Monin–Obukhov length as a cornerstone of the SEBAL calculations of evapotranspiration

Spatial variability in percolation fluxes was studied in field plots cropped with banana plants, which induce very heterogeneous rainfall partitioning at the soil surface, with high subsequent infiltration in Andosols. Percolation fluxes were measured for just over a year at 1 –7 day intervals in eight wick (WL) and gravity lysimeters (GL) that had been buried in the soil at a depth of 60 cm. The results revealed that WL captured unsaturated fluxes while GL only functioned after ponding occurred. The percolation flux measurements were highly biased with both systems, i.e. overpercolation with WL and underpercolation with GL. Percolation fluxes seemed, however, to be mainly unsaturated in the soil types studied. High percolation flux variability was noted on a plot scale, which could be explained by the vegetation structure: total percolation flux (WL) was 2.1-fold higher under banana plants; saturated percolation flux (GL) was 7fold higher under banana plants and almost absent between banana plants. Eighty-eight per cent of the total variance in percolation flux could be explained by the rainfall intensity under the banana canopy, calculated while taking the rainfall partitioning by the vegetation and the initial water status into account. The number of lysimeters required for assessing percolation flux in a field plot can be reduced by taking the spatial patterns of the flux boundary conditions into account.