Stochastic approach for stormwater tank assessment

Stormwater detention tanks are frequently used as a structural measure for mitigating impacts of combined and separated sewer overflows from urban catchments into receiving water bodies. The dimension of those capacity depends on climate patterns and sewer system behaviours and can be estimated by using continue simulation or probabilistic approaches. The last ones are based on a stochastic model able to fit the statistical pattern of observed rainfall records and on an urban hydrology model to transform rainfall in sewer discharge. A key issue is the identification of the optimal structure of the stochastic rainfall model. In this paper a conceptual stochastic model of rainfall is proposed in which storms occur in a stochastic process, where each storm has a random lifetime. Each event is separated by the definition of an inter event time (IET), that is the duration of the dry period between consecutive storms. Point processes are frequently applied where rainfall events are represented through the occurrence of rectangular pulses, which are governed by specific descriptors (volume, duration and IET). The paper focuses on the analytical derivation using the derived distribution approach of the probability distribution of the number of overflows as a function of stormwater capacity. This number is used as a performance indicator of the tank effectiveness as a structural measure for water quality and quantity mitigation in natural water bodies (sea, rivers, lakes etc.). The proposed approach is applied to 21 different sites in Italy, all located in Campania Region, South of Italy. In greater detail, for all locations, Pareto-2 and Gamma probability distributions for rainfall depth and duration provided better fit varying the Inter Event Time Definition (IETD) than the Weibull and exponential model, which were widely used in previous studies. Furthermore it turned out that the Fatigue Life or the Log-Gamma probability distribution provided the best fit for the IET. Finally, the extension of the proposed model to different hydrological conditions is discussed.