Initial statistical analyses of hourly maximum data. Report to Defra, project WS194/2/39 (Reservoir Safety - Long Return Period Rainfall). Draft

This report provides details of preliminary analyses of the systematic rainfall database for durations from 1 to 24 hours, and compares the results with those for durations of 1 to 8 days (Jones et al., March 2007). The analysis reported here considers two aspects of the planned statistical work to develop a revised FORGEX methodology to replace that underlying the Flood Estimation Handbook rainfall depth-duration-frequency model: (i) The assessment of possible families of distributions for describing the annual and seasonal maxima for the full range of durations (1 hour to 8 days), known as marginal distributions. (ii) Evaluation of the scaling assumption, by which the standardisation of observed maxima by a site-specific factor is assumed to remove any differences in distribution between sites so that records from a number of raingauges can be pooled together.