Variability and uncertainty in forest biomass estimates from the tree to landscape scale: the role of allometric equations

Table 7 Out-of-bag model evaluation metrics (pseudo R² and RMSE) from the Random Forest aboveground biomass models

Allometric Equation	RMSE	RMSE percent of mean	Pseudo R²	Percent Bias	Number of Predictors (Top 3 Predictors)	mtry	ntrees
Local	48.1	35.0	0.5936	0.9	11 (Band 1, NDII, DEM)	8	2000
Jenkins	54.8	37.9	0.5377	0.9	10 (NDII, DEM, Band 1)	5	1500
FIA-CRM	40.8	40.8	0.5463	1.1	9 (NDII, Band 2 texture_5 × 5 mean, PAS)	4	1000

These values are the prediction errors only, and do not include allometric error. FIA-CRM Forest Inventory and Analysis Component Ratio Method, RMSE root mean square error, NDII normalized difference infrared index, DEM digital elevation model, PAS precipitation as snow, mtry number of predictor variables randomly sampled at each split in model, ntrees number of trees grown in model

ISSN: 1750-0680