ANN Model Emulators for DSM2

Criterion	Explanation
General Description	DSM2 – ANN (Artificial Neural Network) is an emulator of the DSM2 model to simulate concentrations of conservative constituents in selected fixed locations of the Delta, using flow and concentrations at boundaries.
Model Domain	Sacramento-San Joaquin River Delta. The downstream tidal boundary is specified at Martinez, where salinity is a function of net Delta outflow. Inflow boundary conditions are specified at: Sacramento River, San Joaquin River, Calaveras River, Mokelumne River downstream of Cosumnes River confluence, and Yolo Bypass. Agricultural return flows from Delta islands are also specified as boundary conditions.
Developer	Tetra Tech, Inc., under contract to Metropolitan Water District of Southern California
Hardware computing requirements	Developed for the Windows operating system
Code language	Visual Basic through Excel for inputs and outputs; underlying ANN development in MatLab, called via dlls from Excel Visual Basic code
Original application	Sacramento – San Joaquin River Delta
Public/proprietary and cost	Public, free
Physically or empirically based	Empirically based
Mathematical methods used	ANN model emulation of a physically based model.
Input data requirements	Daily values for major Delta inflows (and associated water quality), Martinez water levels, gates and barrier operations.
Outputs	Volumetric contributions from boundaries and concentrations at fixed locations within Delta.
Pre-processing and post-processing tools	Embedded within Excel.
Representation of uncertainty	None specified.
Prevalence	Recently developed, published peer-reviewed paper, minimal user base.
Ease of use for public entities	Easy to use. Excel user interface is provided for easy input of data required.
Ease of obtaining information and availability of technical support	The model executables and documentation are available.
Source code availability	Available through developer (Tetra Tech).
Status of model development	Latest version was released in 2016. No updates are presently planned.
Challenges in integration	Provides predictions of volumetric contributions from boundaries and concentrations at locations in the Delta. The present interface uses an Excel front end, but a closer integration will require working directly with the ANN dll files, which can be called through many different coding languages.

Reference

Chen, L., S. B. Roy, and P.H. Hutton. 2018. Emulation of a Process-Based Estuarine Hydrodynamic Model, Hydrological Sciences Journal, DOI: 10.1080/02626667.2018.1447112.