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

Accueil du site → Doctorat → Australie → Changes in rainfall-runoff partitioning during decadal drought : influential factors and implications for modelling

University of Melbourne (2016)

Changes in rainfall-runoff partitioning during decadal drought : influential factors and implications for modelling

Saft, Margarita

Titre : Changes in rainfall-runoff partitioning during decadal drought : influential factors and implications for modelling

Auteur : Saft, Margarita

Université de soutenance : University of Melbourne

Grade : Doctor of Philosophy (PhD) 2016

Résumé partiel
Hydrological characterisation and prediction methods typically contain an implicit assumption of stationary hydrologic processes and catchment conditions. The impacts of temporary, or continuous, climate shifts on runoff are usually explored in a context of direct streamflow response to changed climatic forcing (precipitation and temperature) with most studies content to quantify these direct effects. However, the importance of changing within-catchment hydrological dynamics as a cumulative result of change in climatic condition is becoming increasingly recognised. Differences in catchment response are typically studied either as severe but short term impact assessments (e.g. streamflow response to tree die-off during short but intense drought), or as end-point catchment adaptation studies (e.g. catchment intercomparison studies). Both types of studies suggest that if changes in climate affect catchment biophysical characteristics, then the catchment response to climate shifts might not be reliably inferred from the observed response to short-term variability. However, the importance of potential shifts in catchment functioning due to long-term climate variations remains unclear and disputed. Studies synthesising interdecadal climate and streamflow variations with changes in catchment characteristics and processes could fill this gap, but such studies are rare. Partially this is explained by a lack of long-term detailed concurrent records of the relevant variables including vegetation, groundwater storage, soil properties, and humidity, radiation, runoff, and rainfall data. In particular, data for some potentially important sub-surface variables are only available from expensive short-term small-scale experiments, which severely limit their availability.
While these data issues are a major limitation of the current study, I hypothesised that detectable change in catchment functioning following extended downward shifts in rainfall can be observed and distinguished from what was encountered during similarly dry but shorter periods, such as isolated dry years. Observing such differences would prompt questions regarding the primary controls of such shifts in catchment behaviour and whether model performance issues can be related to changes in catchment functioning. This study explores these questions through a combination of statistical data analysis and simulation modelling applied to long-term records from a large set of relatively unimpaired catchments in South-Eastern Australia. Australia is prone to multiyear drought sequences which pose significant challenges to the Australian population and economy, including the recent Millennium Drought (circa 1997-2008). Dry periods like the Millennium Drought are useful real-life cases to explore potential longer-term changes in catchment functioning given that southern Australia, as in many other parts of the world, is expected to have a drier future in which water scarcity will be an ongoing concern for water resources management.
The first part of this thesis investigated whether shifts in rainfall-runoff processes were statistically detectable in South-Eastern Australia during selected prolonged dry periods. Rainfall-runoff processes can be integrated at the interannual timescale by the rainfall-runoff relationship – the functional dependence of annual runoff on annual rainfall. The rainfall-runoff relationship is an extremely useful tool, as it is straightforward to interpret yet comprehensive. Rainfall-runoff relationships inherently include the normal changes in rainfall-runoff partitioning during dry years, enabling direct representation and comparison of rainfall-runoff partitioning during short ( year-long) and long ( decade-long) dry periods. It was found that, for prolonged dry periods, the vast majority of catchments produced less runoff for a given rainfall than the historical rainfall-runoff relationship (i.e. other dry years in the record) suggested. During the Millennium drought this additional reduction in runoff was statistically significant in over half of the catchments. Significant shifts in rainfall-runoff relationships were more often observed in drier, flatter and less forested catchments, whereas features of the dry period climate, including rainfall and temperature anomalies, were not found to be related to the occurrence of rainfall-runoff relationship shifts. Thus, shifts in rainfall-runoff partitioning during prolonged dry periods were governed by factors internal to the catchment rather than by spatial dissimilarities in external forcing, e.g. potential evapotranspiration.


Version intégrale (11,90 Mb)

Page publiée le 12 juillet 2017