Many types of agricultural conservation and hydraulic engineering structures are designed to accommodate rainfall-produced floods of a certain magnitude in order to function safely at a given level of risk. Models of flood attributes such as peak discharges and flood volumes require inputs of so-called “extreme” rainfall that may be expected to occur only very infrequently, e.g. with recurrence intervals of 2 or 10 or 50 or even 100 years. Climate change is expected to change rainfall regimes in a manner that is likely to lead to increases in the intensity and frequency of extreme rainfall events of both short durations (minutes to hours) and longer durations of 1 day or 2 or 3 consecutive days.
Because reliable estimates of flood magnitudes and frequencies based on long time series of good quality observed streamflow data are seldom available at a location of interest, rainfall-based methods of flood frequency estimations are usually resorted to. This requires a probabilistic approach to analysing rainfall or simulated streamflow for design purposes. The terms “design rainfall” and “design streamflow” are then used to describe the
- depth (magnitude, in mm or m3) of rainfall or streamflow, for a critical
- duration (1 day or 2 or more consecutive days), which depends on the size of the catchment, for a desired
- frequency of recurrence (e.g. statistically once in 2 or 10 or 20 or 50 or even once in 100 years, depending on the size and economic importance of the structure). This is commonly referred to as the “return period”. A return period of, say, 10 years implies a statistical probability of recurrence once in 10 years or 10 times in 100 years, and not that it will recur regularly every 10 years.
Historical estimates of design rainfalls of long duration (1:10 and 1:50 year return periods) were computed using the 50-year daily rainfall datasets of each of the Quinaries covering the region, as described in the section Background: Climate Change Modelling and in Schulze et al. (2010).
Future projections of variables based on rainfall are not presented here for reasons given in the section on Mean Annual Precipitation. The future maps of long duration design rainfall can be included once there is a greater model agreement and confidence in the results.