Introduction

Stormsurf is proud to present the latest addition to our suite of surf forecast tools: Wave Models and Weather Models. We built these tools with the goal of providing full global coverage of the ocean-atmosphere interface from several perspectives ranging from global to local.

The Weather Models provide animated forecast imagery for the atmosphere while the Wave Models focus on the oceans surface. It is the interaction of the atmosphere on the ocean in the form of wind that generates waves, so the focus of the Weather Models is naturally surface wind. Likewise the result of that wind is seas, sometimes reaching monstrous proportions. So the Wave Models focus on sea height and the effects those seas have on other areas of the oceans surface over time and distance.

The models also depict these natural processes at a deeper level with the entire wind-to-sea-to-wave lifecycle supported. Wind is generated by differences in surface pressure, with air flowing from areas of higher pressure towards regions of lower pressure in an attempt to create equilibrium. As wind moves over the oceans surface it produces seas or swell, which in-turn have energy. That energy is quantified by it's period (which increases as the swells propagate away from their source). As swell's move into shallow waters they begin to break creating surf, and those waves size can be predicted based on the interaction of the swell/sea size and period. More details concerning these processes can be found in the tutorials (see navigation bar above) with a rudimentary understanding of that information helpful if you want to dive into the deep end.

Our models are broken down into 4 geographic categories:

• Global (whole planet),
• Hemispheric (e.g. Pacific, North Pacific, South Atlantic etc),
• Regional (e.g. Northeast Pacific, East Indian Ocean etc)
• Locals (Margaret River, Hawaii, South California, Cocoa Beach etc).

So depending on your need, the data is available at the macro or micro level.

This is still very much a work in progress with more detailed Wave Model imagery already constructed and waiting in the wings. How much actually makes it the public webserver will depend mostly on whether the existing content is used. Each set of images requires computer CPU cycles and bandwidth, which in turn cost money. So we've focused on images that will benefit a wide audience first and are working our way down to more targeted detailed charts last.

This is just the first installment of the Weather Models, with more local imagery in work.

 

General Features

Presentation Improvements
We used government provided model graphics for years, and like most of you, noted some very useful features and some that we wished we could improve. Our graphics therefore represent what we consider the best of both worlds packaged in a way to enable both the beginner and professional forecaster to get to important data fast. Some of those improvements include: 1) Identifying seas up to the 54 ft, 2) Provide seas height resolution to 1 ft and, 3) provide user friendly color scaling that correlates wind speed to seas height to swell period (wave models). Take a look and notice how 35 kt winds produce 17 ft seas which in turn produce a 13 sec period swell. Of course there are many variables in this mix, but the general pattern is used across the board.

Additionally we have the ability to filter out clutter so that only winds 35 kts or greater and seas 17 ft or greater and periods 13 sec or higher can be viewed (50 kt, 17 sec filters are built but not running yet).

Each image was custom proportioned to fit the specific area being depicted. We used fixed latitude and longitude proportioning rather than stretching or shrinking images to fit a fixed footprint. So rather than saying "all images shall be 600 X 450 pixels" and then trying to morph the image for a region into that box, we instead manually proportioned each individual image. This results in some images having some dead-space around the edges and supporting a wide variety of image dimensions, but from a purist standpoint provides a geographically symmetrical chart and a higher degree of image integrity.

Control
One thing we detest is animations where the user has no real control. Without the ability to interrupt playback of an animation to examine the details of one frame in the sequence, the sequence becomes almost useless, the equivalent of a commercial (eye catching images with no useful purpose). So we built our own custom batch programs to wrap all the images in a Macromedia Flash shell, enabling the user to stop the animation anywhere in it's playback sequence, step forward frame at a time to examine some detail, or even reverse the animation one frame at a time. All with just a click or two of your mouse. The Control Panel for each animation is always positioned in the images upper left hand corner. Of course you have to install the Flash Player, but it is a free, simple and secure download from Macromedia's web site and is used extensively throughout the web.

Download Speed
In addition we have tried to balance providing a large image size (again to allow detailed examination of the product) with minimized download time. Without going into the details, we can provide a native resolution 700 X 500 pixel 7.5 day animation in 6 hour increments with a total download size averaging about 1 meg. And any of the filtered images are much smaller. This is still a challenge if you're using dial-up, but with any form a broadband this is instantaneous gratification.

 

Wave Model Features

Which Wave Model?
We use the NOAA Wavewatch III wave model. FNMOC runs a customized version of NOAA's code, but the underlying equations are the same. There are no other reliable global and regional models. So the overriding differentiator is the source of wind data used to drive the wave model (that's right, wave models use atmospheric models as their input source). NOAA uses the GDAS wind model where FNMOC uses the NOGAPS. NOAA also runs their model 4 time daily versus 2. It was a personal preference, but we chose to go with the publicly funded and maintained solution rather than a military funded source and liked the ability to get updates more frequently.

Resolutions
The NOAA Wavemodel suite includes one global model with a 1 X 1 degree resolution (one data point every 60 X 60 nmiles on the equator) plus multiple regional models with a 0.25 X 0.25 degree resolution (15 X 15 nmiles). We download all of them and use the highest resolution model available to fit whatever region we're depicting. The output from NOAA's supercomputer is not the series of pretty images you see, but is instead is a thing called a grib file. It is basically a binary database containing all the information for every data point on the planets surface (or within the predefined region). Those gribs are then downloaded by local providers like Stormsurf who ingest them into tools to create the graphical images we typically associate with 'models'. The inference is that though many sites claim to have their own custom wave models, in most cases they are only producing a custom graphic from NOAAs freely available gribs. We are doing nothing different here. But what we are doing is presenting the data in a format that is truly useful for both the novice and the professional surf forecaster rather than just making a pretty image with no real forecast utility (a common practice in an attempt to 'dumb down' the data for the masses).

Missing Data
The wave models in their native form have large gaps of missing data near the coastlines. There is often literally no data presented within 3-10 nmiles of many coastlines. This is a result of using square grid boxes and then superimposing them on a coast that is anything but a straight line. Here's 2 examples of the problem:

The black areas indicate where no data is presented in the source grib. This typically occurs at the land-sea interface.

So a certain amount of estimation and projection is used to fill in the missing data, especially on the Local Wave Models. To the greatest extent possible we have used every trick and technique available to make the images as accurate as possible. Still, the combination of having to use the global (low res) grib to start with then zooming way-in to view so some remote island chain can result in some almost comical results in some instances. In other cases tiny islands are not depicted at all, and the swell shadowing caused by them is not accounted for. In the end we decided to publish the results even though in a few instances they were a little bit shaky because in those remote corners of the world "some data is better than nothing". If for your particular local location you see an area that consistently depicts smaller than normal seas, and no known bathymetric feature is present there to cause a shadow, presume it's a case of missing data.

Surf Height
Contrary to popular belief, sea heights do not equate to surf height. You cannot look at the sea height for a point near your beach and determine what the surf height will be. That is because it's the interaction of swell and period that determine surf height. Since swell height is not a standalone variable in the NOAA WW3 grib, we have used some trickery to estimate surf height. It has tested out well for most situations. The great part about these charts is you can actually watch the decay process occur as seas radiate out from a fetch area. But since period is increasing with distance, the total surf height size does not decay nearly as fast as the seas themselves. A variety of surf height perspectives (global, hemispheric, regional, local) are provided for a reason. Since it is physically impractical to provide local charts for every break on the planet, you can use the other non-local perspectives to estimate surf height with a high degree of accuracy for you location if it's not already provided. For instances where swells travel long distance (e.g. from the southern hemi) and then interact with larger locally generated windwaves (like off California in the summer) there is an issue. The WW3 has to make a determination which singular wave event to display. Sometimes it picks the windswell, sometimes the long distance swell, and sometimes is is a patchwork of both swells passing through each other. The issue becomes especially pronounced on the local wave models. The general rule of thumb when using the local models is to pay close attention to the swell direction arrows. They will tell you clearly which of the two swells is generating the surf height depicted on the charts. For small long-distance swells where much larger local windswell is present, the long-distance swell will not display. in those instances we recommend going to the Buoy Forecast section. These are much more sensitive and should clearly indicate the secondary surf height.

Local Bathymetry
The effects of nearshore bathymetry (within 5 nmiles of the coast) are not included in the local wave models. It would be nice, but on a limited budget we have had to make due with what we could realistically accomplish. At a macro level the Wavewatch III does a reasonable job of depicting major oceanic bathymetric features. Swell shadows on some of the South Pacific Islands are supported like Tahiti as is the Galapagos Island. As you look through various regions keep an eye out for shadowing, and often you be surprised at what you'll find. Closer to your break you can assume the surf height depicted on the local charts is the "average" surf height. Clearly there are breaks that amplify wave size for long period swells by some factor. Using your experience you can simply multiply the surf height depicted on the chart by that factor to get an accurate surf height estimate (normal ranges are 1.0-2.0 times). Conversely some breaks will always be less than the surf height depicted. The same technique can be used here (0.0 - 0.99 multiplication factor).

 

Wave Model Highlights

Hemispheric
Global (Lo-Fidelity Mercator): These charts provided a quick globe-at-a-glance view to what is forecast over the next 7.5 days. Sea heights are in 2 ft increments but no directional arrows or wind barbs etc are provided. Contouring (the thin black lines that separate the colors) are turned off when sea heights reach 38 ft and are printed only in 4 ft intervals between 24-38 ft. Otherwise the black line would be so densely packed you couldn't see the actual colors. No longitude/latitude grid provided. These are not serious forecast charts, just a way to get a quick overview of the global forecast.

Rotating Globe (Orthographic): Again, this is not a serious chart, but it is a cool depiction and so we shrunk one and use it in the header of every page. The time period resets itself to 00 hrs as the globe rotates over each ocean (Atlantic. Pacific, Indian).

Orthographic: These are beautiful images and give one a much better sense of how waves travel across the oceans following great circle paths without the distortion associated using a standard flat Mercator projection. No arrows provided on the Sea Height and Max Swell Period charts because they just clutter up the image. Using the animation you can view the travel direction. Arrows are included on the Wind charts since it is wind that determines a swells heading.

Hi-Fidelity Hemisphere (Mercator): These are the meat and potatoes of wave models and made to be used for detailed professional level surf forecasting. 2 ft sea and surf height intervals with arrows included throughout. Note that the arrows on many charts are color and size scaled to sit nicely in the image without overpowering it. Contours on the period charts start at 10 secs to minimize buildup of thick black lines on the leading edge of swell fronts. Note that if seas and surf exceed the limits of the color scale, contours continue in 2 ft intervals so one can manually determine actual values. Latitude/longitude grids included.

180 hr Hindcast: For some regions the 00hr image from each run of the model is archived back 180 hours and then Flashed into an animation. This provides a retrospective view 1 week into the past. This is a great tool for figuring out where that mysto swell came from. Not all regions supported at this time nor are they archived (again, those pesky dollar limitations).

 

Regional
Some of these are built from the much higher resolution regional wave model gribs supplied by NOAA (see 'Resolutions' above). It is possible that the global model will forecast sea heights at say 20 ft, while the regional models will indicate 24 ft, because the regional model can sense higher wind speeds over smaller areas than the global model. That is, the regional models are more sensitive and therefore always a better source than the global model. Note that not all the Regional Models on our menu page use an underlying NOAA Regional model. Some are just zooms of the global model. Real Regional Model are denoted with a '+'. All the same feature applied in the Hemispherics are applicable here.

Filters: For the Gulf of Mexico and the Eastern US Coast we added a 9 sec period filter so that all seas 10 ft or higher, periods 9 secs or higher and winds 20 kt or higher could be identified.

 

Local
Now you have the ability to zoom into a rather small area to see what the surf will be doing 7.5 days from now.

Intervals: Surf height is depicted in 1 ft intervals. Colors are separated by white contours in 1 ft intervals up to 10 ft, then in 2 ft intervals up to 42 ft while the contour remains at 1 ft intervals beyond 10 ft.

Overlay: Wind barbs and swell direction arrows are overlaid on all images. In this way you can determine surf height, swell direction and wind speed and direction from one chart. Pretty cool. Of course you have to know how to read a wind barb (look in the tutorials). Wind barbs are drawn at the highest density possible while still making the charts readable. Swell Direction arrows are less critical and are often filtered to a lower density.

Gribs and Grids: A regional grib is used whenever possible to supply the user with the highest resolution graphic possible. Latitude/longitude grid scales are included mainly to pinpoint geographic features that may not display on the rather low resolution maps supplied with our graphics package (like certain smaller South Pacific Islands).

Buoys and Cities: Buoy and major cities are overlaid to provide perspective. Buoys positions are accurate to within 100 meters.

 

Weather Model Features

Which Weather Model?
There are numerous weather models currently available, all with specific coverage areas and specialized applications. From a global perspective the GFS (Global Forecast System - formally the AVN) is the model of choice and couples best with the Wavewatch III (WW3) wave model. That is, the weather systems depicted by the GFS model accurately manifest themselves over time on the WW3 (suspect that is because GDAS winds are used to feed both the GFS and the WW3) . If you spend any time jumping from atmospheric to wave models, you'll notice that often over the long haul the weather systems depicted in the weather models don't show up on the wave models (or visa versa). So the GFS-WW3 pairing is a natural.

Resolution
The NOAA GFS model is a single global model with a 1 X 1 degree resolution (one data point every 60 X 60 nmiles on the equator), same as the WW3. Again this all is fine when building marco level charts, but as you zoom-in to determine local wind speed and direction for one tiny coastal location, things start to fall apart or the fidelity is not high enough to produce a meaningful forecast. This holds true especially concerning tropical systems. Hence the need for higher resolution (smaller grib squares). Fortunately the GFS now is available in a high resolution flavor with 0.5 X 0.5 degree resolution (30 X 30 nmiles). So we're processing the 1 degree GFS for the Global, Hemispheric and Regional views and we're using the Hi Resolution version for some more localized views.

Levels/Variables
Our models have been built to examine the air-ocean interface. There are hundreds of variables we could display (rain, temperature, snow cover etc..) but from a wave generation and wave riding perspective, there is only one thing that really matters: Wind Speed and Direction. Secondary to that is surface pressure (since it is the stuff that generates the wind). And just for fun we threw in 250 mb winds and height (jetstream stuff) since that has a major influence on how storms (and therefore winds) develop. Of course there's always the money factor too, so we tried to provide the important stuff while containing cost.

 

Weather Model Highlights

Hemispheric
Global (Lo-Fidelity Mercator): These charts provided a quick globe-at-a-glance view to what is forecast over the next 7.5 days at the surface and aloft. Isobars are in 4 millibar increments but not tagged with actual values. Red isobars represent low pressure (i.e. 1010 mbs or less) while blue isobars are high pressure (greater than 1010 mbs). Wind speeds are in 5 kt increments starting at 15 kts (no swell generation potential below 15 kts) Contours are only supplied from 20-45 kts, otherwise the colors would get lost in black lines. No arrows or barbs to define wind direction, though it can easily be inferred. Again these are good overview charts, but not for detailed forecasting use.

Hi-Fidelity Hemisphere (North Pacific, South Atlantic etc): These are a good starting place to get some detail on the macro level scenario. Isobars are in 4 millibar increments with tagged values. Contours are supplied for all wind speeds. Arrows are used to indicate wind direction. All other attributes are the same.

Zoomed Hemispherics (Northeast Pacific, Tropics, etc): Now we're getting into the good stuff. These images are not just cropped versions of the above images, but are custom built from the ground floor up. Isobars now in 2 millibar increments with tagged values. Wind barbs replace arrows and are placed more dense to identify local wind conditions at locations where swell is hitting. These charts therefore due double-duty, Examining the internals of a particular fetch and to forecast local winds. Used in conjunction with the great circle charts (see navigation bar above) one can determine if the fetch is blowing up a great circle path to your beach.

180 hr Hindcast: Like some of the Wave Models above, the 00hr image for select regions back 180 hours are Flashed into animation.

 

Regional
This is just the first installment of a select few regional locations using the hi-resolution GFS model. These are great because a level of detail about wind fields are visible that can not be found using the normal GFS. This is especially true for tropical systems. As such the wind scale is expanded up to 135 kts (155 mph - Cat 5 hurricane) and was most instructive tracking Hurricane Katrina - 8/05). Otherwise they are the same as the Zoomed Hemispherics except with higher wind barb density. These are our main source for forecasting local wind conditions over relatively small areas.

We're in the process of building Local Weather models that cover the same area's as the Local Wave models. More to come...

 

General Comments

Time Zones and Processing: The time on the models is in Universal time (UTC), or the time at the Greenwich Meridian (England) also know as GMT, or simply 'Z' for short. It is the international standard used by all meteorological organizations. The models assimilate all the data that defines the current state of the atmosphere at the time they are run. For example let's say were are working on the 12z run of the models. Current readings from satellite, weather stations and a host of other sources are compiled at or near 12Z. That data is then fed into the super computers at NOAA and the models start crunching that data. It takes a few hours to calculate the 'forecast' conditions for the atmosphere out 7 days using those current readings as the starting place. Then the output from the weather models are run into the wave model computer simulation which takes another couple hours to compute the current and forecast state of the ocean. Then all that output is made available to Stormsurf, who then spends another couple of hours generating the user friendly images we all view on this site. The short of it is that the actual images aren't available for viewing until about 8 hours after 12Z. So the 00hr image (and all the rest of the forecast images) from the 12z run of the model actually become available for viewing by about 20Z.

On top of that one needs to consider local times zones. If for example you lived in Sydney, Australia, they are 10 hours ahead of GMT/UTC or Z. So 0Z Monday GMT is really 10 AM Monday in Sydney Australia and 12Z Monday would be 10 PM Monday in Sydney. So for the time stated on the charts, just add 10 hours to calibrate for that area. From the perspective of one living in Sydney, the picture of the environment at 12Z Sunday (which is 10 PM Sunday Sydney time) is normally available for viewing at about 20Z which would be is 6 AM Monday. Everyone experiences the same 8 hours delay. The models are not real-time. Here a link to help you figure out the appropriate time zone conversions: (http://www.worldtimezone.com/).