DNR/TCOON as well other observation networks provide an increasing amount of real-time data. As part of its research, DNR takes advantage of this information to develop and implement real-time predictive models. We have focused on using machine learning techniques such as artificial neural networks. These techniques are well suited for the modeling of non-linear processes when large data sets are available and also provide predictions virtually instantly once the models are trained.
We typically focus on predictive models to help decision makers and anyone looking for coastal forecasts. The models are developed in collaboration with faculty members of the TAMUCC College of Science and Technology, the TAMUCC Center for Water Supply Studies (CWSS) and the Corpus Christi Weather Forecasting Office. If you have suggestions about existing or new models e-mail us or pass by our offices, we thoroughly enjoy the research and the related discussions.
A set of models were implemented/developed to predict water levels as an aid to navigation in the coastal waters of the Gulf of Mexico and as a tool for emergency management. In Texas tidal predictions do not meet National Ocean Service acceptance criteria as other factors than tidal forcings often dominate water level changes (for example wind stress related forcings and wave climate). The graphs show comparisons of tidal predictions, persistence model predictions (simple but working much better than tides in Texas) and Artificial Neural Network models (perform best overall). Thank you to the sponsors who have made this model possible, including the Texas General Land Office, the NOAA Seangrant and Seagrant Technology programs, and the Texas Research Development Fund. Click to access the real-time water level forecasts.
A single-page view containing measured and predicted environmental data useful in marine navigation.
We are always working on interesting predictive models to help decision making. Other models we have worked on include the prediction of salinity, recreational water quality (enterococci colony forming units) and spring flows in a karst aquifer. If you would like more information or have suggestions for improving our present models let us know.
For more information on DNR forecasting projects, E-mail Philippe Tissot : firstname.lastname@example.org
The computational methodology follows exactly NOS procedures and the program/implementation is described in the following paper: http://lighthouse.tamucc.edu/dnrpub/2002/papers/mostella2002a.doc
Interested in learning more about our forecasting models and methodology? Read it all here.