Delft-FEWS

modules

• connecting Delft-FEWS to external data sources
• validating, interpolating and transforming data
• starting with simple forecasting modelling
• expanding to more advanced modelling
• using some of the advanced forecasting tools
• viewing resuls and disseminating forecasts

connecting Delft-FEWS to external data sources

Of paramount importance in an operational flood forecasting system is an efficient connection to external data sources. Delft-FEWS provides an import module that allows import of on-line meteorological and hydrological data from external databases.

These data include for example time series obtained from telemetry systems such as observed water levels, observed precipitation, but also meteorological forecast data, radar data, etc. Data are imported using standard interchange formats, such as XML, GRIB (for Numerical Weather Predictions) and ASCII.

The import of external data also supports ensemble weather predictions, such as those provided by the European Centre for Medium Range Weather Forecasts (ECMWF).

validating, interpolating and transforming data

Particular emphasis is placed in Delft-FEWS on quality checking of data obtained from external sources, with extensive validation options to check data and serial interpolation (gap-filling) avilities to complete data series where required.

Data hierarchy options allow alternative data sources to be used as a fallback in the forecasting process should the nominal sources be unavailable. The use of these options will ensure the continuity of the flood forecasting process, even if available real-time data is incomplete or inconsistent.

Another set of utilities is available to provide support in transforming data with disparate spatial and temporal scales.

This includes geo-statistical spatial interpolation to derive for example areal weighted precipitation from spatially distributed point sources, or from spatial data such as radar data and numerical weather prediction models.

The data transformation utilities also include methods for temporal aggregation and disaggregation, evaluation of simple equations and typical hydrological functions such as stage-discharge relationships and evaporation calculations.

starting with simple forecasting modelling

Delft-FEWS provides a number of modules to rapidly establish a simple forecasting system where hydrological and/or hydraulic models are not available or under development.

This includes a correlation module that allows generation of forecasts for a downstream location based on the correlation of events at that location and a suitable upstream location.
Another simple module is the lookup table module, which derives warning levels on the basis of a heuristic combinations of critical conditions in the imported data.

Once more advanced models become available these simple modules may be replaced, or retained either to compare to the hydrological and hydraulic model, or as a backup should these models fail.

expanding to more advanced modelling

The philosophy of Delft-FEWS is to provide an open system to allow a wide range of existing forecasting models to be used.

This concept is supported by the provision of a general adapter module, which supports communication to external modules using an open XML based published interface, effectively allowing “plugging-in” of any forecasting model. An adapter between the native module data formats and the open XML interface is typically required. Such adapters are already available to support a wide range of hydraulic and hydrological models.

The great advantage of this open interface is that existing hydrological and hydraulic models and modeling capabilities can easily be integrated in the forecasting system, without the need for expensive re-modelling using a specific model.

The general adapter is currently being extended to support the OpenMI standards for linking models developed under the HarmonIT project.

using some of the advanced forecasting tools

Delft-FEWS provides a number of advanced forecasting tools that can be used in improving and assessing the quality of forecasts.

This includes generic data assimilation methods such as an ARMA based statistical error correction module. A what-if scenario tool can be used to quickly establish the effects of different input scenarios.

A performance module is also provided to assess the accuracy of forecasting models used. The performance module also provides a powerful set of analysis tools to assist post event analysis. A calibration tool provided as a part of Delft-FEWS can help tune a model optimally within the forecasting environment.

viewing resuls and disseminating forecasts

The interactive map display allows geographic navigation, while icons are used to give the forecaster rapid insight in any warning levels being reached.

The time series display can be used to explore data further, or edit input data when necessary. Additional insight in the dynamics of a flood event may be gained through the animated longitudinal profile and flood map displays.

Forecast results can be disseminated through configurable HTML formatted reports, allowing easy communication to relevant authorities and public through intranet and internet.