teixeirak
commented
5 years ago Thanks! That's a good first step. Next step will be to identify cases in which the original ranges are not appropriate. There will be two major cases:
1. Where there is overlap in the size range between two expert-selected equations. For example, Acer rubrum at Lilly Dicky as two equations, one for 1-55cm and one for .25-5.73 cm. (This is a bit of a weird example because the 1-55cm equation covers the full lower end of trees that we measure, but I'll stick with it as an example.) I'd look at several factors:
- check the quality of the equations. If one has low quality (e.g., low sample size, no indication of bias correction) and isn't truly necessary, it probably makes sense just to drop it.
- check whether they cross. It's probably best to write a script for this. If they cross, the crossing point should be the transition from using one equation to the other. If they don't cross and both are needed, I'm not sure what to do.
2. where the equation doesn't cover the full size range observed in ForestGEO plots We need complete coverage, but also to be very careful extrapolating beyond the range for which the allometry was developed. An example would be Amelanchier arborea at Lilly Dicky. The equations only go down to 5cm DBH, but we need down to 1cm. The logical thing to do here would probably be to fill in a generic allometry (in this case and probably many others, perhaps a shrub equation-- see issue #88). Once again, we need to consider where the equations cross in order to decide where to switch between generic and taxa-specific.
At the upper end, there may be individuals larger than those for which the allometries were developed (e.g., a >55cm Acer rubrum at Lilly Dicky). In this case, we may need to switch to the generic equation, but again we hit the issue of equations crossing.
I think that in both cases, we'd be okay extrapolating the expert-selected equations somewhat beyond the observed range, provided that they don't shoot off into unreasonable values.
I'd like to discuss how to handle all this in person (also with @ValentineHerr and @cpiponiot), and come up with a consistent set of rules for determining which equation to apply. In the meantime, let's figure out how many of each of these cases we have.
gonzalezeb
@teixeirak We need to define how systematically we will come up with recommendations for a min or max input (ex. dbh) in the sitespecies table. I am currently transferring the original ranges as a guide to define our suggestions.
This is the definition for these variables: