Repository logo
 

On the estimation of return periods from short sea level records.

Date

1993

Authors

Ranasinghe, Dona Suvineetha Swarna.

Journal Title

Journal ISSN

Volume Title

Publisher

Dalhousie University

Abstract

Description

Techniques for obtaining return periods of extreme levels from short records are investigated focusing on processes which are partly stochastic and partly deterministic. The physical motivation for the problem comes from the need to estimate return periods for short sea level records which contain a deterministic component, in the form of a tide, and a nonstationary stochastic component associated with seasonally changing variance and autocorrelation. Nonstationarity was modeled by a periodic autoregressive state space model which allowed for the seasonality in the autocovariance structure of the surge process. State space representation was able to take the measurement error into account. This model was then used in the estimation of return periods using the exceedance probability method (Middleton and Thompson, 1986) and the revised joint probability method (Tawn, 1992). Extensions were made to the methods to allow for the nonstationarity of the surge process, which was defined in discrete time, and to allow for estimates of sampling variability of the return period estimates.
The 50-100 year return period estimates obtained using just one year of data using the techniques introduced in this thesis are in good agreement with standard estimates based on more than 50 years of data.
Thesis (Ph.D.)--Dalhousie University (Canada), 1993.

Keywords

Physical Oceanography., Statistics.

Citation