Study On Estimation Of E. Pellita's Microfibril Angle And Air-Dry Density By Near Infrared Spectroscopy Technology

In this study, Eucalyptus pellita including 28 families of 6 provenances from Leizhou forestry bureau's seedlot were selected as the material. The standing tree's dynamic MOE, radial MFA, tangential MFA and air-dry density were respectively tested by nondestructive instrument: FAKOPP 2D, x-ray diffractometry and x-ray densitometry. The variances of these wood properties among provenances and families were compared and analyzed. Also the models were established using near infrared (NIR) spectroscopy technology to predict the MFA and density, respectively. This research can supply reference to forest cultivation and heredity improvement, improve the quality of forest and rise the level of wood processing and utilization. The results are as follows:(1) Among different provenances, the difference in MOE is significant at 0.05 level. The provenances with higher MOE are 18597, 18598 and 18197, but 18749 with lower MOE. The difference in radial MFA is significant at 0.01 level, while the tangential MFA isn't significant at 0.05 level. The provenance 18598 has the most significant difference between radial and tangential MFA, and the D-value is 1.33 degree. The difference in air-dry density is extra significant at 0.01 level, and the density range is 0.62-0.69 g/cm3.(2) Among different families, the differences in MOE and MFA are significant at 0.01 level. The families with higher MOE are 230, 247, 132 and 234, but 205, 92, 82 and 158 with lower MOE. The families with air-dry density, no less than 0.70g/cm3, are 5, 43, 132, 230, 247, 270 and 123, while 92, 161, 42, 130, 192, 95, 139, 205 and 200's air-dry density are no more than 0.60 g/cm3.(3) The corresponding families'radial MFA were compared with tangential MFA by ANOVA, then draw the conclusion that 214, 270 and 132 have the significant difference at 0.05 level, 234, 83 and 247 have the significant difference at 0.01 level, while the rest with no significant difference. The tangential MFA is bigger than radial MFA, the D-value is 0.46 degree. The radial variance of radial MFA is the same as tangential MFA. From bark to pith both of the two sections'MFA increase, but a little rally in the bark, also differences between MFA in the pith and the other positions are obvious.(4) The E. pellita's air-dry density was divided into 3 parts in the radial direction. The differences among these 3 parts are very significant. They follow the trend of sapwood>the middle wood>the wood near pith. The average density of sapwood, middle wood and the part near the pith are 0.74g/cm3, 0.64 g/cm3 and 0.53 g/cm3.(5) For E. pellita, some provenance has the potential of using as solid wood processing material, such as provenance 18597 with high MOE , high air-dry density and low MFA. Family 230, 247 and 132 have high air-dry density and MOE. Although family 5 has high air-dry density, its radial MFA is large. Meanwhile, family 234 has high MOE, but its tangential MFA is large.(6) The fast, nondestructive prediction of radial and tangential MFA can realized by NIR technology based on the radial and tangential MFA measured by x-ray diffractometry, and the correlation coefficient(r) of prediction both can reach to 0.92. There is no great difference between radial MFA and tangential MFA, but the error of radial MFA model is bigger than that of tangential's. The standard error of prediction(SEP) of radial MFA and tangential MFA model is 1.44 degree, 1.33 degree. The prediction result of these two models for MFA above 20 degree is a little worse.(7) The fast, nondestructive, exact prediction of air-dry density can realized by NIR technology based on the density determined by x-ray densitometry, and the NIR fitted values have good relationship with the original values. The correlation coefficient(r) of calibraton and validation models is 0.936, 0.921, and the standard error of calibration(SEC) and standard error of prediction(SEP) is 0.044 g/cm3 and 0.048 g/cm3, respectively. The calibration was used to predict the air-dry density of random selected samples, which weren't contained in the density calibration model. The prediction result is good, with coefficient of determination (R2) of 0.9106.