This document describes a study that used near infrared (NIR) spectroscopy to develop calibrations to estimate wood properties (density, modulus of elasticity (MOE), and modulus of rupture (MOR)) of multiple pine species. Samples from 8 pine species growing in Brazil were collected and tested for wood properties. NIR spectra were collected on the radial and transverse faces. Calibrations using NIR spectra provided reasonably strong estimations of density, MOE, and MOR across all pine species. Transverse face spectra produced slightly better calibrations than radial face spectra. The lab-based FOSS static spectrometer produced the strongest calibrations overall.

1. Estimation of the wood properties of
tropical, sub‐tropical and temperate
pine species using NIR spectroscopy
L.R. Schimleck1, J. L. M. Matos2,
R. Trianoski2 and J. G. Prata2
1Warnell School of Forestry & Natural Res., University of Georgia
2Department of Forest Sciences, Federal University of Parana

2. Introduction
Extensive field trials have been established
in South America that aim to evaluate the
growth and adaptability of several tropical,
sub‐tropical and temperate pine species.
To fully assess their suitability for
deployment in plantations wood property
information needs to be collected for
multiple species, which is prohibitively
expensive using lab‐based methods.
Interest exists in using near infrared (NIR)
spectroscopy to estimate mechanical
properties (MOE, MOR). This study aims to
develop multiple pine species calibrations
and to compare calibrations using lab
based and portable spectrometers.

3. Species and sites examined
Samples collected from trials established by CAMCORE and The Center for
the Genetic Conservation and Management of Tropical Pines
Species Age Location
Pinus caribaea var. bahamensis 17 anos Itararé ‐ SP
Pinus caribaea var. caribaea 17 anos Itararé ‐ SP
Pinus caribaea var. hondurensis 18 anos Ventania ‐ PR
Pinus chiapensis 18 anos Ventania ‐ PR
Pinus maximinoi 18 anos Ventania ‐ PR
Pinus oocarpa 18 anos Ventania ‐ PR
Pinus taeda 18 anos Ventania ‐ PR
Pinus tecunumanii 18 anos Ventania ‐ PR

4. Location of sites
1 = Camcore, 2 = C.C.G.M.P.T

5. Sample collection

6. Sample collection
Species Number of Age (yr) Average diameter at Average height
trees breast height (cm) (m)
Pinus caribaea var. bahamensis 5 17 37 27.0
Pinus caribaea var. caribaea 5 17 37 26.3
Pinus caribaea var. hondurensis 5 18 42 25.1
Pinus chiapensis 5 18 46 29.8
Pinus maximinoi 5 18 47 27.6
Pinus oocarpa 5 18 41 26.7
Pinus taeda 5 18 32 18.4
Pinus tecunumanii 5 18 46 25.9
Base of the tree cut to provide two 2.6 m long logs
Ist log used for veneer, 2nd log used for wood
property analysis
10 cm thick slab cut through the pith, and consecutive
static bending samples cut from the slab

7. Wood property evaluation

8. Wood property evaluation

9. Wood properties
Properties Density Elastic properties Compression Shear Hardness
Species (12%) MOR MOE MOR MOE MPa N
kg/m3 MPa MPa MPa MPa
P. c.bahamensis 484 63 6.568 33 9.550 10 2795
P. c caribaea 433 56 6.060 30 10.480 9 2138
P. c.hondurensis 500 64 7.206 36 11.324 11 2667
P. chiapensis 440 61 7.590 36 11.546 9 2511
P. maximinoi 530 70 9.045 40 14.133 11 3383
P. oocarpa 540 68 7.788 41 13.597 12 3403
P. taeda 516 63 8.234 40 13.197 10 3138
P. tecunumanii 561 71 8.878 42 15.109 11 3393

10. Near infrared (NIR) spectroscopy
Widely used to measure parameters that are time
consuming to measure
NIR spectrum closely related to wood chemistry
Applicable to static bending samples and has been
used to estimate a range of wood properties
Calibrations limited to a small number of species or
sites
Global calibrations – rare in forestry (several reasons)
• NIR applied to wood for only a short time
• Most properties are expensive to measure
• Limited networks to share samples

11. Near infrared (NIR) spectroscopy
Transverse Radial

12. Near infrared (NIR) spectroscopy

13. Near infrared (NIR) spectroscopy

14. Near infrared (NIR) spectroscopy

15. Near infrared (NIR) spectroscopy

16. Building calibration models based on NIR and wood property data
Estimation of a parameter involves the following steps:
Collect spectra of calibration samples
Develop a calibration (regression)
(y = B0 + X1*B1 + X2*B2 + ………..+ XN*BN)
Collect NIR spectra of test (or unknown) samples
Estimate parameter of interest for test set samples
using the calibration

17. Calibrations for density (334 samples)
Calibration Number of R2 SEC SECV RPDC
Factors
FOSS Static
Density (Kg/m3) ‐ Radial face 8 0.81 35.6 38.5 2.1
Density (Kg/m3) ‐ Transverse face 8 0.80 36.8 38.3 2.2
FOSS Probe
Density (Kg/m3) ‐ Radial face 4 0.51 57.8 58.8 1.4
Density (Kg/m3) ‐ Transverse face 10 0.68 46.8 49.3 1.7
ASD Probe
Density (Kg/m3) ‐ Radial face 8 0.73 43.2 45.8 2.0
Density (Kg/m3) ‐ Transverse face 10 0.70 45.0 48.5 1.8

18. Calibration for Density (transverse face, FOSS Static)
900
800
Measured density (kg/m3)
700
600
500
8 factors
R2 = 0.80
400
SEC = 36.8 kg/m3
300 SECV = 38.3 kg/m3
RPDc = 2.2
200
200 300 400 500 600 700 800 900
NIR-estimated density (kg/m3)

19. Calibrations for MOE (334 samples)
Calibration Number of R2 SEC SECV RPDC
Factors
FOSS Static
MOE (MPa) ‐ Radial face 8 0.78 1154 1215 2.1
MOE (MPa) ‐ Transverse face 6 0.81 1095 1123 2.2
FOSS Probe
MOE (MPa) ‐ Radial face 7 0.55 1668 1747 1.4
MOE (MPa) ‐ Transverse face 8 0.68 1411 1464 1.7
ASD Probe
MOE (MPa) ‐ Radial face 9 0.73 1308 1413 1.8
MOE (MPa) ‐ Transverse face 6 0.75 1252 1303 1.7

20. Calibration for MOE (transverse face, FOSS Static)
16000
Measured stiffness (MPa)
12000
8000
6 factors
R2 = 0.81
4000
SEC = 1095 MPa
SECV = 1123 MPa
RPDc = 2.2
0
0 4000 8000 12000 16000
NIR-estimated stiffness (MPa)

21. Calibrations for MOR (334 samples)
Calibration Number of R2 SEC SECV RPDC
Factors
FOSS Static
MOR (MPa) ‐ Radial face 4 0.69 8.4 8.6 1.7
MOR (MPa) ‐ Transverse face 6 0.73 7.9 8.0 1.9
FOSS Probe
MOR (MPa) ‐ Radial face 7 0.50 10.6 11.2 1.3
MOR (MPa) ‐ Transverse face 10 0.61 9.3 10.4 1.4
ASD Probe
MOR (MPa) ‐ Radial face 8 0.67 8.6 9.2 1.6
MOR (MPa) ‐ Transverse face 5 0.64 9.0 9.2 1.6

22. Calibration for MOR (transverse face, FOSS Static)
120
100
Measured MOR (MPa)
80
60
6 factors
40 R2 = 0.73
SEC = 7.9 MPa
20 SECV = 8.0 MPa
RPDc = 1.9
0
0 20 40 60 80 100 120
NIR-estimated MOR (MPa)

23. Conclusions
• Wood property calibrations obtained for density, MOE and
MOR using several pine species growing on two sites
• Transverse surface spectra marginally better than radial face
spectra
• FOSS Static system provided the strongest calibrations,
followed by ASD probe and FOSS probe systems