Mere on rain rate distributions and extreme value statistics

A new methodology is described for estimating the 5-minute rain rate distribution from yearly 5-minute maximum rain rate data and yearly accumulated rainfall data published by the National Climatic Center for U.S. locations. The method previously described gives the high rain rate portion of the distribution, whereas the extended methodology yields the complete distribution, which is assumed to be approximately lognormal. The three parameters characterizing the lognormal distribution can be calculated by application of the theory of extreme value statistics. The calculated results agree well with the 20-year data. The accuracy of the calculated results is limited by the instability of extreme rain rate data with a finite time base. Two-year rain rate data measured by a tipping bucket rain gauge at Palmetto, Georgia, are used to demonstrate that the time variation of rainfall process obeys a proportionate relationship, supporting the lognormal hypothesis.