An experimental comparison of methods for estimating rainfall intensity-duration-frequency relations from fragmentary records

The calculation of intensity- duration -frequency (IDF) relationships for rainfall requires records of rainfall accumulated over periods shorter than 1 day when the application is for small and/or urban catchments. Such records are commonly available only for limited periods, and gaps are frequent due to instrumental failure. With short and fragmentary records, it is essential to make the fullest possible use of the information contained in them. This paper compares a number of approaches by which IN relationships can be obtained in such circumstances. Using, as a basis, a reliable 18-year record of hourly rainfall at Eskdalemuir, Scotland, modified records were created which contained gaps of varying degrees of severity, and OF relationships calculated from each artificially fragmented record were compared with the IN relationships given by the full record. The methods were broadly of two types: (a) one in which the period of modified record was divided into sub-intervals (months or years), with analysis of the maximum rainfalls occurring within the sub-intervals, and with allowances for the possible incompleteness of record during the sub-intervals; (b) one in which a '' peak-over-threshold series '' (series of rainfall events exceeding a given threshold) was calculated for each modified record. Of these options, the method using series of monthly maxima with up to 20% missing data is recommended, being simple to apply and exhibiting small or nonexistent bias and small dispersion of the rainfall estimates. (c) 2007 Elsevier B.V. All rights reserved.