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

Accueil du site → Doctorat → Royaume-Uni → 2018 → The effect of land surface hydrological process representation on drought prediction, at a range of spatio-temporal scales

University of Reading (2018)

The effect of land surface hydrological process representation on drought prediction, at a range of spatio-temporal scales

Wright, Azin

Titre : The effect of land surface hydrological process representation on drought prediction, at a range of spatio-temporal scales.

Auteur : Wright, Azin

Université de soutenance : University of Reading

Grade : Doctor of Philosophy (PhD) 2018

Droughts can have a devastating impact on livelihoods. In order to predict and understand drought processes in the earth system, land surface models along with agro/hydrological models are currently used. This study focuses on improving our understanding of how best to model agricultural and soil moisture drought. The specific aim is to assess the sensitivity and configurations of agrohydrological (the SWAP model) and land surface models (JULES and CH/CTESSEL) regarding their drought prediction performance over the UK and Europe. In Chapter 4 the sensitivity of the SWAP agrohydrological model and CHTESSEL land surface model drought-related variables to soil hydraulic parameters (van Genuchten) were analysed by the Sobol’ method. The most important soil hydraulic parameters in the SWAP model depended on the soil texture and soil moisture conditions. The model was found to be most sensitive to van Genuchten a and saturated water content (0s ). In Chapter 5, it was found that the modelled soil moisture of CHTESSEL was most sensitive to the saturated hydraulic conductivity (Ks ) and the saturated water content (0s )- The uncertainty ranges of Ks and 0s were used to produce a drought probability map of Europe for August 2003. In Chapter 6, the performance of the SWAP model under drought conditions for seven crop types was compared. The comparison was based on the modelled agricultural drought indices of Soil Moisture Deficit Index (SMDI), where SWAP was run with dynamic rooting depth, and Evapotranspiration Deficit Index (ETDI). The results showed that the SWAP model simulation of crops’ response to drought was sensitive to the length of crop roots, transpiration reduction coefficient, LAI and soil type. In Chapter 7, CHTESSEL, CTESSEL, JULES with Brooks and Corey hydraulic configuration (JULESBC) and JULES with van Genuchten hydraulic configuration (JULESVG) were compared regarding three types of key approaches that affects model performance in prediction of drought : 1) Stomata/ conductance : The A-gs approach (in CTESSEL) resulted in more water use efficiency within the model in comparison to the Jarvis approach (in CHTESSEL) ; 2) Hydraulic parameterisation : The Brooks and Corey scheme (in JULESBC) resulted in longer and more intense droughts compared to the van Genuchten scheme (in JULESVG) ; 3) implementation of soil water availability and model type : The difference in performance between CTESSEL and JULESVG was greater than comparisons between the stomata ! conductance or hydraulic schemes for each separate model. This work has demonstrated that there are clear parameter sensitivities in both agrohydrological and land surface models, and the values assigned to the soil hydraulic and plant parameters. The sensitivities should be considered carefully when using these models for drought forecasting and prediction. In addition the choice of stomata ! conductance approach and soil hydraulic algorithm has a large effect on drought prediction. This effect is even greater when comparing between different land surface models. Overall this work has demonstrated the importance of undertaking parameter, algorithm and model choice sensitivity studies in order to appreciate the uncertainties in drought prediction and be confident in forecasting future droughts.


Version intégrale

Page publiée le 21 janvier 2023