Table 2 Prediction models of whole-tree WD estimation (VWWD, in g.cm−3).

From: Leveraging Signatures of Plant Functional Strategies in Wood Density Profiles of African Trees to Correct Mass Estimations From Terrestrial Laser Data

Models

Parameters

Performance

a

b

c

d

R²

RSE

AIC

B

CV

(m1) VWWD = a + b*WDStu

0.07842 (0.00682)

0.78915 (0.01076)

  

0.87

0.049

−2629.1

0.86

14.8

(m2) VWWD = a + b*WDStu + c*DBH

0.05455 (0.00679)

0.78326 (0.01011)

0.00048 (0.00005)

 

0.88

0.046

−2732.3

0.75

15.2

(m3) VWWD = a + b*WDStu + c*DBH + d*Sm

0.10013 (0.01077)

0.77299 (0.01012)

0.00042 (0.00005)

−0.05819 (0.0108)

0.89

0.045

−2759

0.74

15

(m4) VWWD = a + b* WDGWD

0.1721 (0.00763)

0.63638 (0.01189)

  

0.78

0.063

−2200.6

1.41

21.8

(m5) VWWD = a + b*WDGWD + c*DBH

0.13406 (0.00783)

0.63614 (0.01105)

0.00067 (0.00006)

 

0.81

0.059

−2320.9

1.19

23.8

(m6) VWWD = a + b*WDGWD + c*DBH + d*Sm

0.18233 (0.01325)

0.62656 (0.01113)

0.0006 (0.00006)

−0.06258 (0.01395)

0.81

0.058

−2338.9

1.18

23.7

  1. The models are based on species average WD extracted from the Global Wood Density database (WDGWD in, g.cm−3), individual tree WD from stumps (WDStu, in g.cm−3) and tree structure parameters (stem DBH in cm and the stem morphology index, Sm). Model coefficients are provided along with standard errors (in brackets). All coefficients are highly significant (P < 0.001). Model performance is characterized using classical fit metrics for model residuals (R², RSE, AIC), the measure of bias (B) and total error (coefficient of variation, CV) computed for AGB estimations (see the Methods section).
Back to article page