How do we know if Real Twig returns accurate volume estimates? We
rigorously tested our method against a high quality reference data set
that were both laser scanner and destructively sampled. The laser
scanning was done in leaf-off conditions with a Riegl VZ-400. The
destructive sampling contains total branch and main stem dry mass, and
also basic density for both the main stem and the branches. We used
different versions of TreeQSM with the same input parameters per tree to
test our model.
SimpleForest
It is important to note that Real Twig was not tested with
SimpleForest during its development. While Real Twig does improve volume
estimates for SimpleForest versus its built in allometric corrections,
there are still improvements to be made, as SimpleForest QSM cylinders
are generally much more overestimated than TreeQSM cylinders, making the
identification of “good” cylinders difficult.
Below are the mass estimates and statistics using SimpleForest v5.3.2
with its built in vessel volume correction, and Real Twig applied to the
same QSMs.
| SimpleForest |
| Metric |
Total Woody AGB |
Main Stem Biomass |
Branch Biomass |
| Mean Relative Error (%) |
51.031 |
27.775 |
174.363 |
| RMSE (kg) |
177.462 |
48.287 |
141.470 |
| Relative RMSE (%) |
42.285 |
14.720 |
154.385 |
| CCC |
0.934 |
0.990 |
0.669 |
| Real Twig (SimpleForest) |
| Metric |
Total Woody AGB |
Main Stem Biomass |
Branch Biomass |
| Mean Relative Error (%) |
15.850 |
11.909 |
41.238 |
| RMSE (kg) |
27.605 |
24.525 |
18.651 |
| Relative RMSE (%) |
6.578 |
7.476 |
20.353 |
| CCC |
0.998 |
0.997 |
0.985 |
