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

SummaryEvaporation measurements using two Bowen ratio energy balance (BREB) systems in a remote high altitude montane grassland catchment of the Drakensberg Mountains, Cathedral Peak, South Africa are reported on. Various methods of data verification and rejection of inaccurate measured air temperature and water vapour pressure gradients are examined. A theoretical analysis, based on the equivalent temperature, results in data rejection procedures using the measurement of the air temperature profile difference. Data rejection is necessary whenever the Bowen ratio approaches −1, resulting in extremely inaccurate and impossibly large positive or negative sensible heat and latent energy fluxes. Using the simplified energy balance, it is shown that when the Bowen ratio approaches the limit of −1, for which the available energy flux density approaches 0 W m−2, conditions are pseudoadiabatic and isobaric and that such conditions can be depicted by the wet-bulb temperature isolines of the psychrometric chart. Disregarding evaporation estimates for which the Bowen ratio values are between arbitrarily chosen values remedies the problem to some extent. With this method, daily total evaporation may be reasonable but 20-min values unreasonable during mainly early morning and late afternoon periods. A more sensitive and dynamic approach is used to prevent BREB data from being excluded unnecessarily and to prevent rogue values escaping detection. Once the rejection procedures were applied, the 20-min BREB latent energy flux estimates compared well with measurements from a weighing lysimeter adjacent the site. Three methods were used to estimate the exchange coefficient K which allowed flux estimation for when BREB data are invalid or lacking. One method involved calculating K from wind speed only and the second method was based on the MOST-dependent temperature-variance method for which the 20-min standard deviation of 1-Hz air temperature data were used. From independent measurements of sensible heat H and latent energy LE, a time-invariant exchange coefficient K was also determined from measurements of the air temperature profile difference. These methods were used when there were invalid water vapour pressure data due to condensation in the hoses or problems with the cooled dew point mirror or when the fine-wire thermocouples were damaged or when there were unreliable estimates of the air temperature gradient. The time-invariant value for K used in one of the methods, 0.239 m2 s−1, was confirmed for a mixed grassland catchment using independent eddy covariance and surface-layer scintillometer measurements of H and Bowen ratio measurements of the air temperature profile difference.