Basic Principles of Automatic Calibration of Multi-parameter Models Used in Operational Systems of Flash Flood Forecasting

摘要

This paper initiates a series of three articles on automatic calibration of operational multi-parameter models used for flash flood forecasting in automated regime. The main point expressed in this series is that proper calibration of even a single model for a single basin is not a trivial problem, but in an tomated forecasting systems, when thousands or even millions of basins should be parametrized simultaneously, priorities change. Now the computational efficiency becomes the most critical; an implemented calibration procedure must be fast and efficient rather than giving "the best" parameters yet computationally expensive. The first article of the series contains critical analysis of the "mainstream" in hydrological models calibration and presents basic principles of the Stepwise Line Search algorithm and its modifications, practically feasible and robust parametrization approaches that would be suitable for automated systems used for flash flood forecasting.