Direct observation of the evaporation of intercepted water over an old-growth forest in the eastern Amazon region

Interception of rainfall by the forest canopy and the subsequent re-evaporation into the atmosphere constitute an important part of the hydrological balance over forests. Estimates of interception vary from 10-30% of annual incident precipitation, depending on forest type, season and precipitation intensity. Furthermore, an appreciable fraction of water vapor in the Amazon is recycled through evapotranspiration and the re-evaporation of intercepted water. We directly observe the evaporation of intercepted water over an old-growth forest site of LBA-ECO (Km 67), located in the Tapajos National Forest south of Santarém, Brazil, in the eastern Amazon region. An eddy-covariance system consisting of two Campbell CSAT sonic anemometers and two Licor CO2/H2O analyzers was operating along with a Vaisala CT-25K ceilometer during periods of time from April 2001 to June 2003. The ceilometer provides 15-second measurements of a backscatter profile from the surface to 7500 m at 30-m resolution. Precipitation can be detected from the ceilometer backscatter profile, allowing us to identify precipitation periods when the forest canopy intercepted precipitation, including light rainfall events that were not recorded by a rain gauge at the site. Do these light rainfall events provide a large fraction of the re-evaporation?

Frequently, sonic anemometers fail or operate intermittently during and just after rain. We identify precipitation and interception events from the ceilometer backscatter profile and observe the water vapor flux from the time the eddy-covariance system began functioning during each event until several hours after the end of each event. We form an ensemble average from many of these precipitation/interception events. We compare the observed evaporation for the events against the half-hourly evaporation reported using the conventional flux reporting method and also against that predicted by commonly-used models (e.g., Penman-Monteith method and the interception model of Gash) to examine the applicability of such methods.