Workflow
Note: Analysis was completed using a provided modelling framework and code developed for the course; my work focused on interpreting model outputs and evaluating landscape change scenarios.
Step 1 – Examine Changes in Forest Cover Composition
Forest cover data derived from Landsat imagery were first explored to examine how the proportions of different forest age classes changed between 1972, 1984, and 1991. Calculating the relative proportion of each class provided an overview of landscape composition and revealed trends such as the increasing proportion of recently harvested areas and the decline in older forest classes.
Step 2 – Build Transition Matrices Describing Forest Change
Transition matrices were constructed to quantify how forest cover classes changed between time periods. Tally matrices were first generated to count transitions between classes, and these counts were then converted into annual transition probabilities that describe the likelihood of a pixel changing from one forest age class to another.
Step 3 – Develop and Validate a Markov Model
A first-order Markov model was developed using the transition probabilities and the initial 1972 landscape composition. The model projected changes in forest class proportions over time, and the simulated results were compared with observed data from 1972–1991 to evaluate how well the model reproduced real landscape trends.
Step 4 – Simulate Alternative Forest Management Scenarios
The model was then used to simulate long-term landscape change under different management scenarios by modifying transition probabilities to represent reduced harvesting rates. Comparing the projections illustrated how management decisions could influence the long-term proportion of mature and old-growth forest.
Tools & Technologies
R Studio
Forest Change Modelling
This lab explored landscape change in Pacific Northwest forests using a Markov modelling approach to analyze transitions between forest age classes over time. Using Landsat-derived forest cover data from 1972, 1984, and 1991, transition probabilities between forest classes were calculated and used to project future landscape composition. The model was then used to simulate alternative forest management scenarios and evaluate their potential impacts on long-term old-growth forest availability.