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

Accueil du site → Doctorat → Australie → The development of a stochastic solar radiation model and its application in estimating evaporation

University of Newcastle (2013)

The development of a stochastic solar radiation model and its application in estimating evaporation

Lockart, Natalie A.

Titre : The development of a stochastic solar radiation model and its application in estimating evaporation

Auteur : Lockart, Natalie A.

Université de soutenance : University of Newcastle

Grade : Doctor of Philosophy (PhD) 2013

Global radiation is an important input for evaporation calculations ; however, limited measurements exist. Current models for estimating global radiation are deterministic and do not give an estimate of the errors associated with the predicted radiation amounts. In this thesis, five stochastic models are developed to estimate the mean amount and associated uncertainty of global, direct and diffuse radiation from sunshine duration data. The modelled global radiation is used to estimate evaporation amounts using a variety of models, including the Penman-Monteith model, radiation-based models, and temperature-based models. Evaporation estimates are compared to determine the influence of uncertainty in the global radiation estimate on evaporation amounts. The second part of this thesis deals with the relationship between temperature and evaporation, with implications for anthropogenic climate change studies. The influence of temperature on evaporation amounts is analysed using a combination of empirical evaporation models and a more physically-based planetary boundary layer model. The results indicate that global radiation can be accurately estimated using all of the developed radiation models (average error = 9%), when compared with measured data. The variance of the errors is greater for cloudy days compared with clear days. The diffuse radiation component is best modelled using a quadratic model (average error = 22%). The direct radiation component is best modelled using a linear model (average error = 23%). Two types of regional models are also developed to calculate radiation amounts at any location. These models have only a small loss of accuracy compared to the locally calibrated models. While the variance of the errors for the locally calibrated models is shown to be location dependent, these regional models allow for the model parameters and the error variance to be estimated at any location, as there is a statistically significant relationship between the model and error parameters and latitude. The regional models are also comparable in accuracy to satellite estimates of global radiation. It is also found that the uncertainty in global radiation leads to considerable uncertainty in evaporation rates, up to ±31% for the radiation-based models. The locations with greater uncertainty in the radiation estimate have an associated greater uncertainty in the evaporation estimate. Temperature increases are shown to have minor influences on evaporation rates. Soil moisture is the most dominant influence. Consequently, temperature-based models are shown to lead to unreasonable estimates of evaporation when temperatures are increased


Version intégrale (11 Mb)

Page publiée le 5 février 2018