. PM me if you encounter any specific challenges that can't be overcome with some googling.

Fair winds,
barnakiel