Science Inventory

Modeling stratification and hypoxia in Long Island Sound using generalized additive models

Citation:

Mudahy, A. AND J. Hagy. Modeling stratification and hypoxia in Long Island Sound using generalized additive models. Long Island Sound Research Conference 2024, Port Jefferson, NY, May 15, 2024.

Impact/Purpose:

This presentation will share information on progress and modeling results related to predicting stratification and hypoxia with the Long Island Sound Research and management community.  By sharing our progress will be gauge interest and gain useful feedback on our research. 

Description:

Seasonal hypoxia, or low dissolved oxygen, in Long Island Sound (LIS) has declined since the early 2000s due to management actions that reduced point source nitrogen loading. To further quantify and better understand drivers of changes in the spatial and temporal distribution dissolved oxygen we applied a spatio-temporal modeling approach using generalized additive models (GAMs). Data included 30 years of monthly monitoring data from the Connecticut Department of Energy and Environmental Protection and the Interstate Environmental Commission, estimates of freshwater inflow and nitrogen loading rates from 17 major rivers compiled by USGS, and data on point source nitrogen discharges to rivers below gauge points or directly to LIS. We calculated Simpson¿s stratification index and a modified index that quantifies thermal vs. salinity stratification. Using GAMs, we modeled density stratification in 2-dimensions over time and DO in 3 dimensions over time. We were able to predict 64.4% of the the seasonal and spatial pattern of stratification. Our model showed a 0.113 J/m2/year increase in summer thermal stratification and 0.021 J/m2/year increase in summer salinity stratification from 1991 to 2021. GAMs explained 90.5% of the variance in seasonal and spatial DO distributions. The model estimated that DO increased by 0.2 mg/L in central LIS, and 0.3 mg/L in eastern LIS, and decreased 0.04 mg/L in western LIS. Despite their usefulness in predicting overall DO levels, GAMs did not effectively quantify the frequency of low DO values associated with hypoxia (<3 mg/L).

Record Details:

Record Type:DOCUMENT( PRESENTATION/ SLIDE)
Product Published Date:05/15/2024
Record Last Revised:06/04/2024
OMB Category:Other
Record ID: 361654