The thermal dissipation technique is widely used to estimate transpiration of individual trees and forest stands but there are numerous conflicting reports regarding the accuracy and universality of the technique. We evaluated the accuracy of the thermal dissipation technique by comparing actual sap flow with sap flow estimates derived from thermal dissipation sensors constructed according to the original design described by Granier and commercially available sensors. Estimated sap flow, with thermal dissipation sensors, and actual sap flow, measured with potometers, were compared using five saplings of white oak (Quercus alba L.), American elm (Ulmus americana, L.), eastern cottonwood (Populus deltoides Bart.), sweetgum (Liquidambar styraciflua L.), shortleaf pine (Pinus echinata Mill.) and loblolly pine (Pinus taeda L.). These species represent ring-porous, diffuse-porous and tracheid xylem anatomies. We inserted two replicate sensors horizontally 20 mm into sapwood on opposite sides of each tree stem to estimate sap flow while simultaneously measuring actual sap flow. An additional pair of sensors were inserted in the same manner and in the same aspect as previous pair, but approximately 0.2 m higher on the stem to investigate variation along the stem.

Results/Conclusions

We observed considerable circumferential variation in sap flow for all species, but only minimal variation when sensors were located at the same aspect. Species with the same xylem anatomy generally did not show similar relationships between estimated and actual sap flow with the exception of two diffuse porous species. In almost all instances estimated sap flow deviated substantially from actual sap flow. The two different types of thermal dissipation sensors we used also produced significantly different estimates in most of the tree species. Estimated sap flow almost always underestimated actual sap flow. New α and β coefficients, generated from the test data, were created for each sensor type. However, the coefficients showed no consistency across the six species, or among species of each xylem anatomy. Our data indicate that it is essential that species-specific calibrations be performed when employing the thermal dissipation technique for estimating sap flow, and consideration should be made for the circumferential variation in hydraulic conductivity in the stem of the species being measured.