Advanced forecasting techniques

The advanced forecasting section introduces techniques for improving forecasting, quantifying, and reducing uncertainty. First, likelihood methods are used to improve the calibration of ecological models. Better calibrations reduce the uncertainty associated with the capacity of the model to reproduce data (process uncertainty). Second, Bayesian methods are used to calibrate parameters and estimate parameter uncertainty in ecological models. Finally, data assimilation with the particle filter is used to estimate initial conditions and reduce initial condition uncertainty.

This section focuses on the use of non-linear ecological models, rather than statistic or machine learning models like in the first section. The concepts in this section are applied to a process-based ecological model in the third section of the book.