A deep learning model and associated data to support understanding and simulation of salinity dynamics in Delaware Bay
Dates
Publication Date
2023-09-08
Start Date
2000-01-01
End Date
2019-12-31
Citation
Gorski, G.A., Cook, S.E., Snyder, A.M., Appling, A.P., Thompson, T.P., Smith, J.D., Warner, J.C., and Topp, S.N., 2023, A deep learning model and associated data to support understanding and simulation of salinity dynamics in Delaware Bay: U.S. Geological Survey data release, https://doi.org/10.5066/P9IK5Y45.
Summary
Salinity dynamics in the Delaware Bay estuary are a critical water quality concern as elevated salinity can damage infrastructure and threaten drinking water supplies. Current state-of-the-art modeling approaches use hydrodynamic models, which can produce accurate results but are limited by significant computational costs. We developed a machine learning (ML) model to predict the 250 mg/L Cl- isochlor, also known as the salt front, using daily river discharge, meteorological drivers, and tidal water level data. We use the ML model to predict the location of the salt front, measured in river miles (RM) along the Delaware River, during the period 2001-2020, and we compare the ML model results to results from the hydrodynamic Coupled [...]
Summary
Salinity dynamics in the Delaware Bay estuary are a critical water quality concern as elevated salinity can damage infrastructure and threaten drinking water supplies. Current state-of-the-art modeling approaches use hydrodynamic models, which can produce accurate results but are limited by significant computational costs. We developed a machine learning (ML) model to predict the 250 mg/L Cl- isochlor, also known as the salt front, using daily river discharge, meteorological drivers, and tidal water level data. We use the ML model to predict the location of the salt front, measured in river miles (RM) along the Delaware River, during the period 2001-2020, and we compare the ML model results to results from the hydrodynamic Coupled Ocean Atmospheric Wave Sediment Transport (COAWST) model. The ML model shows RMSE = 2.52 RM during the five-year holdout period, superior to three overlapping years of COAWST model predictions, RMSE = 5.36 RM, however the ML model struggles to predict extreme events. Further, we use functional performance and expected gradients, tools from information theory and explainable artificial intelligence, to show that the ML model learns physically realistic relationships between the salt front location and drivers (particularly discharge and tidal water level). These results demonstrate how an ML modeling approach can provide predictive and functional accuracy at a significantly reduced computational cost compared to process-based models. Additionally, these results provide support for using ML models for applications in operational forecasting, scenario testing, management decision making, hindcasting, and resulting opportunities to understand past behavior and develop hypotheses. In this model archive, we provide the scripts and configurations to fetch data for the machine learning model, to process the data for the machine learning model, to run the machine learning model and to analyze the functional performance of the machine learning model.
Click on title to download individual files attached to this item.
fgdc_metadata.xml Original FGDC Metadata
View
48.76 KB
application/fgdc+xml
model_archive.zip
461.69 MB
application/zip
Purpose
This release provides code and data for running a machine learning model to predict estuary salinity and using information theory, explainable AI, and comparison to other modeling approaches for analysis.
