Propagating probability distributions of stand variables using sequential Monte Carlo methods

Abstract

A general probabilistic approach to stand yield estimation is developed based on sequential Monte Carlo filters, also known as particle filters. The essential steps in the development of the sampling importance resampling (SIR) particle filter are presented. The SIR filter is then applied to simulated and observed data showing how the 'predictor - corrector' scheme employed leads to a general probabilistic mechanism for updating growth model predictions with new observations. The method is applicable to decision making under uncertainty, where uncertainty is found in both model predictions and inventory observations.

Citation

Gove, Jeffrey H. 2009. Propagating probability distributions of stand variables using sequential Monte Carlo methods. Forestry. 82(4): 403-418.