A Comparative Study for Determination of PMP by Some Statistical Methods in Atrak Watershed, Iran

Probable maximum precipitation (PMP) is the maximum amount of rainfall which theoretically is supposed to fall on a specified point or region. In this paper two statistical methods of Hershfield (in two different approaches of standard and revised, as a point prediction) and multi-station (as a regional method) is compared for 33 raingage stations located in the Atrak watershed in north-eastern Iran. Extreme value type 1 and its two alternatives of transformed and cut-off distribution functions, which can handle a finite limit for rainfall as the return period increases without bound, was adopted for the multi-station approach. There were unrealistic results for the standard approach of the Hershfield method. Varions of dimensionless ratio of PMP for revised approach of Hershfield to the maximum amount of daily rainfall was in good harmony with other studies conducted in vast different climatological regions. The regional PMP by multi-station method were smaller than point PMPs for just 7 stations. This result may not be uncorrect, yet we proceeded for the first time in a reverse algorithm and calibrated the behaviour of the limited reduced variate parameter for the region. The resulting PMP was smaller than just one station. So it seems that the regional PMP may be equal to the maximum of the point PMPs of revised Hershfield method. This result is just drawn for only one region and should be confirmed by other researches.