Modelling and measurement of water productivity and total evaporation in a dryland soybean crop

Simpler crop models simulating evaporation are needed to provide information to farmers, policy makers and decision makers on how to maximise crop yield responses to water. This is becoming important as the frequency and severity of droughts in South Africa is increasing. In this regard, prediction of yield, determination of water productivity and total evaporation (ET) are increasingly becoming essential in water resource management. The overall objective of the study was to compare the FAO AquaCrop daily model output of ET to the residual ET for non-stressed dryland soybean in a sub-humid climate. Energy balance residual ET estimates using an eddy covariance (EC) system and modelled ET using AquaCrop obtained from Glycine max (L.) Merrill grown in the midlands of KwaZulu-Natal, South Africa during the 2012/13 growing season are compared. The modelled and observed yield showed good agreement, while the residual ET was 21.6% less than the modelled. The energy balance closure computed using the daily sums of sensible heat and latent energy fluxes against daily available energy flux for unstable atmospheric conditions was 0.77. A closure of 0.99 was achieved when the EC latent energy flux was replaced with residual latent energy flux. A good fit between the modelled and observed percentage green canopy cover was observed (slope = 0.86, intercept = 15.46%, root mean square error = 10.50% and R-2 = 0.83). Season-long daily residual ET values were consistently low for most of the growth stages compared to the modelled, except for the maturity stage. However, the residual ET comparisons with the AquaCrop model improved after gap-filling was applied to discarded data and for when the EC system failed.