[Insert Fig. 10 here]
The HRM requires a known value of k to correct errors due to probe misalignment. Typically, a constant k is assumed. Burgess et al. (2001) used the Tmax method (Eqn. 15) to calculate k . This is a somewhat circular procedure because the Tmax method requires knowledge of probe locations. It would be preferable to concurrently solve for all three unknowns –∆x 1, ∆x 2 and k – as shown by Chen et al. (2012). Such a calibration approach may not be achievable unless constrained with assumptions or additional calculations (e.g. sap velocity independently inferred via an empirical function in Chen et al. (2012)). We suggest an alternative calibration procedure. First, alignment is checked using the HRM method suggested by Burgess et al. (2001). Second, the non-unique solution is constrained by comparing velocities using the CHPM, which is independent of k . Finally,k is identified with the Tmax method under zero flow conditions (see the misalignment calibration procedures in Methods S4 ). Fig. 11 below compares such calibration results. When the average solution is chosen from the HRM method, sap velocities estimated using the DRM match well with those estimated using the CHPM. In this scenario, any misalignment is well corrected (CHPM is sensitive to misalignment) and k is appropriately estimated (DRM is sensitive to k at low flow).
The three-probe approach used by the DRM allows diagnosis of misalignment corrections. Combined with calculations based on the CHPM, it also allows for detemination of both k and misalignment under moderate conditions (e.g. typical day-time).