First step: Construction of Extreme Rainfall timeseries You may compile timeseries of extreme rainfalls from accumulated intervals within the environment of Hydrognomon or import your existing data e.g. from Excel. Case 1: Compile your own timeseries You need a rainfall timeseries of 10-minutes or hourly or daily time step. Then you will compile extreme rainfall timeseries (annual) with intervals as integer multiples of the time step. (e.g. a 10 minutes rainfall timeseries may produce extreme rainfall timeseries of 10 minutes, 20, 30, 40 minutes, intervals of accumulation; an hourly rainfall timeseries may produce extreme rainfall timeseries of 1, 2, 3, hours intervals of accumulation. Define a new timeseries: Choose Series -> New menu A new window appears:
Set desired time step, title, variable (cumulative). Check Time step is strict only if your data have constant time step. If not leave it unchecked. Copy your data from Excel, or your favorite spreadsheet software: Date format should be: yyyy/mm/dd hh:mm for hourly /ten minute data yyyy/mm/dd for daily data yyyy/mm or yyyy/mm/01 for monthly data Select two or three adjacent columns (the third one contains flags). Paste your data to Hydrognomon:
Under Tools -> Regularize step, run this feature in order to make time step strict if it is not: Now use Tools -> Extremes evaluation menu item to show the window bellow: Set the desired time step for IDF analysis (should be Yearly), IDF variable (use either height or intensity). Do not use Hydrological year, this feature is implemented for the Greek weather (rainfalls start from October). Select the desired multiplier, e.g. a multiplier of 24 will evaluate a extreme rainfall timeseries for duration of 24 hours, if the rainfall timeseries is of hourly time step. Repeat the process by using several multipliers, e.g. a set of (1, 3, 6, 60, 144) applied on ten-minute timeseries will produce extreme timeseries of (10min, ½ hour, 1 hour, 24 hours) duration. See time series properties for each new (extreme rainfall) timeseries. Set the comments and title to contain information about the accumulation interval. Finally save the timeseries as files.
Case 2: Import your existing extreme rainfall timeseries You may add new time series with annual (yearly) time step. Copy your data e.g. from Excel to Hydrognomon and save them to file. Second step: IDF evaluation Load (or compile) desired Timeseries First, load the timeseries of extreme rainfalls into Time series data window (I send you these files). Each column is a timeseries containing the annual values of maximum accumulated rainfall (maximum accumulated 5 min, 10 min, 24 h). The selection of the actual number of timeseries /intervals is upon your experience. A sequence that we are using is that of the following example (5 min, 10 min, ½ hour, 1 hour, 2 hours, 6 hours, 12, 24 hours and sometimes 48 hours). Load the timeseries order by accumulation interval (e.g. from 5 min to 24 hours). See timeseries properties: Check that the time step is Annual, and specify accumulation interval in timeseries title and in comments. The use of hydrological year is suggested if the rainfall in your
Country has periodicity from 1 st of October to 30 September of the next year. In other case you may use the ordinary year (1 st of January to 31 December). Choose timeseries for IDF curves evaluation Use the menu under: Tools -> IDF curves Then a new window appears as follows:
Choose each of the timeseries from the left list (Available timeseries), specify the (source) timeseries variable (Height / Intensity), duration (accumulation interval) in either minutes or hours. Set Time resolution if this information is available (e.g. the base absolute minimum time step from which timeseries of greater interval are accumulated, in other cases choose Time resolution unknown ). Leave the Desired amount to 1/3. Finally press the IDF Analysis button, then, the following window appears:
This is the first screen, showing several (multiple) IDF curves at once. By pressing some of the check buttons on the right panel, you may choose which of the IDF curves to display.
Press the Single cuve tab in order to get extensive analysis for a single curve with a specified return period: Set the desired return period in years (T), e.g. 200. Press the Calculate button to get the alpha (a) parameter of the IDF curve. The above function: i(mm/h) := 77.22 / (d + 0.186)^0.792 it is the IDF curve for a return period of 200 years, using the GEV-Max distribution (you may alter the statistical distribution type, see bellow). Press the Calculate button in the Confidence interval for a panel in order to get confidence interval for a by a statistical simulation process (Monte Carlo). Sample limits and confidence interval are displayed respectively showing the probable variation of the main IDF curve with a confidence level of 90, 95 or 99%.
Press the Data tab to get the sample parameters, also the distribution function parameters and the curvature of the IDF curves: Mean value / standard deviation / skewness are the sample characteristics calculated with moment method. LMoment 1-3 are the sample characteristics calculated with the L- Moments method.
Press the Distribution plot tab to show the fit of sample data to the statistical distribution function: You may choose the desired durations to show from the right list (5 min 24 hours).
Statistical Distributions Use the menu Options -> Distribution to set a Statistical Distribution Type. Default is GEV-Max, statistical parameters calculated with L-Moments method and with kappa parameter constant to a specified value (default is 0.15). Choose a distribution from: Exponential distribution Gamma / LogPearson III Gumbel (Max) EV2-Max (Extreme Values 2) GEV-Max (General Extreme Values distribution, k may or not be specified). Pareto Some of the distributions parameters may be calculated with L-Moments method or else with the ordinary moment method. Use the Specify GEV kappa in order to alter the default 0.15 value. The selection of the GEV-Max distribution for the description of extreme rainfall intensities, is based on works of Koutsogiannis, you may contact him for more information. Choose another Paper (distribution/plot) in order to linearize better the sample values on the Distribution plot. Choose Consider time resolution effect in order to multiply rainfall intensity values with a factor of discretization.