Highlights of the HydAlp project in the last three months
Real-time forecasts of daily runoff
These are planned for the snowmelt period 1999 in the Austrian, Swiss and Swedish basins. The forecast periods will start in April and last for about two months. Links for real-time or near-real time recording, transmission and processing of input data for the hydrological models , including EO data, hydromet data, and numerical weather forecasts, have been established and tested.
ERS SAR and NOAA-AVHRR will be the primary sources for EO input in near-real time, with Radarsat and IRS WiFS as backup or for test purposes. The time required from the ERS SAR image acquisition to the final snow cover product was 6 hours for day-time images and 15 hours for night-time images (with data reception at Neustrelitz, SAR processing at D-PAF, SAR -PRI image transmission via the Internet, and classification at IMGI or SCEOS). For AVHRR (with reception at Dundee) the time required from the satellite overflight to the snow cover product is only 4 hours. This is well within the requirements for real-time forecasting, because the time scale of snow cover changes in the three basins is much longer. Preparations for real-time forecasts included the expansion of the catchments being studied in Switzerland and Austria. In Austria, the customer Verbund expressed the interest in forecasts for three further drainage basins above reservoirs in the same region as the established test basin (Persal) and preparations for these additional forecasts are proceeding.
In Switzerland the original test basin (Rhein-Felsberg) will be replaced by the Engadin basin for the real-time activities due the involvement of a hydropower company as potential customer (which was not the case in the original test basin).
Model calculations of daily runoff were carried out with the two runoff models SRM and HBV for three snowmelt periods for each basin, using EO input. These model runs were essential for calibrating or tuning the model parameters for the various basins and enabled the testing of the model processing flowlines. In the Austrian test basin the runoff simulations were carried out with SRM and HBV, using ERS-SAR and Radarsat as the main EO-data source, with some Landsat TM imagery. In Switzerland and Sweden snow maps derived from AVHRR data were utilised. The runoff calculations in Switzerland are based on SRM, and in Sweden on HBV.
In the United Kingdom EO-derived snow maps have not been used for runoff calculations since no forecasting activities have been planned for this basin.
In the Austrian test basin, simulation runs were carried out with different numbers of satellites images to study the impact of various time sequences of remote sensing input as a basis for the cost/benefit analysis.
Test of HBV with EO data.
In preparation for the real-time forecasting activities, HBV has been used in Sweden for three snowmelt periods to study the impact of remote sensing input on forecasts. To date, test were carried out with snow maps derived from AVHRR, both for short-range forecasts (10 days) and for long-range forecasts.
The EO data are used to correct the snow cover which is estimated in the model from meteorological data. For the three test periods the EO input results in improvements of the long-range forecasts, but for the 10-day forecasts the impact of EO input is ambiguous.
Snow Cover Products and Intercomparison.
Digital snow maps of the test basins were derived from comprehensive EO data sets using the automatic and semi-automatic procedures which had been developed during the project. The snow maps were used as input for the runoff simulation activities. The sensors employed include SAR (ERS, Radarsat), HROI (Landsat-5 TM, SPOT-3 and -4 HRV), and MROI (NOAA-AVHRR, RESURS-O1).
Snow maps from SAR/HROI/MROI(AVHRR) were compared in the Austrian and Swiss basins. Systematic differences between the various sensors were found which result from the different spectral capabilities and spatial resolutions. These differences must be taken into account if time sequences from different sensors are used.
Accessible through EWSE, the HydAlp WWW site at MLURI and sites at three of the partner institutes contain a range of material of interest for potential customers as well as for remote sensing experts. Information is provided on the project objectives, preliminary results, and ongoing research activities. The pages also include descriptions of the test basins and a simplified version of the SRM model with pre-calculated options for runoff simulations.
Return to Hydalp Home Page