Hidden in the Spectrum

Geographic Transfer of an Offshore Wave Index for Coastal Water-Level Oscillations

Manuscript under review

First page of the Hidden in the Spectrum manuscript

Bulk wave measurements leave part of the story behind.

Wave height and peak period summarize an offshore sea state in two familiar numbers. They do not preserve how energy is distributed across the full spectrum.

The study tests whether a low-frequency-weighted spectral index carries information that transfers more effectively from Pacific Northwest development sites to a separate California evaluation network.

49,786observations used to fit the models across five Pacific Northwest coastal pairs

24,878external observations across six California coastal pairs

11.7%reduction in normalized RMSE against the conventional height-period comparator for the preliminary one-minute target

The index gives more weight to lower-frequency energy.

Two sea states can have similar wave height and peak period while distributing energy very differently. Negative spectral moments preserve more of the lower-frequency structure that bulk measurements omit.

X_s= √m_-4

The resulting value is used as a compact offshore descriptor. It becomes one environmental input within ShoreSense, alongside visual observations, tides, weather, local geometry, and other site data.

A model developed in one region was tested in another.

Five Pacific Northwest buoy-gauge pairs supplied the development observations. The fixed relationships were then evaluated at six California pairs spanning Arena Cove to San Diego.

California outcomes did not determine the spectral moment or model coefficients. The evaluation asked whether the ordering would persist across new coastal settings.

Manuscript page showing the Pacific Northwest development sites and six California evaluation sites

The spectral model improved the primary comparison at all six California sites.

The strongest pooled result used preliminary one-minute water levels. The verified six-minute product preserved the pooled ordering with a smaller improvement.

Preliminary 1-minuteVerified 6-minute

Arena Cove

15.6%

6.6%

Point Reyes

17.4%

7.8%

Port San Luis

15.8%

Santa Barbara

7.2%

5.2%

Santa Monica

13.8%

8.3%

San Diego

4.5%

0.3%

11.7%

lower normalized RMSE

for the preliminary one-minute target against the prespecified height-period comparator across the 24,878-observation California evaluation.

From coastal research to environmental intelligence.

The manuscript contributes a geographically tested environmental covariate. ShoreSense places that kind of spectral information inside a larger system that also considers camera observations, local conditions, human movement, and response coverage.

Read the manuscript

See how the research becomes a system.

Explore the sensing hardware, edge models, littoral digital twin, and response-coordination architecture.

Explore the system