Estimating conductive sapwood area in diffuse and ring porous trees with electronic resistance tomography

Accurately estimating sapwood area is essential for modelling whole-tree or stand-scale transpiration from point-flow sap-flux observations. In this study, we tested the validity of electrical resistance tomography (ERT) to locate the sapwood-heartwood (SW/HW) interface for two ring porous (Quercus nigra L. and Quercus virginiana Mill.) and one diffuse porous (Acer rubrum L.) species. Estimates derived from the ERT analyses were compared with the SW/HW interface measured following dye perfusion testing. The ERT results revealed spatial variation in electrical resistance, with higher resistivity in the inner part of the cross sections. Regression analyses showed that ERT was able to accurately account for 97% and 80% of the variation in sapwood area (calculated as R2) for Q. virginiana (n = 19) and Q. nigra (n = 7), respectively, and 56% of the variation in the diffuse porous species (n = 8). Root mean square error (RMSE) values for sapwood areas of the ring porous species were 11.12 cm2 (19%) and 25.98 cm2 (33%) for Q. virginiana and Q. nigra, respectively. Sapwood area estimates for diffuse wood carried greater error (RMSE = 33.52 cm2 (131%)). Model bias for all sapwood area estimates was negative, suggesting that ERT had a tendency to overestimate sapwood areas. Electrical resistance tomography proved to be a significant predictor of sapwood area in the three investigated species, although it was more reliable for ring porous wood. In addition to the results, a comprehensive code sequence for use with R statistical software is provided, so that other investigators may follow the same method.