The influence of catchment characteristics on river flow variability

Hydrology is yet to fully understand the role that catchment characteristics have in determining a river’s response to precipitation variability. This thesis assesses the influence that catchment characteristics have on modulating a river’s response to changes in precipitation throughout the UK. Central to this aim is the concept of the precipitation- to-flow relationship (the transformation of precipitation into river flow), which is characterised using the Variogram, a way of indexing temporal dependence (i.e. the average relationship between river flow on a given day and river flow on the previous days). Firstly, 116 catchments were grouped into four clusters, based on the shape of their variogram, which significantly differed in their catchment characteristics demonstrating that catchment characteristics control how, on average, precipitation is transformed into river flow. Furthermore, over 70% of un-gauged catchments could be clustered correctly using information about their soil type, slope and the percentage of arable land. Secondly, a new method which identifies the changes in the variogram parameters over 5-year overlapping moving windows was developed to investigate temporal changes in the variogram parameters. This method was successfully demonstrated to detect changes in multiple aspects of artificially perturbed river flow time series (e.g. seasonality, linear changes and variability). On average >70% of the variability in the catchment variogram parameters was explained by the precipitation characteristics, although there was large variability between catchments. Finally, the influence that the catchment characteristics have on the temporal changes in the variogram parameters was analysed, demonstrating that rivers in relatively impermeable upland catchments have a relationship with precipitation which is closer to linear and less variable than lowland, permeable catchments. This thesis contributes significant new knowledge that can be used for both assessing how individual catchments are likely to respond to projected changes in precipitation and in informing data transfer to un-gauged catchments.