Understanding radial and azimuthal variation, and tree-to-tree variation, in sap flux density (Fd) as sources of uncertainty is important for estimating transpiration using sap flow techniques. In a Japanese cedar (Cryptomeria japonica D. Don.) forest, Fd was measured at several depths and aspects for 18 trees, using heat dissipation (Granier-type) sensors. We observed considerable azimuthal variation in Fd. The coefficient of variation (CV) calculated from Fd at a depth of 0-20 mm (Fd1) and Fd at a depth of 20-40 mm (Fd2) ranged from 6.7 to 37.6% (mean = 28.3%) and from 19.6 to 62.5% (mean = 34.6%) for the -azimuthal directions. Fd at the north aspect averaged for nine trees, for which azimuthal measurements were made, was -obviously smaller than Fd at the other three aspects (i.e., west, south and east) averaged for the nine trees. Fd1 averaged for the nine trees was significantly larger than Fd2 averaged for the nine trees. The error for stand-scale transpiration (E) estimates caused by ignoring the azimuthal variation was larger than that caused by ignoring the radial variation. The error caused by ignoring tree-to-tree variation was larger than that caused by ignoring both radial and azimuthal variations. Thus, tree-to-tree variation in Fd would be more important than both radial and azimuthal variations in Fd for E estimation. However, Fd for each tree should not be measured at a consistent aspect but should be measured at various aspects to make accurate E estimates and to avoid a risk of error caused by the relationship of Fd to aspect.
Keywords: sap flux; sapwood area; scaling; thermal dissipation probe; water availability.