. 2021 Apr 2;24(7):1792–1809. doi: 10.1007/s10021-021-00618-8

Table 1.

RF Model Efficiency.

Response Variable	Training S^a NRMSE^b (%)	Training T NRMSE^b (%)
PAR	6.8	4.8
T	5.5	5.0
GPP	7.8	7.1
ER	9.9	7.7
GPP including T, PAR	7.2	7.1
ER including T, PAR	8.9	7.2

^aFigures displayed in “Results” section refer to this setup.

^bError estimates refer to the RMSEs on the predicted data not used during training, normalized by the range (that is, the difference between the maximum and minimum value) of the measured data. Lower values indicate less residual variance. See SI Results for the single sites prediction errors.