Wednesday, May 8, 2024

Landmark Labels on ENC

We are one of the first in line to lament the poor coverage of terrestrial charting in electronic navigational charts (ENC) compared to the paper chart coverage we are used to.  And for good reason: we do most piloting relative to landmarks and much of the land mass on ENC is conspicuously blank—which can appear even more moonscape vacant depending how we have the display set up, as shown below.

Unlike viewing raster navigational charts (electronic copies of the paper charts), ENC let the user control many aspects of the display. Above we see an example of choosing to show "Important text only," which is a (misleading) official ENC display option.

If we compare that to what we see on the equivalent paper chart, we see what we are missing in that view.

It is not just the names that can be hidden, but NOAA ENC have very few elevation contours which can often help with piloting. 

Another reason we care about charted place names is a matter of basic safety and prudent seamanship. We teach that it is good policy to always keep in mind a verbal description of where you are, and maybe even note it in the log book that way, ie "Just passing west of Willow Island." Knowing this at all times we are prepared to describe our position over the radio in an emergency—which is much faster than finding, if you can, a read out of the Lat and Lon and reading that with its potential error.  Furthermore it makes the cruise more enduring if you learn these names as you go by.  The say-it-out-loud method is is also how we teach students to learn the stars in cel nav.

Thus these charted place names are valuable to navigation. But things are not so bad as they might appear.  They can be bad, as shown above, but they do not have to be. Below we turn on the text labels to see what we really do have in ENC.

The charted place names are actually all there on the ENC, they are just not as prominent as they are in the paper charts, which have the freedom to use large font sizes for some, and indeed print on a curve.

ENC have strict international rules of font size and orientation, although in some cases they do let labels (and associated symbols) move on the chart so that critical ones remain in view as you change the screen.  A folded paper chart on the chart table may be just hiding a note that a dashed line is marking a restricted Navy firing zone. On the equivalent ENC if you panned that notice off the screen it would suddenly reappear in a new position in view. 

In other words, the use of labels on an ENC is just one more aspect of the new chart reading skills we need to develop for ENC. We have to look at the charts in a new way. One thing that helps with this is the rule that  ENC chart symbols and labels stay the same size regardless of the display scale (zoom level). Thus crucial matters may become more apparent as we zoom into the region of interest.

The Future of ENC

As for other deficits of the terrestrial coverage of existing NOAA ENC, we can be confident that this will improve. First of all, a few nations do a better job with the elevation contours already, and the US certainly has extensive GIS data for all aspects of US mapping. 

To show that big agencies like NOAA should be able to solve this problem fairly soon, we can show how to do this ourselves already.  Beyond its outstanding ENC display presentation, the popular navigation and weather app qtVlm also offers the option to overlay on the chart GIS data as shape files  (.shp), a standard format for GIS data. 

In the sample below, I followed the instructions we have online  to add the roads to Lopez Island and water bodies and elevation contours to Blakely Island, and then (within qtVlm) limit the contours to the 100 ft intervals shown on the paper chart.

Once these are installed, we can get a tool tip presentation of the road names, heights of elevation contours, and related data for water bodies. In fact we learn there are more lakes on Blakely Island than the paper chart showed.

In other words, there is good reason to expect that the terrestrial coverage of future ENC will be even more valuable than that of the paper charts they are replacing. 

We are likely to see this take place first in the printed versions of the ENC called
NOAA Custom Chart (NCC). These are intended to be the (non-official) paper backups of the official ENC viewed on a computer screen or chart plotter.  NCC are user-created online from the NOAA NCC app that produces a PDF chart of the desired region, scale, and paper size, based on the ENC content for that region.  Then it is up to the user to get the chart printed at the chosen paper size.

It is during this NCC production that NOAA could offer the GIS overlay options such as elevation contours, roads, building, water bodies, etc to be added to the PDF they are creating... essentially just as qtVlm offers users the option as shown above. Thus we could end up with a new-generation of paper charts that are indeed superior to what we are now accustomed to.

Seeing this new data in the actual ENC themselves is likely further down the line. Even though a few other nations already have better contours, roads, and buildings, NOAA is likely pretty tied up with their massive process of rescheming all the ENC, which is a major ongoing improvement to the watery parts of the ENC. Not to mention that all nations are in the long process of preparing for the next generation of ENC, where the present IHO S-57 standard will be replaced by the new S-100 standard, which inherently includes a lot of new GIS content. These proposed changes are discussed in our text Introduction to Electronic Chart Navigation.

Monday, May 6, 2024

The World Sees Atmospheric Pressure at Work

This week is the 30th anniversary of the opening of the EuroTunnel (Chunnel) between England and France. The BBC commemorated the event with a story about the first underground meeting of the tunnels being dug from both sides that took place on Dec 1, 1990, four years before the actual opening of tunnel to traffic in 1994. They met roughly mid channel, with TV cameras at hand.

The fellow on the British side with orange t-shirt is Graham Fagg who in 2010 gave a description of the event, which can be heard on the BBC Witness program. In that recording from (3:59) to (4:28) we learn that when the hole was opened up big enough to walk through there came a sudden wind from the British side to the French side that was strong enough to blow his helmet off.  That wind is the subject at hand.

This wind is quite literally what we call in marine weather a channeled wind! It means the pressure on the UK side was higher than that on the French side and the area between the two sides was confined by a narrow channel. We just have a case here of a very narrow channel, not just steep hills on two sides.

Our goal is to estimate what that wind speed was, which is an exercise in resources—meaning, can we find the actual pressures at both ends at that time, and then can we make some semi-reasonable estimates of the wind speed.

Below shows the Chunnel viewed on Meltemus charts of UK in qtVlm, with overlaid ECMWF reanalyzed surface analysis for the approximated break-though time (11 to 12 UTC, Dec 1, 1990) when the wind was noted. (The New York Times had a good article about the event, but gave the wrong time of day due to a time zone error! — no link here as they no longer let non subscribers read their articles.)

The red line is the route of the Chunnel. The isobars are shown at 0.1mb spacing.  The inserts are meteogram plots of how the pressure varied throughout the day at both ends. The pressure gradient across the channel did not change from 11z to 12z, at (1036.0 - 1035.5)/26.9 = 0.5mb/26.9 nmi.

The ambient surface wind at this time was about 10 kts across the channel, but we must use wild approximations to estimate the wind in the tunnel.

We can for example just use the basic formula for wind responding to isobars that leads to the wind we see on the surface. We derive a simple formula for that in Table 2.4-1 of Modern Marine Weather:

 U = 40 kts/ [D x sin(Lat)]

Where D is the pressure gradient expressed in a special way. Namely it is the distance between 4-mb isobars expressed in degrees of Lat. On a map, we put dividers across adjacent isobars, then move that to the Lat scale. If the distance between the two isobars on either side of the point we care about is 180 nmi, then D = 3.0. 

So we have to convert our tunnel gradient to that format starting with: 0.5 mb = 26.9 nmi. 

0.5 mb x (4/0.5) = 4 mb = 26.9 nmi x (4/0.5) = 215.2 nmi = 3.58 Lat degrees (at 60 nmi per degree).

U = 40 kts / [3.58 x sin (51)] = 14.4 kts

Then for surface winds we have a surface friction reduction of 0.8 or so that leaves us with 11.5 kts, which essentially agrees with the observed surface winds—which should not be a surprise as that is the basic procedure used by the models, with a few subtle corrections.

The above is based on the physics of wind flow, but still a large stretch to project that thinking into the tunnel. It is at least a plausibility argument for the rough magnitude of  the wind.

In our textbook in Sec 6.2 on Wind Crossing Isobars (page 146) we give another way to approximate wind flow in channels that is purely empirical, meaning not computed, just observed. It is a rule we compiled based on how the local NWS forecasted wind speed (in the old days) in the Strait of Juan de Fuca and in the Puget Sound based on the pressure differences at each end of these channels. Our composite guideline is this:

Channel wind (kts) = 800 x Pressure gradient (mb/nmi),

which we can easily apply to what we know:

Channel wind = 800 x (0.5/26.9) = 14.9 kts.

So again, we see the order of magnitude of the wind speed we might expect in the tunnel.  And again, we cannot consider this rigorous science; wind flow in restrictions is very complex. We have just confirmed that indeed the wind was going the direction observed, and also about the right speed. We use this same approach to forecast or anticipate wind changes in our own waters based on pressure changes. It is part of our Local Weather web page.

There is also a physiological element of confirmation. The force of the wind is proportional to the wind speed squared. The force of 14 kts of wind is twice that of 10 kts of wind. At 17 kts it is three times stronger than 10 kts. In other words, there is a dramatic difference in what we experience in 10 kts vs even just 12 kts.

We know from our own experience that we could be in a wind of 10 or 11 kts that could blow our hat off if it hit at the right angle. And it would be noted, but not a focus point for any newscaster's story. But if this wind were much more, we know that it would be a focus of the conversation, which it wasn't. Note too that the wind came not at the moment when the flags were exchanged, but later when they had the hole opened up enough to walk through.

In other words, without any math or science considerations, we might guess the wind was about 8 to 10 kts, because less than that would not blow his helmet off, and much more than that would have clothes rippling in the wind and newscasters talking about it, which they did not.

Such visual effects of the wind is not unlike our view of whitecaps. At 10 kts there are some, if we look carefully; at 15 kts they are easier to see; but at 20 kts they are the dominant factor noted when looking at the water. 


Pressure remains a concern in all such tunnel travel due to the piston effect that can create high pressures in front of the train stressing gear and making travelers uncomfortable. The Chunnel has build in pressure escape valves all along the tunnel to prevent this.

Chunnel prices seem to be like Amtrak, which depends on availably, and season. 

Friday, April 26, 2024

Japan Weather — A Sample of What We Can Do For All Global Waters

We had an excellent question come up in class asking simply what are the best weather resources for the waters of Japan?  In our textbook (Modern Marine Weather) and in our training and resources app (Weather Trainer Live) we list all resources available worldwide and even have sections on specific regions, but realize it could be valuable to just focus in and list specific solutions for a sample area, with details needed to actually obtain the data underway.

So we use Japan for this example, stressing that these same sources (or counterparts) are available for essentially any part of the world.


(1) Model forecasts in grib format
The main workhorses we use anywhere will be the global model forecasts from GFS and ECMWF.  These data are available from saildocs with an email request to with this in the body of the message:



These files can then be viewed in any navigation app, such as qtVlm, OpenCPN, TimeZero, Coastal Explorer,  or Expedition or in a dedicated grib viewer such as XyGrib. Background on use of gribs at the Grib School. Luckgrib is a state of the art app for downloading and viewing grib data. 

Sample model forecasts. Red is GFS, blue is ECMWF

(2) Graphic weather maps
We need to check the pure model data from above with actual maps made by human meteorologists, and we have several sources of those. 

Above is a sample surface analysis (12z Apr 28) and below is the corresponding OPC map, which is as far west as they go (135E).

(2a) Weather maps from Japan

Analysis chart

24-hr forecast

48-hr forecast

These are pdfs of about 550 kb, too large for sat phones as a rule, but it won't be long till all mariners have high speed internet offshore, then this type of link becomes more valuable. In the meantime, a supporter on land can download the file, copy the image from the pdf, reduce the file size, and email it to you on the boat.

Graphic weather maps are also available by HF radiofax if your boat happens to have the SSB radio and antenna set up. Japan stations are listed in the Worldwide Marine Radiofacsimilie Broadcast Schedules.  The many JMA maps available this way are listed at the JMH radio Station.

The other important radio related resource for international voyages is the NGA Pub 117, Radio Navigational Aids.  This tells, for example, what time of day you get VHF storm and navigation warnings for different parts of Japan. NAVTEX broadcast times are also given.

HF Fax is frankly an outdated technology (replaced by satellite communications), but if we could access the folder JMA store the images in we could request the same maps by email the way we do the US maps.

(2b) US OPC maps covers NW Japan waters

US maps only go to 135E (sample above), but they could be helpful on the approach from the east. See the Starpath Pacific Briefings page for examples and links. These are easy to obtain by email from Saildocs or FTPmail.

(2c) You can see UK maps of Japan waters at and choose Eastern Asia region. These maps, however, might be just their model output plotted, which does not add knowledge.

(3) Satellite cloud pics
Japan has an excellent satellite image program (Himawari). See index to files here, which is also where you learn the file name you need to ask for.

The latest visible image (b13) for Japan area can  be requested by email from Saildocs is  

The last four digits are the UTC of the image, available every 10 min, i.e., 0000, 0010, 0320 etc.  Himawari data are also excellent throughout the South Pacific.

This sample image is from 2 days later than other examples shown.

(4) Near live ASCAT winds

To get near live ASCAT winds, follow articles we have online about it and use links like the following for Central Japan waters:

You can use the same set of links for Northern Japan by changing the file number to 253, and for Southern Japan use 242.  You can get these online or ask for them from Saildocs by email. For background see

Sample ASCAT pass

(5) Live ship reports
You can also get a list of all ship reports near Japan by sending a blank email to and put the central Lat Lon in the subject line, such as 37.0 N, 140.2 E. (See This gets a list of all the reports plus a GPX file of the reports that can be loaded into a nav app to see actual locations and data.

(6) Ocean Currents.
You can get ocean currents and SST for that region from RTOFS request to Saildocs


For background on currents see

(7) Waves and sea state.
GFS is best for this, again available from Saildocs. There are many sea state parameters (see Grib School list), but these are likely of interest most often:

Significant Wave Height of the Combined Seas (HTSGW)  

Primary Wave, Direction it comes from (DIRPW)

Primary Wave, Mean Period (PERPW)


(8) Tropical cyclone warnings and reports
Primary source is Japan Meteorological Agency, which is also the Regional Specialized Meteorological Center (RSMC).

To get the latest Japan waters text reports from Saildocs, use send Met.11por

To see how to get reports for other parts of the large metarea XI, use send metarea

Other metareas around the world...
(9) Special sources (with thanks to Mark D'Arcy for this reminder)
JMA, like other maritime nations, has weather models of their own, but the grib format is a paid service. You can see their MSM model at, but the higher-res LFM is paid only.

Many maritime nations also have universities or other agencies that run a localized version of the Weather Research & Forecasting Model (WRF). These high-res data can be very useful when available.  Japan and South Korea have WRF data available from selected resources, such as the weather and nav app Expedition

Monday, April 22, 2024

Choosing the Best Sextant Sights

 The criteria for choosing best sights is outlined in the Star Finder Book and in the manual to the StarPilot programs, but this online course has shown us that we need more specific notes on this. Again, this is the type of thing we generally covered in the classroom lectures, so it had been only sparsely covered in the printed materials to date.

The goal of sight selection is always to optimize the accuracy of the fix. If only two sights are available, then these would be ideally about 90° apart in bearing so that intersection errors are minimized. This is the same criteria used in choosing targets for compass bearing fixes in pilotage waters. If instead of 90° apart, the two targets were only 10° apart, then any small error in either of the bearings would cause a large error in the intersection when the LOPs were plotted. 

Use any chart and plot a bearing fix to two objects that are 10° apart and then repeat the plot assuming one of the lines is wrong by 3°. Look at how much the fix changes. Then do the same thing for two objects that are 90° apart at about the same distance off. When they are 90° apart the error will be 5% of the distance to the nearest mark, or something near that. But for the two close bearings the error will be much larger.

For a two-sight fix, this "scissor effect" on the shift of the intersection is minimum at 90°, but from a practical point of view, any LOP intersection angle more than about 30° will reduce most of this error enhancement, and you really don't gain much going to intersection angles above 60° or so.

The same reasoning applies to cel nav fixes. First if you are limited to two sights, then they should be some 30 or more degrees apart... ideally closer to 60 or 90. And, as always, you should take at least 4 sights of each object if you can, alternating at least the first couple of them. The reason for alternating, is to cover the situation where you just get two sights then something happens. If these are two of the same body, you are left with only an LOP, but if of different bodies you have a fix. Not the best you might have gotten, but at least a fix. For the record, I have been on two vessels where my sight taking was interrupted early in the process, and in both cases, a cel fix at the time was crucial. In the first case, the boat under spinnaker in the ocean broached on a wave, which kept us busy with immediate sailing issues even more crucial than improving the quality of the fix. In the second case, a pressure cooker exploded below decks, and the first aid issues again took precedence over the navigation. Needless to say, both of these examples are rare cases, but the goal of sound navigation is to develop procedures that cover you even in unusual circumstances. 

Figure 1. When LOPs are closer than some 30° apart, the fix errors are greatly enhanced (red lines) due to unavoidable uncertainties in the bearing lines. Here the error shown is 3°.

Also in cel nav we have limits on altitudes. Generally you would choose sights above about 15° and below about 75°. This is for two separate reasons. Low sights, especially down within say 5° of the horizon are more influenced by refraction. Refraction is the one uncertainty we do not have much control over in cel nav. We routinely make refraction corrections, but we are always vulnerable to abnormal refraction. In other words, mirages do indeed exist, and some are very prominent from the water in some circumstances. Mirages are an impressive demonstration of the presence of abnormal refraction. In the open ocean, when there is no land or vessels over the horizon to see mysteriously floating above the horizon in a mirage, we have no way to know that abnormal refraction is present, so we have to just be vigilant. There are special tables in the Nautical Almanac for correcting for abnormal refraction based on temperature and pressure, but what is not stated in these tables is the fact that the uncertainty in these corrections are about as large as the corrections themselves. Indeed, we do not even include these in our routine procedures, unless you are forced to take low sights, in which case they probably statistically would be right more often than wrong. Sights within some 5° of the horizon might be off by as much as 5 miles or so, even with the special corrections. Not always, and maybe even not likely, but definitely possible.

The best bet is to just avoid low sights whenever possible. Refraction correction is about 35' on the horizon, then 10' at 5° and 5' at 10° and then it just gets smaller as the elevation (Hs) gets larger. Look at the altitude correction for stars, since that is pure refraction.

The reason for avoiding high sights is completely different. There are two reasons for avoiding high sights. One is they are harder to take because the bodies are nearly overhead, which makes it difficult to tell which way to point the sextant as you rock it. For high sights it is easy to be misled into thinking you have the body aligned with the horizon when you do not. Hence if you do get stuck and need to take very high sights, be aware of this issue when rocking the sextant.

The other problem with very high sights is the sight reduction process itself. For high sights the LOPs cannot be accurately approximated as straight lines (which is our normal procedure), since the the circle of position now has a relatively small radius. Later we will add a section on processing high sights, for now the issue is just to avoid them if possible. If you are eventually using the StarPilot for sight reduction, then this issue is taken care of automatically, but when sight reducing and plotting by hand, we need special procedures for sights above some 75°. It is not difficult, and does not require special tables or compuations, but it is different.

Summary so far: for two sights only, choose the two bodies as close to 90° apart as possible and find bodies that are above 15 and below 75 degrees in height, and then take 4 sights of each to average for the two best LOPs. Sight averaging is covered in the course book, chapter 11.

But... with all that said, two stars (even in the right elevation range) are not the best option in the first place. Three stars are much more valuable for an accurate fix. Even if you take multiple sights of the two bodies, which reduces your statistical errors from any one measurement, you still are left with just two LOPs, the average from each set. You do get a picture from the plot of the LOPs what this level of uncertainty is — the more they are spread out the more uncertain the fix is — but you do not learn anything about systematic, or constant errors that might apply to each sight.

That is the value of choosing 3 sights that are about 120° apart. In this configuration, any constant error in each sight simply makes the triangle of LOPs (called the "cocked hat") larger, but the center of the fix remains an accurate position. This is not the case with 3 sights that are 60° apart, even though the final cocked hat of intersections might look identical. As time permits, we will add numeric examples to illustrate this important point, for now, however, the main goal is to explain the rest of the criteria beyond the geometry.

Figure 2. This is the way 3 LOPs would appear if there were no errors at all in the sights and they were reduced using the true position as the AP for each sight.

The top picture is for 3 sights taken 120° apart, the bottom is for 3 sights taken 60° apart. 

Assume that Hc = 30° 20' and Ho = 30° 20' for each sight

While the choice of geometry (selecting 3 bodies as near 120° apart as possible) will always be the dominant criteria in selecting bodies (along with above 15 and below 75° high), there are other criteria as well. The other factor is brightness. You can take more accurate sights of ones you can see clearly. So when all else is equal, or about equal, then choose the triad that includes 3 of the brightest stars. For example, if you have 3 stars that are very near 120 apart, but one of them is a magnitude 2.5 star, then you would almost certainly get a better fix from 3 that were, say, 130 and 110, 120 apart. In other words, you can tweak your choice to give up 10 or 20 degrees in optimum angle in exchange for brightness. There is a big difference in apparent brightness between a magnitude 2.3 star and a magnitude 1.5 star. See the table in the Star Finder book which coverts the magnitude scale to perceived brightness.

Figure 3. Now we show the same sights as above, but now assume there is a contstant 5' error in each sight, ie the sextant read 5' too high on each sight. We now have Hc = 30° 20' and Ho = 30° 25' for each sight, which gives a = 5' T 060, 180, and 300 for the top sights, taken 120° apart and a = 5' T 300, 000, 060 for the bottom sights taken 60° apart. They are all again reduced from the true position.

Note that the center of the top sights is still the proper fix, even with a constant 5' error, but in the bottom case, if we chose the center of the triangle as the fix, we would not get the right answer.

The main point is, we do not know what the error is, so we can't guess ahead of time where the fix should be for 3 sights 60° apart. We only know that the final uncertainty is larger than we would guess from the size of the "cocked hat" of intersections.

Once you have chosen several possible triads that have comparable quality on spacing and brightness, the final criteria would be to choose the triad that has the 3 stars at about the same height. This is again because of refraction. A star at 70° has a different refraction correction than one at 20° and if you happen to be in a case with abnormal refraction, you will magnify this effect by having stars at different heights. Again, the goal is to take advantage of the 120° geometry. If we have a refraction uncertainty, then to first approximation it will be the same error for all 3 stars if they are at about the same height. And if the error is the same for each sight, it will cancel out with the 120° geometry. We have of course removed the main effect of refraction by limiting all sights to above 15°, but this is now the third level of choice criteria, which is really fine tuning the process. The first filter kept our unknown errors below a 2 or 3', this final choice might help us get to the optimum accuracy of some 0.5' or so.... all providing we have taken into account the motion of the boat properly. 

A graphic reminder that when we have a choice, we choose sights above 15° and below 75°. Naturally, if there are no other options, we take any sights we can and keep in mind the special issues of each region.

If you do not advance all sights properly, then you loose accuracy according to your speed and time in the sight process. If I am moving at 6 kts, and take 30 minutes to do my sights, then I have a 3 mile uncertainty floating around that will mask much of this fine tuning in star choice if I do not correct for it. This again, is a virtue of the StarPilot or other computer or calculator based sight reduction. All sight reductions automatically advance all sights to the time you ask for the fix.

In the top picture, we give up a superior 120° spacing in favor of a brighter star that has fairly good spacing.

In the bottom picture, we sacrafice a bit of spacing for 3 stars at about the same altitude... or more to the point, to avoid one that is rather different than the other two.

All of these choices are fluid. The general criteria is discussed in the text, and from that you make your best choices and try options if you have the opportunity. Or take them all and do the fixes in various triad combinations to learn more of the practical matters.

For the record, in the StarPilot program, which is the only software available that actually sorts out and selects best sights from any sky, we use as a default weighted criteria: 70% on geometry, 20% on brightness, and 10% on relative altitudes, with Hc max = 75 and Hc min = 15. Each of these criteria can be adjusted by the user.