Skip to main content

Table 2 Predictors, pseudo R-squared (R2), and relative root mean square error from the cross validation (RMSECV %) of the two prediction methods

From: Modelling aboveground forest biomass using airborne laser scanner data in the miombo woodlands of Tanzania

Prediction method

Predictorsa

R2

RMSECV %

LMM (Model 3)

PF70, TF0, TF5, TF8, PL20, PL30, TL7, TL8

0.69

46.8

k-NN (k = 10)

PL80, TF0, TL2, PF80, TL4, TL8, PS40, TL5, TF2, TL7, TL6, PF70, MaxF, PL90, PL50, MeanL, TF7, PF10, PL60, TF3

0.58

55.9

  1. aPF10, PF70, and PF80 = Percentiles of the first echo canopy heights for 10 %, 70 %, and 80 (m); PL20, PL30, PL40, PL50, PL60, and PL90 = Percentiles of the last echo canopy heights for 20, 30, 40, 50, 60, and 90 % (m); TF0, TF2, TF3, TF5, TF7, and TF8 = Canopy densities corresponding to the proportion of first echoes above fraction #0 (1.3 m), #2, #3, #5, #7, and #8;TL2, TL4, TL6, TL7, and TL8 = Canopy densities corresponding to the proportion of last echoes above fraction #2, #4, #6, #7, and #8; MaxF and MeanL = Maximum and Mean of the canopy height distributions of the first and last echoes, respectively