Machine learning and geostatistical approaches for estimating aboveground biomass in Chinese subtropical forests

Forest Ecosystems

Table 2 Summary of the modeling performance measures derived from the three RF training models

Year	Variables used for model prediction	RMSE (t∙ha^− 1)	RMSE (%)	Bias (t∙ha⁻¹)	Bias (%)	MAE (t∙ha⁻¹)	R²
1992	TM7_1, second77, B2, SAVI, TM14, TM17, B5, TM37, VRI, B7	9.05	71.57	−0.01	−0.05	4.04	0.86
2002	mean55, RSI, mean77, TM37, VRI, TM12, TM27, TCD, elevation, TM34	43.68	64.34	0.35	2.29	16.68	0.90
2010	mean77, B1, TM34, HH_HV, TM2_1, TM25, ARVI, B2, RFDI, EVI	37.51	56.21	1.09	1.02	13.88	0.95