A Spread Forecast Inherits Every Error in Its Fuel Map
A landscape spread run needs a fuel type for every cell of the terrain, and it gets one from a mapped fuel layer built by classifying satellite imagery and field plots into a set of standard fuel models. Those layers are produced at tens of metres per cell and refreshed on multi-year cycles, so they can be blind to last season's fire, to a thinning project, to grazing, to insect kill, and to the brush that grew back after all of them. The model is then perfectly precise about a landscape that no longer exists. What makes this checkable rather than a slogan: swap one cell's fuel model for a neighbouring plausible one and re-run. The predicted rate of spread in that cell can change several-fold, which tells you the forecast is not reporting terrain, it is reporting a classification of terrain. The rule is two questions before trusting any landscape run — when was this fuel layer built, and what has happened on the ground since. The confusion resolved is that map resolution is not map accuracy; a finer grid of stale classes is still stale. After this Concept you can judge whether a spread run's fuel input deserves the confidence its output implies.
This Concept is waiting for its first lesson!
A landscape spread run needs a fuel type for every cell of the terrain, and it gets one from a mapped fuel layer built by classifying satellite imagery and field plots into a set of standard fuel models. Those layers are produced at tens of metres per cell and refreshed on multi-year cycles, so they can be blind to last season's fire, to a thinning project, to grazing, to insect kill, and to the brush that grew back after all of them. The model is then perfectly precise about a landscape that no longer exists. What makes this checkable rather than a slogan: swap one cell's fuel model for a neighbouring plausible one and re-run. The predicted rate of spread in that cell can change several-fold, which tells you the forecast is not reporting terrain, it is reporting a classification of terrain. The rule is two questions before trusting any landscape run — when was this fuel layer built, and what has happened on the ground since. The confusion resolved is that map resolution is not map accuracy; a finer grid of stale classes is still stale. After this Concept you can judge whether a spread run's fuel input deserves the confidence its output implies.
Are you a teacher? Sign in to start contributing.
Sign In