Thermal-dissipation sap flow sensors may not yield consistent sap-flux estimates over multiple years

Sap flow techniques, such as thermal dissipation, involve an empirically derived relationship between sap flux and the temperature differential between a heated thermocouple and a nearby reference thermocouple inserted into the sapwood. This relationship has been widely tested but mostly with newly installed sensors. Increasingly, sensors are used for extended periods. After several months, tree growth, wounding, or other changes in water flow path may impair sensor performance. To quantify changes in sensor performance over time, we installed 23 sensors (one per tree) in 16-year-old Douglas-fir [Pseudotsuga menziesii (Mirb.) Franco] and red alder (Alnus rubra Bong.) in the western Cascades of Oregon and measured daily average sap flux (J_s) from April through July 2001 and 2002. We assumed the measurements from 2001 to be unimpaired and the response of J_s to vapor pressure deficit (δ) to be consistent under the same edaphic conditions. Differences from this assumption were attributed to "temporal sampling errors." During the study, soil moisture (θ), did not differ on similar calendar dates, yet the slope of J_s versus δ decreased significantly in the second year. In 2002, J_s in Douglas-fir was 45% less than in 2001; in red alder, 30% less. Variations in δ could not explain the differences. A correction for temporal sampling errors improved estimates of J_s from sensors used for more than one season. Differences in temporal sampling errors between the two species reveal underlying causal mechanisms. Evidence is presented that cambial growth causes errors in Douglas-fir.