Adequacy of Villalobos method to adjust eddy covariance latent heat flux

Latent (LE) and two sets of sensible (H) heat fluxes (H-ec and H-t2) were measured with an eddy covariance system (sonic anemometer, krypton hygrometer, two fine-wire thermocouples) over grass (6 days, summer 1997) and wheat (7 days, late spring 1999) in NE Spain. The objective was to evaluate a method to correct eddy covariance underestimation due to horizontal sensor displacement, based upon the similarity of covariance loss for both LE and H. First, this assumption was examined by regression analyses of measured lysimeter LE (LElys) versus eddy covariance LE (LEec) and H-ec versus H-t2. In general, regression slopes of LE and H were broadly similar, although this similarity depended upon the atmospheric stability conditions. For grass, L-ec was significantly lower than LElys while for wheat there was a close agreement between LElys and LEec. Ratios of horizontal sensor displacement to measurement sensor height above zero plane displacement (s:z(d)) were about 0.36-0.40 for grass and 0.11 for wheat. A further regression analysis was performed to compare LElys with corrected eddy covariance LE (LEecv ) values. That correction significantly improved eddy covariance LE values measured over grass but only in some cases, mostly under unstable (near to neutral) atmospheric conditions. It can be concluded that low s:z(d) ratios (less than 0.1) are preferable to reduce loss in covariance of LE. If the use of higher s:z(d) ratios cannot be avoided, the Villalobos method may reduce the expected loss in covariance but only under limited conditions. In these situations other solutions to reduce loss in covariance should be investigated further.