Model optimization

This section covers advanced techniques for optimizing model parameters and initial states, in addition to the classical methods explained in the Quickstart section.

The examples in this section use the Lez dataset, which contains daily data from 3 nested gauges spanning a total area of 169km².

• Fully-distributed optimization using a uniform first guess
• Fully-distributed optimization with regularization terms
• Multi-site optimization
• Multi-criteria optimization
• Improving the first guess using Bayesian estimation
• Variational Bayesian calibration
• Pre-regionalization using polynomial mapping
• Pre-regionalization using artificial neural network
• Calibration-validation splitting