Climate change is likely to impact the seasonality of precipitation and flows in Mediterranean mountain catchments. This will be manifested by increasing rainfall intensity, and runoff magnitude frequency, even in arid areas particularly the Moroccan High Atlas, where extreme flood events are still difficult to predict, as this type of data is rare to be collected in a short time step. Flash floods are generally poorly observed hydrological extremes in regions where measuring stations are sparse and poorly distributed spatially, and characterized by high interannual precipitation variability. Considering the important influence of the uncertainties of the calibration parameters of the hydrological models; this makes it difficult to predict and understand their seasonal behavior, such as in our case in the Zat and Ghdat watersheds. The climatology of the High Atlas watersheds is characterized by a seasonal contrast, where the climate is dry and hot in summer and may affect the erodibility of soils, with infrequent flash flood events, followed by a stormy autumn causing extreme flooding due to the impermeable nature of the soils affected by the summer heat, then mixed winter precipitation characterized by two different types: liquid downstream and snowy upstream with a moderately stable runoff, followed by the spring where snowmelt represents a large fraction of the streamflow in the study during the spring which causes flooding as the river flow is higher. The aim of this study is to understand the behavior of the Zat and Ghdat watersheds and the seasonality of runoff and precipitation, while taking into account the uncertainty of the model used. The analysis has been developed by using instantaneous rainfall and runoff data at 10 minutes time scale, from the 2 measuring stations located downstream of the watersheds (Sidi Rahal and Taferiat) respectively during the period from 01/09/2011 to 31/08/2018. More than 100 episodes were simulated and calibrated on HEC-HMS software. Then a satellite approach was elaborated in order to quantify the snowfall season that causing flooding. However, to analyze the uncertainty of the calibration parameters we applied the Probability Distribution Function and the Monte Carlo simulations to analyze the Curve Number and the Time of Concentration uncertainty; the results showed that observed and simulated hydrographs were highly correlated for both basins. Moreover, the model's performance was evaluated by obtaining a Nash coefficient ranging from 61. 9% to 90% for calibration part, and that the runoff flooding parameters uncertainty are due to the snow melt process.