Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Australie → 2014 → Evapotranspiration : application, scaling and uncertainty

University of New South Wales (UNSW) 2014

Evapotranspiration : application, scaling and uncertainty

Ershadi Esmaeilabadi, Ali

Titre : Evapotranspiration : application, scaling and uncertainty

Auteur : Ershadi Esmaeilabadi, Ali

Université de soutenance : University of New South Wales (UNSW)

Grade : Doctor of Philosophy (PhD) 2014

Résumé
Evapotranspiration (ET) represents one of the key components of the terrestrial hydrological cycle. In this research, a number of different aspects of ET estimation are studied in order to better understand the causes of uncertainty in flux estimation and also to identify potential approaches to improve ET prediction. First, the differences amongst commonly used ET models are evaluated and discussed. This analysis includes : i) an intercomparison of models over various land surface conditions ; and ii) a study on the role of model structure and resistance parameterization on flux estimation. Results from these investigations provide guidance into making an informed choice on an ET model. It is shown that a modified form of the Priestley-Taylor model outperforms more complex Penman-Monteith and energy balance type estimation approaches. Moreover, it is identified that for Penman-Monteith type models, the resistance parameterization is more influential than model structure.Second, the effects of spatial scaling on ET estimation are studied by undertaking an in-depth evaluation of the Surface Energy Balance System (SEBS) models response to the aggregation of input forcing : a necessary step in undertaking regional to global scale estimation of surface heat fluxes. Aggregation of input forcing showed limited effect on the land surface temperature (LST) and available energy, but reduced ET at the image scale by up to 15% and at the pixel scale by up to 50%. The main source of such errors was determined to be due to the role of the roughness parameterization.Finally, a Bayesian technique is developed for the explicit quantification of uncertainties in process-based models. In a case study using the SEBS model, the Bayesian technique illustrated that the main reason for discrepancy between simulated and observed sensible heat fluxes was a result of errors in the local observations of the LST. This was related back to footprint difference between the in-situ LST sensor and the eddy covariance system used for independent flux estimation.Overall, results of this research provide new insights into the process of evapotranspiration, and offer new ideas for future research aimed at improved realization of the evapotranspiration process.

Mots clés : Upscaling ; Latent Heat Flux ; Energy Balance ; Bayesian inference

Présentation

Version intégrale (9,60 Mb)

Page publiée le 29 novembre 2014, mise à jour le 2 juin 2017