Overland flow resistance & flood generation in semi-arid environments: explaining the restrained draining of the rain in Spain

Abstract

Resistance equations developed for pipe flows and open channel flows cannot be applied to model overland flows uncritically. The formulation of these equations employs several assumptions that are specific to the conditions in which they were developed and cannot be universally applied. The hydraulic behaviour of overland flow is distinct from that of pipe and channel flows and can be characterised by a high degree of variability both over space and over time as roughness elements are progressively inundated with increasing depth. A novel methodology of measuring overland flows in the field at a high- resolution permits examination of the interaction between flow variables and surface roughness. Reconstructing the water surface from elevation data and flow extent provides an estimation of the distribution of flow depths and offers a complementary perspective to more conventional approaches. Overland flows are observed to be highly variable both across and between hillslopes. The distribution of flow depths can be modelled using a two-parameter gamma distribution; both parameters show distinct variations with distance downslope and represent the progressive inundation of roughness elements with increasing depth. The flow interacts with soil surface form where it is capable of eroding its bed and the observed slope- independence of rill velocity can be explained by a feedback between flow state (as characterised by the Froude number) and surface roughness. While the existence of this interaction is affected by soil-type, the soil is observed to have little influence on the relationship between surface roughness and overland flow. Resistance is found to be spatially variable; some of this variability could be explained by the classification of areas of similar microtopogiaphy as identified in the field. This classification can be approximated by a thresholded index-based classification and provides a tool for up-scaling to the hillslope scale. Relating roughness to resistance is not straightforward. Complex natural soil surfaces vary in innumerable ways. Traditional roughness measures fall short of providing an adequate description of the complex soil surfaces observed in the field. A variety of alternative measures are developed, each of which captures a different attribute of surface form. These measures are tested to examine their influence on overland flow resistance and a suite of roughness-resistance models is developed which includes the effect of hillslope position to different degrees. Modelled flow resistance can be separated into a constant term and a depth-dependent term and can be easily incorporated into models of hillslope hydrology. This resistance is observed to decline where a hydrological connection, once established, is then maintained. Examination of the concept of hydrological connectivity in a semi-arid context suggests that the interaction between runoff generation and transfer determines not just flood peaks but also total flow amount. It is suggested that flow resistance and hence runoff transfer should be afforded the same detailed consideration as infiltration parameters, i.e. a spatially distributed and variable value (as a function of depth) that can be organised into discrete units akin to those developed for runoff generation. The parameterisation of both infiltration and resistance in this way provides a crucial interaction through the redistribution of soil moisture and runoff over hillslope surfaces. Through this mechanism, the observed complex and nonlinear runoff response to storm events may be explained as these attributes interact with rainfall characteristics and flow network development. Further understanding of this interaction could have practical implications for catchment management and affect the prioritisation of land management decisions.