Estimation of evapotranspiration fluxes at the field scale: parameter estimation, variability and uncertainties

The estimation of evapotranspiration (ET), a key process within the Earth's surface water and energy balance, remains an important challenge for a wide range of disciplines such as surface hydrology, irrigation management and meteorology. However, notwithstanding the considerable progress recently made in our understanding of the physical and biological processes governing ET, the accurate quantification of ET is very tricky to achieve, even at a limited spatial scale. In this study, we combine field measurements with numerical experiments to tackle issues related to the quantification of ET and the associated uncertainties for a maize cropped field using two different approaches, i.e. the agro-hydrological modelling and the soil water balance approach. For the agro-hydrological modelling, we mainly focus on the estimation of field-scale soil water content and on the identification of root water uptake parameters. With regard to the field-scale soil water content, we put forward that the within-field variability is large and that the maize crop plays a non-negligible role in the development of the soil water content patterns both at the field and at the maize row scale. For deriving root water uptake parameters (RWUP), we develop and test two different approaches, i.e. the simplified soil water balance and the inverse modelling approach. Using numerical experiments, we show that the simplified soil water balance approach produces quite accurate RWUP. On the other hand, the inverse modelling approach is only successful for some soils and for some conditions due to instability and nonuniqueness issues. For the soil water balance approach, we show that the accuracy of the local ET estimate is strongly dependent on the estimation method used to derive the bottom fluxes and that the use of pedotransfer functions is of little interest. For field-scale ET estimates, we show that the variability of ET is large both at the field scale (due to the variable crop growth) and at the maize row scale (due to the maize row layout). To produce accurate field-scale ET estimates, we suggest to scale up maize row scale ET estimates using the concept of temporal stability or using a covariant such as the Leaf Area Index. The results of our study suggest that the estimation of water fluxes or associated state variables for a row cropped field requires a two-step upscaling strategy, from the local scale to the row, then from the row to the field scale.