Water Level Forecasting Tools

Four primary types of forecasting are presently used to predict water levels along coasts around the world. These forecasting tools are:

• Harmonic Analysis
• Constant Water Level Model
• Finite Element Models
• Persistence Model (Constant Water Level Difference) Forecast

Harmonic Analysis

Harmonic analysis is the most widely applied method of tidal prediction and is the method that NOAA/NOS applies to generate tidal forecasts. Harmonic analysis of water levels is based on the calculation of the gravitational influence of the sun and the moon. These celestial bodies attract all items on earth because of their mass and the resulting gravitational forces. This influence is most significant on oceans as they have a large amount of free moveable water. Because the orbits of the earth around the sun and of the moon around the earth are periodic, it is possible to make accurate longtime water level predictions.

Harmonic Analysis Applied to Neural Network Forecasting

A harmonic forecasting application was developed by DNR to facilitate NN water level forecasts. The harmonic forecast serves as one of the input for the NN model.

For more information on Harmonic Analysis:

• http://www.cbi.tamucc.edu/~pmichaud/pub/harmonic-analysis.pdf
• http://co-ops.nos.noaa.gov/restles1.html
• http://co-ops.nos.noaa.gov/harmonic.html

Predictive and Demonstative Tools:

• DNR Harmonic Forecasts
• http://www.coastal.udel.edu/faculty/rad/tide.html
• http://www.sci.tamucc.edu/~pmichaud/harmonic/HarmApplet.html

Constant Water Level Forecast

A very simple method is to assume that the water level in the future is the same as right now. This method is extremely simple and needs no computation at all, but it ignores the influence of both astronomical and meteorological forces. It may be used for short time predictions (< 3 hours) and also for 25-hour predictions, when the influence of the moon is similar.

Finite Element Models

Finite element models are complicated and need a large amount of calculating power. Furthermore they take a long time to compute their results and hence are difficult to apply for real-time calculations. They divide the given space into a finite number of elements. Then the physical behavior for every piece is calculated. Based on the conditions of one element they compute the behavior of the neighbor elements. So it is vital for this model to know the exact physical process and the exact geometry of the location. Furthermore, they usually have problems with the conditions at the borders of their model. Since finite element models can also take into account meteorological influences, their prediction time is limited to 3 to 4 days. Examples of finite element models used for water level and current modeling are the Princeton Ocean Model (POM) and the parallel Advanced Circulation model for oceanic, coastal and estuarine waters (ADCIRC).

Persistence or Constant Water Level Difference Forecast

This method assumes that the difference between the harmonic analysis and the measured water level will be the same for the forecasted water levels. It is therefore based on both harmonic data and the influence of the wind on the water level up to the forecast time. This method is only applicable to short periods of time (< 48 hours).

The persistence model was applied in this study to provide a simple yet effective method to improve upon harmonic predictions over short time periods.

Return to "Understanding Water Level Observations and Forecasting Methods"

<< BaselineANNApplicationToTexasEstuaries | ForecastsSiteMap | PersistenceModel >>