Effect of Record Length and Recent Past Events on Extreme Precipitation Analysis

Risk of possible damage to important hydraulic structures needs to be kept at the minimum by suitably modelling environmental parameters like rainfall for extreme values of desired return period. Efficient estimation of such meteorological extremes depends on the historical records available at the site of interest. Each of the sampled data is essentially a signal from the natural system and in any statistical analysis uncertainty about the underlying phenomenon gets reduced with increase in the record length. In this article, the effect of record length on the extreme value estimates of daily rainfall at Colaba and Santacruz using theoretically appropriate generalized extreme value (GEV) model has been analysed. The study indicates that estimates for different return periods get stabilized with the increase in the length of record. Data analysis-based recent past records at Colaba give comparatively higher estimates which can possibly be attributed to increased variation and observance of more number of extreme events during the recent past. The heavy rainfall of 944.2 mm recorded at Santacruz on 26 July 2005 has shown an extraordinary effect on extreme value estimates. A possibility of temporal dependence in the series requires further studies by parameterization of trend in the GEV model.