| Abstract: | To reduce flooding risks and improve urban drainage management, there is a need to increase the forecasting accuracy for rainfall models on small typical urban time and space scales. Increased rainfall forecasting accuracy will in turn improve runoff prediction and thus, prevent flooding hazards, decrease pollution discharge through combined sewers, increase waste water treatment efficiency, etc. For this purpose, we analyzed the parameters of a two-dimensional stochastic advection-diffusion model including a Fourier domain method and an extended Kalman filter algorithm for investigation of motion, shape, size, and intensity distribution of convective rainfall. The resulting set of model parameters (advective velocity, apparent turbulent diffusion, and development/decay of rainfall rate) is used to study convective rainfall variability. It appears that the speed at which the rainfall cell is advected is not dependent on the cell development stage or apparent diffusion. Instead, there is a dependence between the source/sink term and apparent diffusion. This can be explained by the turbulent updraft of warm air which results in large rainfall intensity increase. This strong turbulence results in larger diffusion (and vice versa). The behavior of the model parameters is therefore physically explainable and relevant. The results can be used as first choice of parameter values when modeling convective rainfall over ungaged areas. |
