Multi-species test and calibration of an improved transient thermal dissipation system of sap flow measurement with a single probe

International audienceTransient thermal dissipation (TTD) systems provide a simple way to measure xylem sap flow with dual or single Granier-type probe, which gives lower energy consumption and higher accuracy due to its lessor sensitivity to thermal interferences. A new system, TTD heat within 5 min (TTD5), proposed on the reduction of the heating duration. This evolution captured interest through decreased energy consumption and increased temporal resolution. Within our study, the first objective was to test and calibrate this new system with a single probe for young rubber tree - Hevea brasiliensis. The second objective was to explore the sources of variability in calibration such as species, individual cut-stems and probe-wood contact. The complementary species consisted of two diffuse-porous species (mango tree - Mangifera indica, eucalyptus tree - Eucalyptus camaldulensis) and one ring-porous species (teak tree - Tectonia grandis). Twenty-eight response curves were assessed over a large range of flux densities from 0.5 to 10 l dm-2 h-1. The incremental rise of temperature from 30 to 300 s (T300-30) after commencement of heating was sensitive to flux density over the complete range. Compared with the full signal at 300 s, the incremental signal markedly reduced the variability between response curves within species and between species. Moreover, a new index K2, defined as (T0 - Tu)/T0, normalized the responses between 0 and 1. However, the responses had a non-linear trend above 5 l dm-2 h-1. Within diffuse-porous wood type, the species did not differ in calibration, whereas the ring-porous species was markedly different. A sigmoid function provided the best fit for the diffuse-porous species. Individual stems were identified as the main source of within-species variability in calibration. The normalizing K2 index removed the influence of probe-wood contacts, controlled through drilling difference; however, there was still an effect of individual stems interacting with flux density (P = 0.019). Replications of cut-stems and response curves are necessary to assess a reliable averaged calibration. In conclusion, the applicability of the TTD5 system with a single probe has been confirmed and several sources of variability in calibration have been evaluated