Extreme runoff events in small catchments the need for a proper modelling strategy for early warning systems

Dieter Gutknecht
Technische Universität Wien, Karlsplatz 13/223, 1040 Vienna, Austria.
Tel. : +43-1-588-01, Fax: +43-1-505-6212,
E-mail: gutknech@bimb.tuwien.ac.at

Extreme runoff events in small catchments are characterised by their random occurrence, the intensity of their runoff response, and by their concomitant circumstances that cause severe damages. Warning of such events must be based on a sound understanding of the processes that occur during these events and a proper modelling approach that accounts for the impossibility to accurately forecast extent, time and location of the events.

To gain better insight into the mechanisms that characterise extreme events, three events that occurred in small catchments in Austria in recent years have been analysed. Investigations comprised the analysis of the scarce hydrometeorological data available, augmented by field observations that were directed towards the collection of auxiliary data based on various detectable traces of the events.

The evidence obtained from theses investigations suggests that such extreme events have a special character. It manifests itself in the occurrence of heavy bursts of rain and hail, in uncommon runoff phenomena such as overland flow channelisation and outbursts of subsurface flow. These outbursts have been described as "fountain-effects" by some observers. Models for extreme events should be designed to describe these phenomena. Difficulties in model building arise due to the fact that (1) measurements that allow parameter estimation are usually not available, (2) each event exhibits an "individual" character, (3) runoff production mechanisms change within the course of the event, and (4) "new" runoff processes occur which cannot be observed under normal (less extreme) conditions.

Model building therefore has to resort to some strategy of extrapolation beyond the common range of observations. To facilitate this and to gain more information on the occurrence of special runoff mechanisms and the conditions under which they occur, a concerted analysis of a number of events under varying conditions seems to be necessary. It is advocated in this paper that the conduction of studies that are directed towards the comparison of observations in different catchments with various runoff producing factors may assist in the proper definition of the dominant factors that control the runoff processes under extreme conditions.

Model building for extreme events may then focus on the proper simulation of the particular factors that dominate extreme runoff behaviour in the catchment of interest. Simulation runs for a set of possible extreme storms and initial conditions may provide a range of possible runoff events that may serve as a basis for evaluating various warning situations.