Seasonal variation of boreal forest surface conductance and evaporation

Long term measurements (June 1994 to December 1996) of evaporation were mad e in a boreal forest in central Sweden. Fluxes were measured continuously w ith eddy-correlation systems from a 100 m tower. Surface conductance and po tential evaporation were estimated using the Penman-Monteith equation. The overall average evaporation during the 947-days observation period was 1.07 mm d(-1). The average evaporation from June to December 1994 was 1.25 mm d (-1), 1.07 mm d(-1) during 1995, and 0.97 mm d(-1) during 1996. Maximum dai ly rates were typically 4 mm d(-1) around mid-summer in 1994 and 1995 and s lightly less in 1996. During the winter period from November to March, the evaporation sometimes reached 0.5 mm d(-1). Generally, the actual evaporati on followed the dynamics of the potential evaporation fairly well. As a tot al for the entire period, the actual evaporation accounted for 918 mm, or 6 9% of the potential evaporation (1332 mm). There were three major time peri ods with a considerable evaporation deficit: the late summer (July-August) 1994, the late summer (August) 1995, and the spring to early summer (March- June) 1996 (Table I). The first two periods were correlated with low soil m oisture while the last probably was caused by low winter temperatures. A su rface conductance model showed little dependence on soil moisture. Radiatio n and vapour pressure deficit were the most important factors. When soil water was not limiting, the diurnal courses of surface conductanc e showed a steep increase in the morning followed by an almost linear decre ase, the coniferous trend caused by stomatal control. During periods of wat er stress a midday depression occurred after the morning maximum. During wi ntertime, the surface conductance hardly ever approached to zero. A simple model for surface conductance based on global radiation and vapour pressure deficit fitted the estimated data fairly well except for calm nig hts with poor turbulent mixing. This was apparently due to water vapour sto rage within and above the canopy. Including the friction velocity in the mo del parameterisation improved the fitting considerably. (C) 1999 Elsevier S cience B.V. All rights reserved.