State and parameter update in a coupled energy/hydrologic balance model using ensemble Kalman filtering

SummaryThe capability to accurately monitor and describe daily evapotranspiration (ET) in a cost effective manner is generally attributed to hydrological models. However, continuous solution of energy and water balance provides precise estimations only when a detailed knowledge of sub-surface characteristics is available. On the other hand, residual surface energy balance models, based on remote observation of land surface temperature, are characterised by sufficient accuracy, but their applicability is limited by the lack of high frequency and high resolution thermal data. A compromise between these two methodologies is represented by the use of data assimilation scheme to include sparse remote estimates of surface fluxes into continuous modelling. This paper aims to test the combined use of coupled energy/water budget model and data assimilation schemes to assess daily evapotranspiration at field scale in a typical Mediterranean environment characterised by sparse olive trees. The continuous model was applied at hourly scale using remote multispectral images in the short-wave and standard meteorological information. The model was validated by means of contextual micro-meteorological information adopting the best available parameterisation (including root zone depth). The validation suggests an accuracy of about 35 W m−2 for the hourly turbulent fluxes and of about 0.3–0.4 mm d−1 for the daily ET. Successively, two data assimilation schemes based on the ensemble version of the Kalman filter were tested under the hypothesis of absence of information about the root zone depth. The application of a dual state-parameter filter (2EnKF) allows to obtain results very close to the ‘optimal’ ones independently from the value adopted as initial of root zone depth. Moreover, these results were obtained both by assimilating synthetic ‘perfect’ observations and ‘real’ remotely-derived estimations of latent heat flux. The methodology, which combines a coupled energy/water budget model and a dual state-parameter assimilation scheme, seems to be suitable to provide precise estimations of daily ET also when information on root zone depth are absent or not enough accurate.

► Introduction of a data assimilation scheme in a coupled water-energy balance model. ► Assimilation of latent heat fluxes derived from a residual remote sensing-based model. ► Optimal results using state-parameter update scheme based on ensemble Kalman filter. ► Reduction of the uncertainty related to root zone depth parameterisation.