How Reliable Are Heat Pulse Velocity Methods for Estimating Tree Transpiration?

Transpiration is a significant component of the hydrologic cycle and its accurate quantification is critical for modelling, industry, and policy decisions. Sap flow sensors provide a low cost and practical method to measure transpiration. Various methods to measure sap flow are available and a popular family of methods is known as heat pulse velocity (HPV). Theory on thermal conductance and convection, that underpins HPV methods, suggests transpiration can be directly estimated from sensor measurements without the need for laborious calibrations. To test this accuracy, transpiration estimated from HPV sensors is compared with an independent measure of plant water use such as a weighing lysimeter. A meta-analysis of the literature that explicitly tested the accuracy of a HPV sensors against an independent measure of transpiration was conducted. Data from linear regression analysis was collated where an R-2 of 1 indicates perfect precision and a slope of 1 of the linear regression curve indicates perfect accuracy. The average R-2 and slope from all studies was 0.822 and 0.860, respectively. However, the overall error, or deviation from real transpiration values, was 34.706%. The results indicate that HPV sensors are precise in correlating heat velocity with rates of transpiration, but poor in quantifying transpiration. Various sources of error in converting heat velocity into sap velocity and sap flow are discussed including probe misalignment, wound corrections, thermal diffusivity, stem water content, placement of sensors in sapwood, and scaling of point measurements to whole plants. Where whole plant water use or transpiration is required in a study, it is recommended that all sap flow sensors are calibrated against an independent measure of transpiration.