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

Titre : Evaluating the influence of the land surface and air temperature gradient on terrestrial flux estimates derived using satellite earth observation data.

Auteur : Khan, Sameera.

Université de soutenance : University of KwaZulu-Natal

Grade : Master of Science MSc Hydrology 2020

Résumé
One of the most challenging processes of the hydrological cycle to determine accurately especially in arid and semi-arid regions, is actual evapotranspiration (ETa). Numerous approaches are recognised and have been established to account for ETa at various spatial and temporal scales. Satellite earth observation (SEO) methods have been utilised as an alternative to conventional methods to estimate ETa, as they provide estimates over larger geographical scales. Satellite-based ET models have been shown to provide fairly reliable estimates of terrestrial fluxes and ETa. However, these models have the tendency to perform poorly in water stressed environments due to an inherent limitation in their conceptualisation, which relates to the temperature gradient (To-Ta). Due to the dynamic nature of the To-Ta gradient, the study aimed to establish whether the selection of an image based upon the satellite overpass time influences the accuracy of the modelled flux and ETa estimates. For this purpose, the Surface Energy Balance Systems (SEBS) model was implemented using SEO data using Moderate Resolution Imaging Spectroradiometer (MODIS) imagery. The simulated fluxes and ETa were compared against in-situ Eddy Covariance (EC) data, as well as ETa estimates obtained from MOD16 to quantify the influence of To-Ta. The study was undertaken during the 2015 dry period within the Luvuhu and Letaba Water Management Area, situated in the semi-arid north-eastern region of South Africa. This period coincided with a large El Niño induced drought, which provided an ideal opportunity to assess the model’s ability to adequately simulate ETa during conditions of water stress. The results of the investigations undertaken in this study indicated that both the ETTerra and ETAqua largely overestimated ETa when compared to in-situ riparian ETa measurements, yielding a Relative Volume Error (RVE) of -123.04% and -159.41%, respectively. Overall, the SEBS derived MODIS Aqua estimates compared relatively favourably with the in-situ measurements. The aggregated 8-day ETTerra and ETAqua generally overestimated ETa, whilst ETMOD16A2 tends to underestimate ETa during summer months when compared to in-situ ETa. The degree of overestimation of ETMOD16A2 was lower than the aggregated ETTerra and ETAqua estimates. The SEBS results and the MOD16 product emphasized the importance of the satellite overpass times and the limitations that are observed in the SEBS model. The MOD16 product and satellite-based ET models can be used to assist in decision making and can provide long-term data records over remote areas.

Présentation

Version intégrale (2,35 Mb)