A Study On Deriving Storm Intensity Formula

Extreme raifalls occur frequently in recent years due to accelerated climate warming and urbanization. This always leads to poor drainage or even internal waterlogging. Such a situation gives rise to higher demands on drainage system construction. Rainstorm intensity formula is the basis of urban rain drainage system design and directly influences project investment and urban safety. However, many big cities in our country adopt formula established in 1980s, which causes many problems during practical application. Even worse, some small and medium-sized cities have got no formula enacted at all. So they can only reference to intensity formula of neighboring big cities. This is obviously unsuitable. Wubu county is just in such dilemma which makes it urgent to work out intensity formula reflecting the local rainstorm regularity in order to guide plan and design of drainage system.This paper first sorted out 3 groups of i-t-P data table on basis of original precipitation data from 1995-2014 provided by Wubu weather bureau through comparative analysis using annual maximum method and non-annual maximum method, frequency adjustment using three frequency curves. Then use 4 methods solving non-linear parametric equation to calculate 12 groups of rainstorm intensity formula. These formulas were tested by Root mean square error together with calculated value and actual value error rate to compare and select the best frequency distribution lines and storm intensity root and branch formula.Compare the two sampling method, annual maximum method has the advantage of sampling simpleness, data availability, good independence, high frequency rainfall intensity reasonability, wide-range application than non-annual maximum method. So sampling process utilized annual maximum method in this paper. Samples selected by this method were adjusted by three frequency curves. Pearson III type distribution curve best fits the data, Gumbel distribution curve comes second place and index distribution curve performed the worst fitting effect by comparing fitting error.The fitting effects revealed that the order of excellence of 4 optimization algorithm under same distribution curve follow a sequence of marquardt method, Gauss-newton method, golden section method and the least square method. The order of excellence of 3 distribution curve under same optimization algorithm were Gumbel distribution curve, index distribution curve and pearson III type distribution curve which is not in consistence with frequency adjustment. So regularity characteristic, frequency adjustment, fitting error of evaluated formula and so on of original precipitation data should be taken into consideration when selecting optimum frequency adjustment model.It was preliminarily determined to select root-mean-square error less than 0.05 mm/min as the standard to judge the evaluated formula.7 groups of intensity formula were selected and then measured by error rate between calculated value and actual value of precipitation. Best storm intensity formula was that sampling using annual maximum method, adjusting frequency by Gumbel frequency curve and evaluating parameters by marquardt method. The determined formula of Wubu county was i=6.5850(1+0.9641gP)/(t+12.7312)0.7090 (2-20a) and i=6.5050(1+0.90171gP)/(t+14.2082)0.6871(2～100a) fitted absolute mean square error were 0.0284 and 0.0358 respectively when return period was 2-20a and 2-100a, and comparative mean square error were 3.58% and 4.02% which meets the demand of less than 0.05mm/min and 5% by the code. In consideration of actual situation of Wubu county, return period of drainage design was 2-20a and that for urban flood control project design was 2-100a. When return period of detailed project design matches that of branch formula, it is suggested to use branch formula to calculate the designing storm intensity as its accuracy was higher.