Känslighets- och osäkerhetsanalys av parametrar och indata i dagvatten- och recipientmodellen StormTac
Sensitivity analysisUnceartainty analysisStormwater modelRecipient modelMonte Carlo simulationParameters.
Three methods of sensitivity and unceartainty analysis have been applied to the operative stormwater- and recipient model StormTac. The study area is the watershed of lake Flaten in the municipality Salem. StormTac?s submodels for stormwater, pollutant transport and the recipient are cosidired. In the sensitivity assessment, the model parametres and inputs were varied one at a time by a constant percentage according to the ?one at a time? (OAAT) method and the response of the outputs were calculated. It was found that the stormwater- and baseflow were most sensitive to perturbations in the perciptation. Unceartainty analysis using Monte Carlo simulation was performed in two different ways. (1) All model parametres and inputs were included with defined unceartainties and the resulting unceartainty for the target variable was quantified. Thereafter, whith the purpose to estimate the contribution of all the parametres and inputs, the cumulative uncertainty for the target variable, each parameters/inputs unceartainty was omitted one at the time. The most crucial uncertainty for the storm water flow was the runoff coefficient for forestland and the perciptation (i.e the differens between the 90- and 10-percentile for the storm water flow was reduced whith 44 % and 33 % respectively). (2) To identify optimal parameter intervals, the probability for an acceptable value of the target variable was plotted against each parameters value range. The result suggests that for some of the parametres i StormTac, the ranges should be changed.