Comment by baxtr
20 hours ago
Interesting. Have you tried back testing your model agains fire data to analyze how well you could predict fire likelihood?
20 hours ago
Interesting. Have you tried back testing your model agains fire data to analyze how well you could predict fire likelihood?
No, at the moment classifying disturbances is only done retrospective. Predicting fire likelihood is its own large area of research.
An open case is more in terms of temperate forests which are increasingly starting to burn, for example in Austria or other central European countries. These tend to be understory fires, which (for now) only rarely catch the crown.
For these fires the tree crown stays green and you can't see any change until later in the year or the next year, where the trees die due to damage the fire did to their cambium and roots. Correctly attributing this later decline to fire damage is a bit challenging.