All the links I use are on my page.

Stormsurf has a tutorial and a do your own prediction page, too.
This page provides a basic explanation of the mechanisms that underly the waves we surf from their generation to their cresting. The winds thousands of miles away create a ripple, and five days later you are driving around a bottom turn and pulling into a tube. But how can you learn to tell when surf will hit your beaches? This page is here to help. This page will explain what a wave group is, how to identify a significant swell, how to determine if the winds are pointing at your beach, guesstimating the dominant wave period, arrival time, arrival size, how to use local buoys to refine predictions, and some brief notes on predicting local winds. But first, wave groups.

Wave groups - how waves travel

The waves we surf locally are wind generated. When wind blows over water, it creates waves that travel in the direction of the wind. As each waves passes through water, it causes the local water to move in a circular pattern. Down and back, then up, then forward, then down and forward. This circular pattern creates momentum. Because each individual wave creates its own momentum pattern that is matched by the momentum pattern of other waves at the same speed, waves will tend to synchronize with each other and form wave groups. These wave groups are a wave train, a set of waves travelling together. The group moves in the same direction as the wave, at half the wave speed. Within the group, each wave moves forward from the back of the group to the front. At the front, the wave has to create momentum from water that has none, so the front wave gradually disappears. At the back of the group, the wave leaves the water with some momentum that gradually builds into a wave. The waves in the middle share in the transfer of momentum, and do not change in size, but move within the group. Waves move, like on a conveyor belt, from the back of the group to the front.



Wave group conceptual


If you ever have a chance to fly over a solid swell, you can watch wave groups travel, and waves moving within the group. If the groups are well-separated, and the swell clean, these groups become sets of waves at the beach. The number of waves per set is based on how "fully developed" the seas become. If the wind is consistent in direction and velocity for a long time, over a large area, the wind will create lots of single wave travelling at the same speed. As they synchronize with one another, the wave groups will increase in the number of waves per group. With particularly well developed seas, wave groups with 10+ waves can result. More often, there are fewer waves per group, and the storm has variable winds from hour to hour and mile to mile which prevent the seas from reaching a larger wave group size.

These groups of waves, formed by wind, will move in the same general direction as the winds. As a rule of thumb, the wave groups will spread about 30 degrees in either direction from the mean source wind direction. The wave density, at the beach, is a function of area of waves generated, and the inverse of the square of the distance from the beach. Larger storms generate more dense swells at the beach, and further away storms generate less dense swells at the beach.

The really important things to know about wave groups are that the wave groups travel at roughly half the wind speed, and the waves travel twice as fast as the group.

The entire point of predicting swells from far away is finding winds over the ocean that are blowing at you with an adequate velocity, and then calculating the arrival time based on wind speed.

Finding a significant swell

Let's start a prediction run. I watch the FNMOC models for the Pacific Ocean, NGP, Previous 12-hr Precipitation Rate [mm/12hr] and Sea Level Pressure [hPa], and look for low pressure systems. Low pressure systems usually come with winds, and create swell. Sometimes I find swell generators on QuikSCAT instead. Quikscat is a scatterometry site - they take "roughness" readings over the ocean and use them to estimate the winds. It is important to note their data is based on waves - not winds. But, they present their data interpreted as wind data.

Suppose I find someplace where winds are over 40 knots covering a substantial area, and the WWIII Zero time model shows seas over 25 feet. I need to know a few things. First, are the winds really pointing at us. Second what wave period can I expect. Third, when will it get here. Fourth, how big will it be when it arrives.

Are the winds pointing at me?

Now, with a load of waves over the ocean, I need to know if they are coming my way. For that, you need to understand the big circle. Most radar and maps are mercator plots - every 1 by 1 degree latitude-longitude square looks the same size. That would work if the world were flat. But it is not. So, many years ago, the concept of traveling by the "Great Circle" was developed. The concept basically lets you use coordinates, and uses a globe (or ellipse) model of the earth. With an appropriate "Great Circle" calculator, you plug in latitude and longitude of the starting and stopping points, and it gives you the heading from each point, and distance. Here is my favorite. Actually, my favorite is the program "geod" from the government here. The program is really anal about formatting - the web page is easier to use. In any case, it is useful to see, for example, that a swell on the dateline at 45 degrees N latitude needs a heading of 75 degrees (ie: north east) to come into San Francisco at 305 degrees (or from the northwest). If that seems confusing, think about it this way. If you start in Alaska and head north, after you cross the pole you are heading south. We live on a globe, not a mercator plot of the earth.

Using a Great Circle calculator, I plug in the coordinates for San Francisco, and for the center of the swell mass. Here is a closeup of the storm winds.




San Francisco is at 37d45'N 122d49'W, the swell origin is close to 54d00'S 124d00'W. Use the position in the swell where the wave models show the largest waves are being generated. Do not use the positions where the winds are pointing in exactly the right direction. The winds will generate waves with a splay of about 30 degrees in either direction, so the angle is like horseshoes and hand grenades - close is good enough. Use the strongest position in the swell, it will generate the longest period waves. The calculator tells me something like this

echo 37d45'N 122d49'W 54d00'S 124d00'W | geod +ellps=WGS84 -p -I +units=kmi

OUTPUT: 180d41'56.584" 0d56'22.202" 5489.364

This calculator is the geod calculator provided by the USGS. The output says this swell is coming from 180 degrees. The winds at origin need to be pointing at 1 degree, or due north. And the swell is 5490 nautical miles away. I go back and check the QuikSCAT plot to see if the winds are actually blowing close to 1 degree. THEY ARE!!! Hurray - a swell is coming - but when??? And how big???


What is the wave period?

There are two complementary approaches to determining the wave period. The first is to look at the QuikSCAT data and estimate directly from it. In this example, we would see at the center of the winds pointing at us, that winds are in the 40-50 knot range. This means if the seas fully develop, the swell should be predominantly 14 and 17 second wave periods.

The basic relation between wave period and wind speed is in the following table
Wind Speed (knots) Period Generated
30 11 sec
37 13 sec
42 14 sec
50 17 sec
60 20 sec


The second approach is to look at the WWIII Time Zero model run. This is the current wave heights. Look at 124W 54S

Wave heights in the 24 to 27 foot range will generate predominantly 14 second wave periods, with some 17 second, and some 12 second periods. Wave heights close to 35 feet at origin will have periods of 17 seconds, with some 20 and 14 second energy. At 45 feet, it is mostly 20 second energy, with some 25 and 17 second energy. In the case of our example swell, we estimate mostly 14 second energy, with some 17 and 12 second energy.

When will it get here?


Now I take the 5489 nautical miles and my handy dandy calculator and divide by 26.52, 21.84, and 19 nautical miles/hr. I get
Wave Period (sec) Arrival time (Days)
17 8 days 14 hours
14 10 days 11 hours
12 12 days 0 hours
We should start to see swell in 8.5 days, and see some swell for another 4 days.

Table relating swell period to speed. Please note that these speeds are the transit speeds, or wave group speeds. The individual waves in the group will move at twice the group speed.
Period (in seconds) Speed (knots)
11 17.16
13 20.28
14 21.84
17 26.52
20 31.2


How big will it be?


Now for swell attenuation. The basics are that swells lose 25% of their height for each travel day. In addition, as swells approach shallower water, the circular motion of the water will begin to be interfered with by the ocean bottom, and the swell will lose even more energy. So, we use the basic formula,

Swell height at origin * 0.75^(num days travel) * 0.75
The last factor, the 25% loss in height locally, is specific to San Francisco. Different locations will lose more or less, in relation to how much "shallow" water there is on the way to the beach. But back to our example. Assume 26 ft seas at origin. After 8 days the swell should be down to 2.6 ft 17 seconds. After 10 days, a little under 2 ft 14 seconds. As it turned out, this swell generated waves continuously for several days, and the peak energy came in close to 3 feet 14-17 seconds. Predictions more accurate than this may be obtained, but only if the swell is plotted day-by-day, and the variability in wind attenuation factored in. The 25% loss in wave height each day is due to wind resistance. Don't forget these 17 second period waves are moving through the air at 50 miles per hour! A substantial wind impacting a travelling swell will have a large impact. More than 30 knot cross or headwinds will attenuate a swell 40 to 50%, instead of the normal 25%. A tailwind of 20 knots will allow no swell attenuation at all in a day.

Another point to keep in mind is something I mentioned earlier. The winds at the source need to be consistent. You will rapidly get a hang of this if you plot fetches for each 12 hour period (or each QuikSCAT pass). If the fetch moves towards you, in 12 hours, by about 300 miles, this is perfect. The fetch is moving at close to the same speed as the swell. The swell size will be larger, and longer period, than otherwise expected, because the seas will more fully develop. If the fetch, on the other hand, forms and degrades rapidly, or moves too fast, seas will be smaller than otherwise forecast.

Some other details


As mentioned above, the swell will also lose energy in shallow water. What is considered shallow? Well, the wave groups cause circular motion of the water, to a depth of roughly half the wave period. So, when the water depth gets shallow relative to half the wave spatial period, the swell will start to lose energy. More energy is lose the more shallow ocean floor is crossed.
Table relating swell period to wavelength. Each value is twice the distance travelled at that swell velocity in one swell period multiplied by two (the two factor is from the wave-to-swell velocity difference).
Period (sec) Wavelength (feet)
11 636
13 890
14 1030
17 1520
20 2100

So, for a 20 second swell, even 500 feet of water will cause significant swell attenuation, and refraction. This is terminal for the US East Coast spots, as the continental shelf will scrub the longer period swells down to nothing. In San Francisco, we lose another 25% of swell height due to the relatively brief continental shelf. Longer period swells will lose less attenuation travelling from their source to the near-deep water because they spend less time in transit. But, they will lose more size when they approach shore, as they "feel" the ocean floor deeper than shorter period swells. The swell spots that really pump have water 2 km deep as close as possible to shore.
If you've grasped these concepts, start charting swells and taking notes, and you will be able to predict swell conditions with the best. I am completely serious, this is all there is to it. But take notes, and try to find out why your predictions are off, when they are off.


Refining predictions using buoys

What has been described, so far, is a way to use the source data, wave heights and wind speeds, and use them to predict the swell. Of course, the prediction runs without further input of data, and there can be more than a week of uncertain weather conditions en route to your beach. So, it is important to use buoy data to refine your predictions.

For any direction of swell, look for a buoy it will hit on the way. A 230 degree south swell will come straight to use from the Christmas Island Buoy. A 290 degree NW swell comes straight through the SE Papa buoy. A 270 degree swell comes through the California buoy. A 180-190 degree south swell will hit Southern California buoys about a half day before it hits us. By using the raw spectral data and directional data, when available, from these buoys, you can anchor a prediction in the 12 to 40 hours before it reaches your shore. The Great Circle calculator can tell you the distance from the buoy to you, and its direction, and you can use the data from the tables on this page to tell you how long each period will take to reach you from the buoy. With local buoy refinements, you can estimate swell arrival time within an hour, and height within 10% of actual height, which is much better than any wave forecaster can do with only swell origin data.
Wind prediction

First of all, let me tell you, charting swells is NOTHING compared to trying to predict local winds in SF. I am BAD at it. But, I see nothing else doing much better, so here is what I use.

First, the NWS marine forecast has a wind forecast. If it says "winds less than 10 knots and variable", then we are ON. No winds morning or afternoon, probably high pressure is in place. If it says "Winds NW less than 15 knots", then we'll probably, but not definitely, have null winds in mornings and problems in afternoons. NW winds 15-25 (or more) knots is BAD NEWS. Anything NE or SE or E will blow close to straight offshore. The local climatology focuses things directly onshore or offshore. Anything remotely offshore will blow pretty much right into the wave face.

Okay, NWS forecast gives us SOME idea of winds out to 2 days. But further than that I use the FNMOC model. The model has wave heights and wind vectors on the same page. If the wave heights show a little area of light blue and elevated wave heights right at the coast, that is usually a high pressure driven local wind gradient - windswell, and usually bad winds at the beach. If there are no wind vectors anywhere near SF, that is high pressure, and good winds morning and afternoon.

Also, use pressures to help you out. Higher pressure over the Great Basin (eastern Nevada, western Utah, southwest Idaho) is our preferred offshore wind pattern. Lower pressure over land, and high pressure over the ocean, usually means onshores.
Hope this helps.
blakestah AT blakestah DOT com