Frequency Analysis Of Rainfall Runoff Quality In Urbanizing Catchment

Rainfall runoff has recently been considered as a major pollution source in manyrapidly urbanizing areas. For the random distribution of the runoff pollution, afrequency analysis method is used to analyze the runoff pollution. This can help tounderstand and analyze the overall condition of rainfall runoff pollution, risk controlof water quality, as well as support to select the most appropriate solution for runoffpollution problems. In this study, a runoff quality frequency analysis method based onlong-term simulation was introduced. The objectives of this study are1) to estimatethe frequency distribution of Event Mean Concentration/Event mean Pollutant Loadper unit area (EMC/EPL);2) to evaluate the effect of initial rainwater interception onthe frequency distribution;3) to evaluate the impact of seasonal variations andAntecedent Dry Period (ADP) on the frequency distribution. The main contents ofthis paper are described as follows:(1) Established the runoff quality frequency analysis method and then applied itin an urbanizing Shiyan River catchment in Shenzhen.(2)Calibrated and verified parameters of rainfall runoff model usingNondominated Sorting Genetic Aalgorithm (NSGA-II) and artificial calibration.Uncertainty analysis was developed using Generalized Likelihood UncertaintyEstimation (GLUE) method and the measured data are selected from onlinemonitoring system.(3) The Shiyan River catchment was used as a case study. Based on the historydata of1688rainfall events, the frequency distribution was estimated using Matlabprogramming. Furthermore, the impact of seasonal variations and ADP on thefrequency distribution was evaluated.(4) Based on the result of (3), the impact of initial rainwater interception on thefrequency distribution as well as the impact of initial rainwater interception on the frequency distribution of different seasons and ADP were also evaluated,The following conclusions can be arrived at:(1) The results of parameters calibration and validation indicated: theNash-Sutcliffe (NS) coefficients of IHACRES and water quality model are0.83and0.84, respectively. This result demonstrated that the simulated data match themeasured data closely. The results of uncertainty analysis indicated: the confidenceinterval of EMC is [-22.98%,+23.74%] and the confidence interval of EPL is[-32.70%,+41.42%] at90%confidence level. There lies in great uncertainty in themodel simulation.(2) The results of case study indicated: the EMC/EPL showed considerablevariations between rainfall events. The average of EMC is300mg/L, the coefficient ofvariation(Cv) is0.76.The average of EPL is3t/km2, Cv is1.67.(3) The results of the effect of seasonal variations and ADP on the frequencydistribution indicated: at high EMC/EPL, the cumulative frequency increases with theincrease of ADP; the cumulative frequency of the period before flood season is higherthan the other two. At low EMC/EPL, the cumulative frequency for different ADPand seasons are basically the same.(4) The results of the effect of initial rainwater interception on the frequencydistribution indicated:88.3%,32.6%,18.6%and8.5%of rainfall events will haveEMC (COD) greater than40mg/L when0mm,5mm,10mm and20mm of rainfallrunoff is intercepted, respectively.After interception, water quality can still not meet the quality standard if therainfall has long ADP and those in the period before flood season.Some othertreaments are still needed for better management.(5) The method introduced in this study can help to determine the entirepollution of the receiving waters, analysis the risk of water quality control and supportto select the most appropriate solution for runoff pollution problems.