Towards extended range stream flow drought predictions in Switzerland using reforecast data and snow water equivalent climatologies: Homogenisation of a snow water equivalent climatology
Event details
| Date | 01.04.2014 |
| Hour | 16:15 › 17:15 |
| Speaker | Dr Stefanie Jörg-Hess, Mountain Hydrology and Mass Movements, WSL Birmensdorf |
| Location | |
| Category | Conferences - Seminars |
Abstract:
Several drought events in recent years have highlighted the vulnerability of Switzerland to hydrological droughts. To improve risk management and preparedness we work on forecasts of stream-flow droughts. First, we explored the performance of streamflow simulations with the hydrological model PREVAH forced with meteorological information from numerical weather prediction. The discharge simulations with COSMO-LEPS as input performed poorly with focus on low flow. However, when replacing precipitation with observed or bias corrected COSMO-LEPS precipitation and using the other meteorological variables from COSMO-LEPS, the performance of the low-flow simulations was comparable to simulations forced with observations of all variables. For operational applications, a probabilistic reforecast VarEPS from ECMWF was used as meteorological input. These simulations showed skill up to a lead time of 32 days for the Thur and other catchments. In a second step we wanted to improve the initial conditions for the hydrological model. Runoff was not assimilated because of the presence of hydropower in most catchments in Switzerland. Thus we focused on the improvement of the snow storage information. We developed a homogenised snow water equivalent (SWE) dataset for the period 1971-2009. This dataset was used as SWE information at initialisation of the hydrological model. The SWE dataset and calibration method will be presented in detail in this presentation.
When trying to derive consistent time series for SWE products, changing data availability represents a considerable challenge. In an attempt to improve the product consistency of a gridded SWE dataset, we tested the potential of quantile mapping to compensate for mapping errors in a longer SWE product consisting of fewer snow stations, compared to a shorter SWE product consisting of more snow stations. The SWE calibration was done with quantile mapping by making seasonal, regional and altitude-related distinctions. The calibrated SWE maps were accurate when averaged over regions and time periods, where the mean error was approximately zero. However, deviations were observed at single grid cells and years. When we looked at three different regions in more detail, we found that the calibration had the largest effect in the region with the highest proportion of catchment areas above 2000 m a.s.l. The added value of the calibrated SWE climatology is illustrated with practical examples: the verification of a hydrological model, the estimation of snow resources anomalies and the predictability of runoff through SWE.
Short biography:
After the master in Environmental sciences at the ETH Zürich I started my PhD at the Swiss Federal research institute WSL in the Hydrological Forecasts Group. My research interests are the hydrological cycle during drought events and hydrological modelling. The PhD project is part of DROUGHT-CH which is founded by the NRP61.
Several drought events in recent years have highlighted the vulnerability of Switzerland to hydrological droughts. To improve risk management and preparedness we work on forecasts of stream-flow droughts. First, we explored the performance of streamflow simulations with the hydrological model PREVAH forced with meteorological information from numerical weather prediction. The discharge simulations with COSMO-LEPS as input performed poorly with focus on low flow. However, when replacing precipitation with observed or bias corrected COSMO-LEPS precipitation and using the other meteorological variables from COSMO-LEPS, the performance of the low-flow simulations was comparable to simulations forced with observations of all variables. For operational applications, a probabilistic reforecast VarEPS from ECMWF was used as meteorological input. These simulations showed skill up to a lead time of 32 days for the Thur and other catchments. In a second step we wanted to improve the initial conditions for the hydrological model. Runoff was not assimilated because of the presence of hydropower in most catchments in Switzerland. Thus we focused on the improvement of the snow storage information. We developed a homogenised snow water equivalent (SWE) dataset for the period 1971-2009. This dataset was used as SWE information at initialisation of the hydrological model. The SWE dataset and calibration method will be presented in detail in this presentation.
When trying to derive consistent time series for SWE products, changing data availability represents a considerable challenge. In an attempt to improve the product consistency of a gridded SWE dataset, we tested the potential of quantile mapping to compensate for mapping errors in a longer SWE product consisting of fewer snow stations, compared to a shorter SWE product consisting of more snow stations. The SWE calibration was done with quantile mapping by making seasonal, regional and altitude-related distinctions. The calibrated SWE maps were accurate when averaged over regions and time periods, where the mean error was approximately zero. However, deviations were observed at single grid cells and years. When we looked at three different regions in more detail, we found that the calibration had the largest effect in the region with the highest proportion of catchment areas above 2000 m a.s.l. The added value of the calibrated SWE climatology is illustrated with practical examples: the verification of a hydrological model, the estimation of snow resources anomalies and the predictability of runoff through SWE.
Short biography:
After the master in Environmental sciences at the ETH Zürich I started my PhD at the Swiss Federal research institute WSL in the Hydrological Forecasts Group. My research interests are the hydrological cycle during drought events and hydrological modelling. The PhD project is part of DROUGHT-CH which is founded by the NRP61.
Practical information
- General public
- Free
- This event is internal
Organizer
- EESS - IIE
Contact
- Dr Hendrik Huwald, CRYOS