Long-term ensemble forecast of snowmelt inflow into the Cheboksary reservoir under the differently constructed weather scenarios

Abstract

Abstract.Development and verification of the long-term ensemble forecast of snowmelt water inflow into the Cheboksary reservoirone of the eleven major river reservoirs of the Volga-Kama Reservoir Cascade, are described. The used forecasting procedure is based on a combination of semi-distributed hydrological model ECOMAG that allows for ensemble calculation 10 of inflow hydrographs with two types of weather ensembles for the lead-time period: (1) observed weather data constructed on the basis of the ESP methodology and (2) synthetic weather data simulated by a weather generator. Ensemble forecasts represented both in deterministic and probability forms were verified by producing hindcasts of water inflow into the Cheboksary reservoir for 35 seasons in 1982 -2016 from the 1st of April for 3 months ahead. Four inflow characteristics were forecasted: volume and maximum discharge of the inflow, as well as amount of days with inflow above two specified 15 thresholds. Results of operational forecast of the recent freshet of 2017 are additionally presented. We found that statistical distributions of the forecasted ensembles calculated under both types of weather scenarios are very close to each other.However, confidence intervals of the forecast statistics is much larger for the ESP-based ensemble. Sensitivity of the forecast performance to changes in the forecast horizon and issue date is analyzed and so-called "forecastability maps" are presented as useful tool for such analysis. 20