California Wave Models

In October 2016 the server running CDIP’s original swell model code reached its end of life; both the server and that model have been retired. Now the latest version of CDIP’s spectral refraction model is being used to generate all model products. This model was released in 2010 as a significant advance over the original model. With lower levels of numerical noise, the spectral refraction model is able to accurately predict wave periods and directions in addition to wave heights.

For scientific details on the models, please refer to our model description page.

Since output from the original swell model remains highly popular among users, CDIP has started generating the same products and formats using the latest version of the wave model. There are, however, a number of ways in which the new output differs from the old:

  1. Predictions in all regions incorporate proper time lags for the propagation of swell. E.g. where the old model would show wave heights all across the Southern California Bight instantaneously rising and falling with readings from the Harvest buoy, the current model shows offshore changes propagating across the bight hour-by-hour. (Figure 1)
  2. While the old model used a single buoy to make predictions, the current model combines input from multiple offshore buoys. This generates better predictions but can sometimes result in minor discontinuities on the wave maps, where input from one buoy starts or stops. (Figure 2)
  3. Uncolored areas at the edges of the map represent locations where model predictions are currently unavailable. Whereas the old model would assume a common wave field at all deep water locations and make predictions across the entire map, the current model assesses output sites on a point-by-point basis, producing predictions only where appropriate. A location’s update status is determined by its position relative to the buoy network, its swell exposure, and buoy data availability. (Figure 2)
  4. The ‘Deep Water’ spectrum plot and the ‘Deep Water Swell’ summary text no longer come from a single buoy. Instead they are estimated for the center of the map from multiple buoys using the appropriate time lags.
  5. Model resolution has been standardized across the entire state. In many areas - e.g. across all of Northern California - this results in improved resolution. In a few areas, however, resolution will be lower than with the old swell model.
image1 image2
Fig. 1 Fig. 2
Full image 1 Full image 2
More details on the CDIP MOP System.

About the Southern California Swell Model

Note: While this document specifically describes the Southern California model, most of the information and FAQs apply to all of the nowcast models.


The swell model maps are created by a linear spectral refraction wave model developed by Bill O’Reilly. The model is initialized with input from the deep-water buoys in CDIP’s wave monitoring network.

For more detailed information please see “The California coastal wave monitoring and prediction system” by O’Reilly et al. (Coastal Engineering, Volume 116, October 2016, Pages 118–132).


The model only simulates waves arriving from outside the islands (wave periods of 8 seconds and longer). It doesn’t consider any local generation of seas.

To create the image, deep water wave data are collected by Datawell buoys off the California coast. These data are transferred to the Coastal Data Information Program at Scripps at 30 minute intervals.

The buoy data are processed and combined to produce estimates of the directional wave spectrum at each point on the map. These estimates are time-lagged to properly predict the arrival times of swell at locations across the map.

The buoy data are also used to create an estimate of the deep water directional spectrum for the middle of the model domain (i.e. for the center point of the map). This 2D spectrum is shown at the bottom left of the Southern California image, and is used to calculate the N/S summary parameters given on the map.


Frequently asked questions

  • What is the circle at the bottom left of the swell image? It’s a wave spectrum (swell intensity as a function of period and direction). It shows the relative intensity (using color, red = highest relative energy density, or itensity), direction (on the compass) and period (by distance from the center) of the swell. The colors in the compass plot are not related to the values given on color scale for Hs at the top of the image.
  • Sometimes the red peak in the spectrum does not match the peak period and direction with the largest wave height in the N. Pac/S. Pac swell table on the image. Why is that? The values in the table are calcualted by summing up wave energy over all swell periods for N. Pac directions (240-335 degrees) and S. Pac directions (155-240 degrees). It is not unusual for a swell event to have a large, sharp peak in the spectrum (i.e. its energy is concentrated in a narrow range of wave periods and directions), but have less total energy than another concurrent swell with a broader distribution.
  • What does the Hz value inside the spectrum represent? Cycles per second. So a dot of color on the outer part of the circle indicates a swell at .12 cycles per second or more commonly “every 8 seconds.” If a dot is near the center of the circle, the swell period is higher. So if it’s around the .04 Hz value, that swell component is “every 25 seconds.”
  • Where is the center of this “radar” geographically? The directional spectrum is estimated for the middle of the map, e.g. for Southern California a point between Catalina and San Nicolas Islands. This central point is most representative of the region as a whole.
  • What do Hs, Tp and Dp stand for? Hs=Significant wave height of swell, or roughly the average height of the 1/3rd highest waves (feet); Tp=Peak period of the swell (seconds); Dp=Compass direction from which the waves are arriving (degrees), e.g. 180=from the south, 270= from the west.
  • What is the spatial resolution of the images? The image of the entire Southern California Bight has a resolution of 0.01 degrees or approximately 1000m. The more detailed regional images have a resolution of 0.001 degrees or approximately 100m in water depths less than 60m (that is why these images look like they have smaller pixels close to the coastline).
  • What is the spatial resolution of the bathymetry (ocean bottom topography) grid used by the model? 100 x 100 meters.
  • Why is there sometimes a rather large disparity between the Southern California Bight swell model and the local models (the San Pedro Channel model, the Long Beach model and the San Diego Bay model)? The local models are generally more accurate. They use “local” buoys to model both sea (short period waves) and swell (long period waves). The Southern California Swell model is for swell only (T=8 sec and longer) and uses our offshore buoy at Point Conception. They do look significantly different when a local wind sea is present.
  • Why is the time on the model sometimes two hours old (or more)? Data from the buoys is collected in 30-minute files and marked with the start time, so in general the latest file has a time from one hour ago. And because the swell model products are updated just once an hour, the time on the models can be about two hours old before the next update runs.

About the CDIP Wave Forecast Model

These experimental coastal wave forecasts are a joint research effort by:


The CDIP coastal wave forecast model

The coastal wave height forecasts are actually the combined result of two wave models:
  • The NOAA/NCEP implementation of the Wavewatch III wave model for deep water regions (depth > 300m). The Wavewatch III model is a wind-wave generation and propagation model. That is, the global wave forecasts are made based on global surface wind forecasts. [See the NOAA Wavewatch III model web site].
  • The Coastal Data Information Program’s implementation of a spectral refraction wave model for shallow water (10m < depth < 300m). This is a propagation-only model (no wind-wave generation). It models the effect of bathymetry (underwater topography) on waves as they travel from deep water towards the coast. [See the CDIP swell model web page]. This implementation is slightly different than the swell version. We also include the propagation (but not generation) of shorter period local seas based on input from the Wavewatch III model.

How the forecasts are made.

  • Every 6 hours, CDIP receives detailed wave forecasts from NOAA for its deep water locations along the California coast.
  • NOAA’s WW3 forecasts are used to initialize the CDIP wave propagation model and make predictions of wave heights across the continental shelf to the coast (10m water depth).

A Description of Forecast Model Products.

  • Offshore Wave Height Forecast Plot

    This is the plot displayed at the top of the CDIP Recent-Forecast page. These are forecasted significant wave heights from the NOAA global Wavewatch III model for the two deep water sites off California used in the coastal wave models:
    • Pt. Conception - (34N 121W, used to make coastal predictions in southern and central California).
    • San Francisco - (37N 123W, used to make coastal predictions from Monterey Bay to Pt. Arena).

    The plot is designed to provide a quick look at whether there is a big deep water wave event on the horizon that may impact the coastline. The storm threshold line of approximately 13 ft. is based on historical storms in southern California. Forecasts of offshore waves exceeding this threshold do not necessarily mean damaging coastal waves will occur (for example the waves can come from the northwest and southern California is sheltered by Pt. Conception). It is provided as a guideline for when you may want to look at more detailed plots of coastal forecasts.
  • Wave Height Maps (San Diego to Pt. Arena)

    These are similar to CDIP real-time swell maps, but include short period local seas. Regions along the coastline are clipped from the larger modeling area and rotated so that 3 forecast days can be stacked on a single plot. The wave height scale on these plots is fixed between 0 and 27+ feet. As with the Southern California swell maps, the time lag for waves to propagate from offshore waters to the coast are not accounted for in these images.
  • Coastal Wave Height Plots (Southern California Only)

    These are also similar to CDIP’s real-time swell predictions of alongcoast wave height. They are a plot of the model results along the 10m depth contour. They are not breaking wave heights. Generally, the 10m depth contour is outside the surf zone (area of depth-induced wave breaking). However, when the wave height exceeds roughly 15 ft., 10m is the outer end of the surf zone and the plotted height will be larger than what would actually occur at 10m depth. Nevertheless, it does provide an idea of how much wave energy is reaching the coast relative to other locations. Higher wave energy generally leads to higher water levels and wave runup at the adjacent shore.
  • Coastal Wave + Tide Plots (Southern California Only)

    These plots combine predicted tides and forecasted wave heights in 10m depth (described above) on a site by site basis in southern California. The combined height is defined as the “Potential Flooding Index”. They provide a clear view of when forecasted storm waves are going to be coincident with high tides.


Known Limitations

  • Forecasting Local Seas

    The CDIP coastal wave propagation models do not include wind-wave generation by local winds between the Wavewatch III deep water site and the coast. Unlike the CDIP swell model, we have included the propagation of seas in the coastal predictions using the Wavewatch III forecasts of seas at the deep water sites. For the coastline north of Pt. Conception, the resulting errors should be relatively small during large wave events because the continental shelf is narrow and there isn’t much distance (fetch) between deep water and the coast for additional wave generation.
    In southern California, the story is once again more complicated given the distance between deep water and some coastal areas, and the existence of numerous offshore islands. For the south-facing coastline in Santa Barbara County the coastal forecasts will most likely underpredict local seas. The CDIP wave propagation model assumes almost complete blocking of seas from the south by the Channel Islands and ignores local wind generation in the Santa Barbara Channel.

Frequently asked questions

  • What is meant by ‘storm threshold’ in the offshore forecast plot? The definition of a major storm is arbitrary and is based upon historical precendent. In Seymour et al (1984), a list was shown of hindcast and measured large wave events in Southern California during the first 83 years of the century when significant wave heights exceeded 10 feet for more than 9 hours. Seymour (1996) updated this list, but extended the height threshold to 13 feet to qualify as a major storm (because it appeared that smaller storms had been undercounted in the early part of the century when atmospheric data to support wave hindcasts were sparse.) Because the forecasts used here are based on 12 hour update intervals, a modification to the threshold in Seymour (1996) has been made so that wave heights must exceed 13 feet for 12 hours to be counted as a major event. References: Seymour, R.J., R. R. Strange III, D. R. Cayan, and R. A. Nathan].1984. Influence of El Ninos on California’s wave climate. In: Proc. 19th Int. Conf. on Coastal Engineering, B. L. Edge, ed., ASCE, Houston, Texas, 3-7 September, 1984, Vol 1: 577-592. R. J. Seymour.1996. Wave Climate Variability in Southern California, J. of Waterway, Port, Coastal, and Ocean Engineering, ASCE, July/August 1996, Vol 122(4): pp. 182-186

Experimental 5-Day Wave & Wind Forecasts : SIO and Torrey Pines

These experimental wave forecasts are being made in support of Southern California Beach Processes Study field operations and the Nearshore Canyon EXperiment (NCEX).
The forecasts are made with a local, shallow water, spectral refraction model initialized with deep water forecast spectra from the global NOAA Wavewatch III Wave Model at a location offshore of San Diego (32N 118W). The model forecasts update every 12 hours. The real-time measurements on the plots are updated hourly.
The forecasts include both long period swells and short period, locally generated seas. The Wavewatch III model does not resolve the sheltering effects of islands offshore of Southern California and the unaltered deep water forecasts tend towards overprediction. To crudely compensate for the island blocking of sea and swell, regional spectral refraction calculations at the 32N 118W Wavewatch III forecast point are used to remove energy from the forecast spectra from directions blocked by islands. The modified forecast spectra are then used to initialize the Scripps local refraction model to forecast wave conditions at CDIP buoy measurement sites and at the end of the Scripps Pier.
Wind forecasts from two data sources are also shown when available. 1) The offshore AVN wind forecast from NCEP at the 32N 118W Wavewatch III forecast point (green line), and 2) The US Navy COAMPS wind forecastfor the measurement sites (red line, interpolated from a ~10km COAMPS output grid).
Our experience to date is that the wave model scheme underpredicts the heights of the larger local sea events, which appear as a rapid increase in the wave height concurrent with short peak wave periods (4-8 seconds), and high local winds. This is most likely owing to excessive directional blocking of the Wavewatch-III energy in our modified deep water spectra. In addition, the timing of the initial arrival of southern hemisphere swells are usually off during the summer, appearing as a gradual increase in energy over the course of 24-36 hours, when the measured arrival energy appears later and builds more rapidly. However, the forecasted time of the peak in the height of the swell event is usually pretty close.