Rainfall interception modelling: Is the wet bulb approach adequate to estimate mean evaporation rate from wet/saturated canopies in all forest types?

• Saturated crowns of individual sparse trees behave as wet bulbs.
• Evaporation from fully ventilated canopies is well estimated by the wet bulb approach.
• When applicable, this approach may be preferable to the Penman–Monteith model.
• Fully ventilated canopy conditions do not depend solely on crown cover fraction.
• Proper evaluation of wet canopy evaporation is critical to Gash’s interception model.

SummaryThe Penman–Monteith equation has been widely used to estimate the maximum evaporation rate (E) from wet/saturated forest canopies, regardless of canopy cover fraction. Forests are then represented as a big leaf and interception loss considered essentially as a one-dimensional process. With increasing forest sparseness the assumptions behind this big leaf approach become questionable. In sparse forests it might be better to model E and interception loss at the tree level assuming that the individual tree crowns behave as wet bulbs (“wet bulb approach”). In this study, and for five different forest types and climate conditions, interception loss measurements were compared to modelled values (Gash’s interception model) based on estimates of E by the Penman–Monteith and the wet bulb approaches. Results show that the wet bulb approach is a good, and less data demanding, alternative to estimate E when the forest canopy is fully ventilated (very sparse forests with a narrow canopy depth). When the canopy is not fully ventilated, the wet bulb approach requires a reduction of leaf area index to the upper, more ventilated parts of the canopy, needing data on the vertical leaf area distribution, which is seldom-available. In such cases, the Penman–Monteith approach seems preferable. Our data also show that canopy cover does not per se allow us to identify if a forest canopy is fully ventilated or not. New methodologies of sensitivity analyses applied to Gash’s model showed that a correct estimate of E is critical for the proper modelling of interception loss.