The goal of this project was to efficently calibrate hydrological networkx models utilizing optimization techniques. I used Spotpy for the optimization alogrithms and intergrated into the pywr ecosystem.
- Script to run any climate change scenrios
- Parent script to start calibration
- Bulk of the calibration code with parameter setup, iterating the model, comparing results through a loss function, and tweaking parameters
- Script that runs the model with modified parameters and formats the outputs
- CSVs used in choosing parameters nodes and their value ranges
When I was put onto this project claibration was done by hand and took a long time to get correct. A calibrated model of Merced's model is shown below. With our new calibration approach the model's accuracy improved by 48%
