| Abstract: |
Recent results have shown that the extent of summertime hypoxia in the Chesapeake Bay has not responded as expected to decreased nutrient loads from the watershed. In this study, we use a combination of spatial and temporal statistical models to explore the hypothesis that changes in the stratification of the Chesapeake Bay are a major factor in the unexpected patterns of Bay hypoxia. Strong stratification has the potential to impact hypoxic conditions by causing decreased water column mixing and thereby reducing the replenishment of oxygen to bottom waters. Through a related team effort called the Chesapeake Bay Environmental Observatory (CBEO), we have compiled data from multiple agencies and researchers collected since 1950. We make full use of this rich data resource to develop, test, and compare statistical models for evaluation of hypoxia and stratification. Specifically, we implemented a kriging technique to calculate the hypoxic volume of the Bay for each summertime data collection cruise from 1949 to 2009 and implemented a calculation of vertical stratification strength that uses density data collected along the main channel of the Chesapeake Bay. Time series analysis of these results revealed some interesting findings: that the observed increase in hypoxia volume in recent decades is actually only occurring in the early summer, and that the early summer stratification strength of the Chesapeake Bay has been increasing. We will present results that demonstrate a significant relationship between the increasing early summer hypoxia and stratification of the Bay and explore the spatial extent of these changes. Finally, we will describe some of our on-going modeling efforts including linear multi-regression, non-linear regression, and data mining techniques designed to evaluate possible climatic factors for their relative impacts on the increasing stratification. Overall, this study demonstrates how comprehensive spatial and temporal statistical modeling of the Chesapeake Bay monitoring data can reveal previously unnoticed trends and suggest relationships that can be tested further with targeted monitoring and modeling. |
