Which probability distributions does the IDF Curve Tool support?
In short: The IDF Curve Tool offers eight frequency-analysis methods: the non-parametric Empirical Plotting Position (EPP) method and seven parametric distributions—Gumbel, Gamma, Exponential, Log-Normal, Weibull, Generalized Extreme Value (GEV), and Log-Pearson Type III (LP3).
Empirical and parametric choices
EPP uses Weibull plotting positions and is useful for comparing fitted curves with observed ranks. Because it is tied to the sample, it cannot extrapolate beyond approximately the observed record length. Parametric distributions estimate a mathematical model and can calculate larger return-period quantiles, but those extrapolations depend strongly on tail behavior and sample size.
Gumbel is the classical two-parameter extreme-value model. GEV adds a shape parameter and can represent different upper-tail behaviors. LP3 fits a Pearson Type III distribution to base-10 logarithms and is widely used in U.S. flood-frequency practice. The other supported distributions provide alternative shapes for positive rainfall intensity data.
How to compare fits
The Goodness-of-Fit table compares the parametric distributions for a selected duration using log-likelihood, AIC, Delta AIC, the Kolmogorov–Smirnov statistic, and a KS p-value. Use those rankings together with Q-Q plots, hydrologic judgment, sample size, and sensitivity of the design result; no single metric guarantees the correct tail model.
Review the distribution formulas and the tutorial comparison table.