Kalman filtering and application to storm surges
MetadataShow full item record
Over the last decades with the advancement of computational power and access to data, the ability to create advanced forecasts and hind-casts too has grown. Recently, with an increase of the global population moving closer towards coastal areas. There is a much stronger presence to understand severe weather systems and there impact on the local population as well as the economy. With that there is still much work to be done within the field of weather forecasting specifically in tandem with real time decision making.This project will consider forecasting in the event of extreme weather systems. Precisely we will focus on the role of storm surge and investigating novel techniques in trying to increase accuracy in sea-level prediction models and decreases error associated with tidal gauge systems along the coast of Texas. Under the thought experiment that this will be used for some decision making process. Along with including an optimal warning time derived from the prediction methods.Tidal prediction methods have either been of two differing methods, statistical or deterministic. For typical usage most tidal predictions are given by deterministic methods, i.e. used by solving hydrodynamic equations in tandem with their astronomical constituents. Statistical methods have since been developed with the added benefit that we can include live measurements to improve tidal predictions based off a time series of observations.For high-impact weather systems we do not have the ability to solve the same hydrodynamic equations so quickly and readily as to aid emergency services and workers. Accurate models are desirable not only from a human life standpoint but from an economic standpoint as well. It may be the case that community is not endanger but it may affect local businesses that rely on precise forecasts. In these cases we must insist on the exact time closures must be necessary in order to minimize any economic impact or loss.In these cases we choose to employ a combination of both statistical and deterministic methods. The Kalman filter is one such method of combining these two methods in order to increase the accuracy in model.
This thesis meets the standards for scope and quality of Texas A&M University-Corpus Christi and is hereby approved.