Determination of log moisture content using early-time ground penetrating radar signal

AbstractFast, reliable and non-destructive measurement of log moisture content (MC) is important to optimize the forest value chain. We investigated the use of early-time ground penetrating radar signals to determine MC of stacked logs of black spruce, quaking aspen and balsam poplar in mill yards. Two approaches are presented: a linear fitting between the average envelope amplitude (AEA) and MC and a partial least square (PLS) regression between the signal amplitude and MC. We show that PLS regression enable us to greatly improve the prediction of MC in comparison with the AEA method. Moreover, the PLS technique allows us building models which integrate the signal variability due to the different species and log states (thawed and frozen). Models acquired with measurements collected on the logs ends produced usually higher accuracies (with ranging from 0.67 to 0.95 and root mean square error of prediction [RMSEv] ranging from 6% to 13%) than models acquired with measurements collected through the bark of...