Dwyer, I.J., Reed, D.W. 1995a. Allowance for discretization in hydrological and environmental risk estimation. Report No. 123, Institute of Hydrology, Wallingford, 45 pp.
For full report: Click here
Disasters, natural or otherwise, are often precipitated by extremes of an environmental variable such as rainfall or wind speed. In order to make provision for alleviating the effects of such events, reliable risk estimation is required. This is usually obtained by measuring the relevant variable and analysing the data in an extreme value context. The data, however, inevitably contain some form of discretization (such as averaging over discrete time steps) which can degrade the risk assessment.
The report examines the effect of data discretization upon the estimation of period maxima, A correction model is proposed for converting fixed maxima, derived from discretized data, to true maxima as would be derived from continuous data. The model can be applied over a range of event durations.
The report takes a particular interest in rainfall extremes, for which discretization effects can be marked. The results suggest that previously reported correction factors are too low. The analysis of hourly rainfall data from various locations has enabled correction factors for daily rainfall to be discerned according to climate regime. Wind speed and air temperature extremes are also examined. A relationship between correction factors and effective fractal dimension is demonstrated, providing a means for deriving correction factors for other environmental variables.
The report concludes by consolidating the results into practical recommendations.