Flood frequency analysis for nonstationary annual peak records in an urban drainage basin

摘要: Abstract Flood frequency analysis in urban watersheds is complicated by nonstationarities of annual peak records associated with land use change and evolving stormwater infrastructure. In this study, a framework for flood developed based on the Generalized Additive Models Location, Scale Shape parameters (GAMLSS), tool modeling time series under nonstationary conditions. GAMLSS applied to maximum discharge Little Sugar Creek, highly urbanized watershed which drains core Charlotte, North Carolina. It shown that able describe variability mean variance selected parametric distribution as smooth function via cubic splines. analyses Creek (at drainage area 110 km 2 ) show flow 0.01-annual probability (corresponding 100-year stationary conditions) over 83-year record has ranged from minimum unit 2.1 m 3 s - 1 5.1 . An alternative characterization can be made examining estimated return interval would have an exceedance 0.01 assumption stationarity ( 3.2 Under conditions, definitions period should adapted. model, ranges value more than 5000 years 1957 almost 8 present (2007). The also used examine links between population trends frequency, well rainfall. These are future decades.