Extreme rainfall: Global perspective

The study of rainfall extremes is important for design purposes of flood protection works, in the development of flood risk management plans and in assessing the severity of occurring storm and flood events. Such study unavoidably relies on observational data, which, given the enormous variability of the precipitation process in space and in time, should be local, of the area of interest. While general statistical laws or patterns apply over the globe, the parameters of those laws vary substantially and need local data to be estimated. Because of their global coverage, satellite data can be insightful to show the behavior of precipitation over the globe. However, only ground data (observations from raingages) are reliable enough for rainfall extremes and also have adequate length of archive that allows reliable statistical fitting. The study of the record rainfalls throughout the globe provides some useful information on the behavior of rainfall worldwide. While most of these record events have been registered at tropical areas (with a tendency for grouping in time with highest occurrence frequency in the period 1960-1980), there are record events that have occurred in extratropical areas and exceed, for certain time scales, those that occurred in tropical areas. The record values for different time scales allow the fitting of a curve which indicates that the record rainfall depth increases approximately proportionally to the square root of the time scale. Clearly, however, these record values do not suggest an upper limit of rainfall and are destined to be exceeded, as past record values have already been exceeded. In addition, the very concept of the probable maximum precipitation, which assumes a physical upper limit to precipitation at a site, is demonstrated to be fallacious. The only scientific approach to quantify extreme rainfall is provided by the probability theory. Theoretical arguments and general empirical evidence from many rainfall records worldwide suggest power-law distribution tail of extreme rainfall and favor the Extreme Value type II (EV2) distribution of maxima. The shape parameter of the EV2 distribution appears to vary in a narrow range worldwide. This facilitates fitting of the EV2 distribution and allows its easy implementation in typical engineering tasks such as estimation and prediction of design parameters, including the construction of theoretically consistent ombrian (also known as IDF) curves, which constitute a very important tool for hydrological design and flood severity assessment.