Bowen ratio evaporation measurement in a montane grassland: data integrity and fluxes

One of the most important factors affecting water supply from any catchment is the evaporative loss from the vegetative community, which in the Drakensberg, South Arica, is a fire-determined climax grassland. The Bowen Ratio Energy Balance micrometeorological method (BREB) allowed accurate and reliable evaporation measurements on a long term basis. The research location was Catchment VI, Cathedral Peak Forestry Research Station, in the foothills of the Drakensberg, Natal, South Africa at 29.00 S, 29.25 E, at an altitude of 1935 m and with a predominantly north-facing aspect - average slope of 0.27 (about 15^o). It is 0.677 km² 68 ha) in area and varies in altitude from 1847 to 2076 m. Two locations were used in the catchment - an upslope site at which two Bowen ratio systems were employed for simultaneous measurement comparisons and a location within a riparian zone. A trend in the value of a generalized exchange coefficient K was found that agreed well with the sparse literature available on this subject. An overall average for K was 0.239 m² s^-1(for LE greater than 200 W m^-2). The lack of variability between measurement time intervals showed that the ratio of the exchange coefficient for sensible heat to that for latent heat, K_h/K_w, is not changing markedly from one measurement period to another - implying that the ratio is close to unity and the assumption of a generalised K coefficient is reasonable. An automatic data exclusion procedure was used to replace evaporation values when the Bowen ratio approached -1. Excluded data values were replaced by equilibrium evaporation, usually at early morning and late afternoon times. The 1992 annual total showed that during this particularly dry year, measured evaporation exceeded the annual rainfall. Daily total evaporation typically decreased from a summer maximum of 17 MJ m^-2 (7 mm) to a winter season minimum close to zero. In 1992, the one Bowen ratio system was moved to a riparian location and the other remained upslope. For January 1992, the measured daily total evaporation at the riparian location was about 30 % greater than that upslope and greater than that upslope on two out of three days. The daily total energy balance components, for each day of January 1992, for the riparian and upslope locations were statistically different. Net radiant density, latent heat and soil heat density were greater for the riparian location compared with that upslope but the sensible heat density was significantly smaller. Over the study period, peak summer evaporation rates appeared independent of rainfall or soil water content, but were related to the time between burns. The winter season evaporation rate varied little between one year old, two years old and burnt canopies. The total evaporation for 39 months was 3330.7 mm, compared with a rainfall total of 3618.4 mm and a total net radiant density of 12470.82 MJ m^-2. The Bowen ratio energy balance method was suitable for long term monitoring of daily total evaporation and capable of detecting the effects of management fires.