FIA data doesn't provide crown area or diameter as a value. Yet, there are measures like the Crown Ratio, that measure the ratio of a stem covered by the crown and other measures of crown cover. We may need to find a better proxy or other datasets to create a meaningful estimate of crown size for a large set of FIA tree points

DeepForest is impressively good approach, but may not be enough in reliably use crown estimates at the individual level. For example, extracting boxes for NEON VST points taller than 3m, we can notice uncertainty rising as the crown radius gets bigger. In particular, there seems to be an under-representation of the crown size (Fig 1). This error propagates in the DBH-CR relationships (Fig2). However, it seems that using 20x20 plots average, the effect of this error may be reduced, but it requires a more quantitative backup.

If that is confirmed, makes probably sense to approach the study using the FIA subplot as an elementary unit, also because it would back up the logic behind graph descriptors (proxies of plot level interactions/demography)

• Using Jucker et al.
• Using FIA
• Using FIA + NEON VST
• Integrating multiple sources (optional)

use the file previously created to extract FIA like plots from DeepForest boxes

Start with fuzzy coordinates

• climate
• soil