Digital Soundings and Water Depths from OFS Forecasts

The Operational Forecast System (OFS) model forecasts tide height and tidal currents for 15 locations around the country — a true revolution in modern marine navigation.

What is probably less known, is that we can potentially get the actual water depths for any point on the chart from these same forecasts.  These values should match the charted soundings and depth contours—to the extent that they are right, and indeed the OFS model bathymetry data are right as well.

Plus, we have to assume that the logic presented here is valid for extracting this information. So a main reason for this post is to have a way to ask the experts if this is a sound process.

When we then add the tide heights to the digital depths we have the forecasted water depth at any point in space and time, which would be another revolution in marine navigation. The concept of digital water depth has been planned to be part of the future S-100 electronic navigational charts (ENC), but I would like to show here that this is essentially available now.

When one of the OFS forecasts in netCDF format is downloaded from the NOAA AWS server and then opened in Panoply, we see these parameters from the San Francisco Bay model (SFBOFS).

u_eastward and v_northward are the vector components of the tidal current.

zetatomllw is the tide height, which is always relative to (above) MLLW.

But we also have

h, which is the depth of the water below MSL and 

zeta, which is the depth of the water above MSL.

(The parameter called Depth is just the number of depth layers where data are provided, which is 21, from 0 to 100 m.)

The diagram below shows how these parameters are related.

There are stand alone programs such as CDO that lets users combine parameters in a netCDF file and make a new file with the new parameters. So we have experimented with the process.

It seems we can get the total water level by just adding h and zeta, since they are both relative to MSL, even though that is not a datum used for this purpose in charting.

To obtain digital values of the soundings at any point on the chart, we need the depth relative to MLLW, not the h values in the native files, which are relative to MSL. The actual charted depths will be deeper than h by the difference between MLLW and MSL. 

But we can compute that value, which varies across a chart, because it is just the difference between zetatomllw and zeta, as shown in the diagram. Since in the nautical chart world, MLLW is the sounding datum defining zero tide height, this difference is just the tide height equivalent to MSL, which is a datum that NOAA lists for each of their tidal stations.

It is presented at tidesandcurrents.noaa.gov on each tidal station's home page. Below is a sample from Redwood City, CA.

We can then make a plot of this difference in Panoply and check for what it thinks this value is at each of the locations where the value is known. That plot looks like this:

The places where MSL is known in this area are shown in this figure.

In Panoply you can interrogate a point in a plot to get location and value, which we did at each of these locations. Samples are below.

The results are summarized in this table:

The agreement is good over a fairly large range of values, so it appears that this is a valid way to extract the MSL depth from the OFS data that we can use to compute chart depth from h.

Below is an example of a custom GRIB file made in the manner described and viewed in qtVlm—a popular free nav app for Mac and PC. It shows digitized chart depths in the region of SFBOFS just outside of the Golden Gate Bridge.

This shows the depth in feet, with a color gradient background designed to match the standard depth contours on US ENC. We end up with a display that is similar to an ENC depth area object  (DEPARE), but now we have digital values of the soundings any place on the chart. The famous Four Fathom Bank (yellow patch) stands out very nicely.

It will take more testing to be sure this is a productive useful addition to our navigation. We can now display the digital soundings (chart depths) and the digital water depth, which is chart depth + tide height.

It is a promising development, and new use of the OFS forecasts, but it will take some work to test its value.

Predicting Tidal Currents in the Swinomish Channel

Historically there have been no NOAA predictions for the tidal current speed in the Swinomish Channel flowing past La Conner, WA, but local mariners know it can be strong—over 2 kts at times.

Years ago we noted that this current should be predictable from the tide difference at the north and south ends of the channel—a driving force called a hydraulic head. The current flows from the high-tide end toward the low-tide end, and the bigger that difference, the stronger the current.

There is a NOAA tide station at La Conner (#9448558), 1.4 nmi into the 6-nmi long channel from the south and another right at the north end of the channel (#9448682). Typical channel widths are 300 to 400 ft, but navigable waters are narrower—dredged to minimum of 12 ft over 100 ft width. I note that the only chart of the Channel (US5WA31M) apparently has the dredge depth wrong at 6 ft. We have reported this to NOAA.

Historically mariners did not have digital tides on board, so it was tedious to copy the tides from tidesandcurrents.noaa.gov and transfer them to a spread sheet to make the current forecasts. I am not sure that method ever caught on with local mariners; we originally worked on this for a specific Seattle to Bellingham kayak race.

Whether or not that procedure ever became popular does not matter, because we have now an all new way to get presumably good forecasts for the channel current with the click of a button, thanks to the newly available results of the Operational Forecast System (OFS) model along with also new developments in how to read that data.

The driving force of the current and the way to forecast it remains true, but we have much better data now on the actual tide heights. The Salish Sea and Columbia River OFS model (SSCOFS) forecasts the current in the channel every hour out 3 days, and the model is recomputed every 6 hr to take into account changes in local environmental factors that can have major effects on tides and currents. The big advantage of the model forecasts is they take into account local values of wind, pressure, and river runoff, which the static harmonic NOAA station predictions do not account for.  

For the moment, we can see these new current forecasts online two ways. One is the NANOOS presentation, shown below. This works great for the Salish Sea data but there are not as good IOOS presentations for the other 14 regions around the US where we have OFS predictions.

The above source is established and dependable, but NOAA has a new online viewer called OceansMap and it promises to include the currents and a lot more information, with versatile display options. As of now, it is still in beta form, so sometimes not all data are available. Also for now we can only see the time scale in EST (PST +3h).

Historically there have been several local guidelines for predicting the channel currents based on a single value of the tide height at La Conner or Seattle. Google "Current speed in Swinomish Channel" to see a few of them. And indeed some may work in some average conditions, but they cannot be dependable because of the wide range of environmental changes plus the area frequently has unusual tide patterns, such as the above example taken at random with high tide nearly all day long—essentially a diurnal pattern where it is normally semi-diurnal. 

In the picture below we used the SSCOFS data viewed in OceansMap and stepped through a days data, one hour at a time, and recorded the tide heights at the north and south ends, as well as the current near La Conner.  Then we did the old school method of subtracting the tides and plotted that along with the current.

So we see very nicely that this is the driving force of the current, which just confirms our original approach with modern data.  

But we no longer need this analysis. We just open one of the two apps above and get the current. Eventually the OceansMap will be the working tool as it has that neat meteogram optional display at the bottom that could be copied and saved in a phone.

Below are a couple more of these meteograms that compare currents in the channel with tide height at La Conner. Sometimes there is a correlation; other times not.

Earlier models of the Victoria Clipper, traveling from downtown Seattle to Victoria, BC, did sometimes transit the Channel when conditions in Puget Sound were very bad. This OFS data should make that planning better in those cases where it might come up again. The new Clippers are larger, so such careful planning is even more crucial.  

The channel has a mean tide height of 6.1 ft, with mean high water of 9.4 ft and mean low water of 2.7 ft. With dependable tide and current predictions, the Channel might be a more frequent option to Deception Pass for low-powered vessels, which then have the bonus of a visit to La Conner, which is a popular NW destination for good food, good art, and friendly people.

_______

For completeness, let me add that the wonderful GRIB versions of the OFS data we can get from Expedition and LuckGrib are fantastic for optimum routing over inland waters, but they must compromise on the grid size when converting the NetCDF to the GRIB format, and the resulting resolution is not adequate to use for the narrow Swinomish Channel. For this we need the two resources cited above. Also it seems Panoply which does load the full original data files, cannot resolve the data to that level either.

The New Revolution in Tide and Current Predictions.

First published on OceanNavigator.com  / updated here Feb 17, 2025

We had one tidal revolution in 2021 when NOAA announced that they were discontinuing the decades-long use of Table 2 lists of secondary station corrections, and that there will no longer be any sanctioned annual Tide and Current Tables. Plus there will no longer be any international tidal data published by NOAA. Going forward, the way we get official tide and current predictions is go to tidesandcurrents.noaa.gov and create a monthly or annual table for specific stations as PDFs and then print them. It takes four pages per year, per station. This is a superior system, as we rarely needed the global coverage in the historic annual tables, plus the use of Tables 2 (one for tides and one for currents) was tedious, and, indeed, we learn now, not accurate in many cases. We still see in 2025 the discontinued 2020 Tables 2 in some third-party tide or current books, but it is important to know that much of that content is wrong.

The USCG have also now recognized that historic tidal predictions have not been valid since 2021 and the new round of license exams have removed all Table 2 references in lieu of the modern approach of direct data from tidesandcurrents.noaa.gov.  This is also now updated in all electronic navigational charts (ENC). All previous references to “NOAA Tide and Current Tables” have been removed.

But with that revolution still unknown to many mariners, we have a new one!  NOAA’s new Operational Forecast System (OFS) now produces digital tide and current forecasts that are superior to the traditional NOAA predictions, which are based on harmonic constants for each station. We now have tidal current forecasts uniformly over the full waterways, out two or three days, in fifteen regions and two channels around the country. 

Figure 1. Regions where there are OFS digital tide and current forecasts. See tidesandcurrents.noaa.gov/models

The beauty of the OFS model forecasts is they take into account the local values of wind and pressure, as well as unseasonal river runoff. The models are updated four times a day to account for changes in these local environmental factors that affect tide height and current flow. The model data has a latency of  about 2 hours, meaning a 3-day set of hourly forecasts run at 12z will be available to mariners at about 14z.

The other huge improvement are the OFS current directions. Traditional harmonic currents are presented as pure reversing currents with just two directions, being the average flood and average ebb directions. But most open water currents are rotating currents to some extent, which do not have just two directions. An example is in Figure 2.

Figure 2. A comparison of OFS model forecasts at the location of a specific harmonic station on Feb 4, 2025, UTC to the harmonic predictions at that station. 

Where to get OFS tide and current forecasts.

For the time being, NOAA presents the OFS forecasts as graphic animations such as shown in Figure 3. These animations are not a very precise way to access this very precise data, but they are working on other presentations. In the meantime, third party navigation apps have solved this problem for us, which we come back to shortly.

Figure 3. Sample OFS currents as presented by NOAA. To access these, start at tidesandcurrents.noaa.gov/models and  choose a region on the left. Then scroll to the bottom of the page, and under currents click Forecast Guidance. If that link is not there, then click any subdomain indicated on the main image, and then look for the current link.

NOAA is also working on a new OceansMap web app that promises to be a sophisticated digital display that replaces the Figure 3 animations. A sample from the beta version is shown in Figure 4.

Figure 4. San Francisco OFS currents displayed in the forthcoming NOAA OceansMap web page. When completed, it will also show the harmonic predictions as well as the NDBC buoys that measure currents for direct validation checks. Please keep in mind that this is still a developing beta and all features may not work yet as intended.

In the meantime, we also have presentations from other agencies. A particularly nice one is from the Northwest Association of Networked Ocean Observing Systems (NANOOS) for the Salish Sea region shown in Figure 5.

Figure 5. NANOOS presentation of the Salish Sea OFS data, continuously updated. Click any point on the map to read the set and drift of the current. The value of this data has been confirmed by sailors in Port Townsend Bay. Other members of the Integrated Ocean Observing System (IOOS) have related presentations of OFS data.

These graphic presentations show us the general flow of the tidal currents, revealing patterns we would never know from the isolated harmonic station predictions alone, but for actual navigation underway we need the digital data in GRIB format. This way we can load it into navigation programs and compute optimum routing for all classes of vessels, but this is specifically crucial to sailors and low-powered craft. The problem is the official data are only published in NetCDF format, which most nav apps cannot read.

But mariners can be grateful to two marine navigation apps who have taken it on their own to convert this crucial data to GRIB format. They are Expedition and LuckGrib. The former is a popular racing and performance PC app, and the latter is a state of the art marine weather data source and display for Mac and iOS. Both can incorporate OFS currents into optimum inland routing computations. Both apps also allow users to export the OFS grib files they created. LuckGrib has a two-week, full-function demo period, so users can experiment with this OFS data and other features it offers.

Figure 6 shows the Salish Sea OFS grib file exported from LuckGrib and then loaded it into qtVlm, another nav app which is chosen here because it can display the OFS forecasts as well as the NOAA harmonic station predictions so we can compare the two current sources. qtVlm is a free app for Mac or PC.

Figure 6. Salish Sea OFS forecasts (black arrows) compared to the harmonic forecasts (colored arrows). In the gray labels, M is the OFS Model forecasts, and T are the Tabulated NOAA harmonic predictions. The yellow labels show the model values we would not know from harmonic predictions alone. We see that at the harmonic station locations (circled) the speeds are usually pretty close, which gives us confidence that the model data are right at other locations. The differences in directions between model and harmonic currents can be larger in between the peak and slack samples shown here. We are reminded in the slack data (bottom picture), that slack water is rarely still water —  the model data makes this even more apparent.

Usually tidal currents affect our navigation more than the tide heights themselves, but there can be exceptions, and the OFS model includes tide heights that can help with this. One example would be predicting current flow along a narrow that has no harmonic predictions nor OFS current forecasts for the channel. Such cases are usually controlled by the tide height at each end, with current flowing from the higher-tide side toward the lower-tide side.  An example is shown in Figure 7.

Figure 7. Salish Sea OFS tide height forecasts (background colors, mostly green) in the region of the Swinomish Channel that flows past La Conner, WA, displayed in qtVlm along with the harmonic predictions (small meters) at several locations.  Local knowledge calls for the current to start flowing north sometime between 2.5 to 4 hours before high water (HW) at La Conner, and last till the same interval past HW — and even this broad prescription is known to be sensitive to the state of local river runoff and the range of the tide. The OFS tides include the effects of present and forecasted winds and river runoff, so it is likely that the OFS tide forecasts can be used for more precise predictions of these currents. In this example, there is a notable slope in tide height across the channel 6 hours before HW.

_______________

When this post was first made, we could not read SSCOFS on the OceansMap and we could not see any details of the currents in Swinomish channel from the GRIB coversions as they have to average the grid and cannot go to that resolution.

But now we see that the OFS can indeed predict currents in Swinomish Channel and they look very promising. 

The above is from the OceansMap viewer; below is from the NANOOSH viewer.

Now we are in a position to test the local knowledge guidelines, which do not account for variations in environmental factors, and likely create a more dependable way to know the currents.  Could be as easy as just looking at one of the two sources above.

 

Squall Forecasts

First published on OceanNavigator.com 

It is likely known that we can get GRIB formatted wind and pressure forecasts from numerical weather models such as GFS. But it is probably less known that we can get usable squall forecasts as well. We get this from the output parameter composite reflectivity (REFC), often called “simulated weather radar,” which is effectively what it is. Once we are in an area of squalls, we can indeed watch them and maneuver around or with them using our marine radar, but it is often valuable to know when they are likely, how severe they might be, and how they will move.

It was not that long ago that navigators beat themselves up chasing atmospheric instability parameters such as CAPE (convective available potential energy) and CIN (convective inhibition) and LI (lifted index) hoping to piece together a usable probability of squalls and their severity—with, I venture to guess, much the same success I had, minimal at best. Now we have a new generation of navigators who can skip all of that, and let the models do the stability analysis, and report it to us as a nice weather radar image right on our chart screens. It is in a sense like new navigators now never having to struggle with the Table 2 tide and current secondary-station corrections, which were, thank goodness, discontinued in 2021.

We can see what live weather radar looks like nationwide at radar.weather.gov. Our textbook Modern Marine Weather has an extended section on the interpretation of REFC.

Figure 1. Sample weather radar. From Modern Marine Weather.

Figure 2. Unofficial guidelines for relating dBZ to squall intensity. From Modern Marine Weather.

The units of reflectivity (Z) are complex and logarithmic (see noaa.gov/jetstream/reflectivity), so they have been simplified to decibels as dBZ. There is no official scale for squall wind intensity, but we made a rough correlation with thunderstorms (rain based) in Figure 2, which has proven practicable. We thus anticipate severe squalls for dBZ values above 40 or so. Squall conditions are most severe with fastest onset where the dBZ gradient is steep, meaning color change from blue to red is narrow.

Besides the global model GFS, the regional model HRRR also provides REFC. Gribs of both models are available by email request from Saildocs. REFC is also included in the high-res NAM models. A sample is shown in Figure 3.

Figure 3. REFC display from the NAM-Puerto-Rico model downloaded and displayed in LuckGrib (luckgrib. com). Forecasted squall winds of 32 kts, gusting to 36, in an area with REFC about 62 dBZ.

This is just a 4-hr forecast, but the general information would have been known earlier. These data are best in the regional models with higher resolution and more frequent updates. The GFS and NAM are only updated every 6 hr, but the HRRR is updated hourly, so it can be useful for near-live squall forecasting in local waters.

You can test these forecasts by looking at the actual weather radar for the same region and time, as shown in Figure 4.

Figure 4. Sample weather radar at about the same time and place as the NAM forecast in Figure 3.

To practice with this, look at the national radar map to find squalls (Florida has the most) and then compare to an HRRR REFC forecast for the area. The hourly updates of HRRR extend out 18 hr, except those run at the synoptic times (00, 06, 12, 18 UTC) that extend out to 48 hr.

Note in passing that the HRRR forecasts for all parameters might be your best local weather forecast available, especially in remote parts of the country.