There are two main schools of estimation of extreme flood characteristics in the world hydrological practice. The first approach is based on fitting a statistical distribution to available measurements of flood peak discharges and extrapolating this distribution to estimate the floods of needed low exceedance probabilities. The second one uses the concept of probable maximum flood. Neither approach practically utilizes available meteorological observations (that contain important information on

more »

... possible variations of runoff generation) and both are based on implicit assumptions that the physical mechanisms of runoff generation do not depend on the magnitudes of the water inputs and land use changes. To overcome these shortcomings, a new technique based on coupling the Monte Carlo simulation of meteorological inputs with application of the detailed physically-based model of runoff generation processes is suggested. The paper illustrates the implementation of this technique for estimation of the extreme flood characteristics for the Seim River basin (the catchment area is 7,460 km 2 ). The model of runoff generation is based on the finite-element schematization of river basins and includes the description of snow cover formation and snowmelt, freezing and thawing of soil, vertical soil moisture transfer and infiltration, overland, as well as and channel flow. The Monte-Carlo simulation is based on stochastic models of daily precipitation series, daily air temperature and daily air humidity deficit (for continuous simulation during autumn-winter-spring seasons) or distributions of snow water equivalent, depth of frozen soil, and soil moisture content before snowmelt (for simulation during only spring period). The calculated exceedance probabilities of the flood peak discharge have been compared with ones calculated using long-term runoff data. As an example of the application of the developed technique, changes of the Seim River flood runoff characteristics resulting from changes of basin land use are given.