Calibration of thermal dissipation sap flow probes for ring- and diffuse-porous trees

Thermal dissipation probes (the Granier method) are routinely used in forest ecology and water balance studies to estimate whole-tree transpiration. This method utilizes an empirically derived equation to measure sap flux density, which has been reported as independent of wood characteristics. However, errors in calculated sap flux density may occur when large gradients in sap velocity occur along the sensor length or when sensors are inserted into non-conducting wood. These may be conditions routinely associated with ring-porous species, yet there are few cases in which the original calibration has been validated for ring-porous species. We report results from laboratory calibration measurements conducted on excised stems of four ring-porous species and two diffuse-porous species. Our calibration results for ring-porous species were considerably different compared with the original calibration equation. Calibration equation coefficients obtained in this study differed by as much as two to almost three orders of magnitude when compared with the original equation of Granier. Coefficients also differed between ring-porous species across all pressure gradient conditions considered; however, no differences between calibration slopes were observed for data collected within the range of expected in situ pressure gradients. In addition, dye perfusions showed that in three of the four ring-porous species considered, active sapwood was limited to the outermost growth ring. In contrast, our calibration results for diffuse-porous species showed generally good agreement with the empirically derived Granier calibration, and dye perfusions showed that active sapwood was associated with many annual growth rings. Our results suggest that the original calibration of Granier is not universally applicable to all species and xylem types and that previous estimates of absolute rates of water use for ring-porous species obtained using the original calibration coefficients may be associated with substantial error.