Date of Award
2020
Degree Type
Thesis
Degree Name
Master of Science in Statistics
Department
Computer Science and Statistics
First Advisor
Jing Wu
Abstract
The marine ecosystems cannot survive without dissolved oxygen (DO). Low oxygen events (hypoxia) in the ocean cause stress on the benthic community and, hampers their growth rate initiating mortality. To monitor the concentration of oxygen, different water quality monitoring sites have been established across the globe. The Narragansett Bay fixed-site water quality monitoring network (NBFWQMN) is a facility that regularly measures oxygen level as well as other important water parameters (temperature, salinity, pH level, and Chlorophyll) at different locations of Narragansett Bay (NB). Missing observation is a common phenomenon for this times-series dataset and, can occur for various reasons. In this study, we analyzed time-series data of dissolved oxygen (DO) after taking into account the missing data. Variability of DO across any water-body depends on diffusion from the atmosphere, respiration of organic matter in the water column and in the sediment and advection of saltwater. The oxygen concentration in water also depends on instantaneous temperature, salinity, and freshwater inputs from nearby rivers. In this study, we used time-series data of temperature, salinity, and river discharge as covariates for DO time-series. In addition to the response variable, some of the covariates also have missing data. In this thesis, we applied dynamic linear model to handle the time-series data with ignorable missing response and covariates.
Recommended Citation
Alam, Shajratul, "ANALYSIS OF MISSING DATA IN MARINE DISSOLVED OXYGEN TIME SERIES USING DYNAMIC LINEAR MODELS" (2020). Open Access Master's Theses. Paper 1903.
https://digitalcommons.uri.edu/theses/1903
Terms of Use
All rights reserved under copyright.