Using fractal analysis in short-term river flow forecasting

This article discusses an approach that improves the accuracy of short-term river runoff forecasting through the use of fractal diagnostics and fractional differential equations. A method for calculating the fractal dimensions of time series is considered, based on a correlation integral, allowing for the selection of the optimal number of phase variables for modeling hydrological processes. It is shown that the use of fractional differential equations corresponds to the actual structure of time series and improves forecast quality, particularly during the rise and fall of spring floods and autumn floods. A comparative analysis of the effectiveness of forecast models is provided. It is confirmed that the proposed approach outperforms traditional methods, increasing the reliability of forecasts. The method is recommended for operational water flow forecasting using modern computer equipment.