calibration_python

This is for python scripts associated with the process of automating calibration of SnowModel

This is separate from the acquisition of SNOTEL data (or other) for the purposes of calibration. Those scripts reside in the preprocess repositories

01_Calibrate_SnowModel.ipynb

This Notebook runs allows a user to select calibration parameters. SnowModel will be run in line mode for all parameter combinations at each SNOTEL location. Objective functions (RMSE, R^2, NSE, KGE, MBE) are computed between modeled and observed SWE.

Requires: par_base.json in directory with notebook

Files saved out:

• .csv of all parameter combinations
• .nc of calibration metrics for each station for each calibration run

02_Calibration_postprocess.ipynb

This notebook evaluates the output of the calibration runs based on:

1. OFs averages over the domain

• OFs are averaged across the stations for each calibration run
• The highest score for each of is used to identify the top model parameter combinations

2. OF rankings

• each OF is ranked independently
• the rank scores are summed across each model run
• The summed scores are then ranked and the top scores are used to identify the top model parameter combinations

Using both of the above method, SnowModel is run again in line mode and a top parameter combination is identified.

A final full-domain calibration run is initiated using the top parameter combination.