Short name: Delft-FEWS
Long name: Delft Flood Early Warning System
Platform type: Delft-FEWS is a platform that integrates data and models.
Delft-FEWS allows a wide range of existing forecasting models to be used, supported by a module called the General Adapter, which communicates to external models through an open XML or NetCDF based interface, effectively allowing “plugging-in” of any forecasting model, algorithm or script.
Numerous specialized adapters are already freely available for a wide range of hydraulic and hydrological models, including HEC-RAS, HEC-HMS, ISIS, Mike11, OpenDA, OpenStreams, RTC Tools, SOBEK, DELFT-3D, Flood Modeller Pro and HBV.
Usage: Short to long range stream flow forecasting, flood forecasting and early warning. Delft-FEWS was designed to support the flood forecasting process and due to its flexible and modular structure, it is also very well suited to support day-to-day operational management, real-time control and forecasting and warning in other disciplines, like water quality, reservoir management, irrigation, hydropower, navigation, groundwater, drought and dike strength monitoring.
Background: Delft-FEWS is a platform developed already 20 years ago initially as a flood forecasting and warning system. Essentially it is a sophisticated collection of modules designed for building a hydrological forecasting system customised to the specific requirements. The Delft-FEWS business model is based on shared investments by the Delft-FEWS community. Deltares acts on the community’s behalf to implement new features while maintaining the highest standards of software integrity through rigorous design, development, testing and release procedures. This allows users to actively contribute to the knowledge base of Delft-FEWS, which currently bundles decades of hydrological knowledge and software development, and stay up to date with the international best practices in operational (flood) forecasting. The Delft-FEWS community is extremely important in the exchange of knowledge and ideas and the sustainable development of hydrological forecasting science.
Delft-FEWS has published a vision for 2025 which is implemented through yearly roadmaps.
Input data: Supports over 175 import formats.
Delft-FEWS provides import modules that allow data to be imported from a variety of external sources such as web services, external databases and many different file formats. These data include for example time series obtained from telemetry systems such as water levels and precipitation, but also water quality sample data, meteorological forecast data, radar data and numerical weather predictions. The import of external data also supports ensemble weather predictions.
Input format: Many ++
Data are imported using standard interchange formats, such as CSV, XML, GRIB, Hdf5,NetCDF as well as some 150+ propriety data formats.
Data Latency: Workflows that digest (import) data can be scheduled at any interval. Data availability at the provider´s end determines the optimal choice of interval. Usually import runs do not take long so the (external) data can be visualised within minutes after data provision and start of the (triggered / scheduled) import task.
Model Output Time-Series: Models running in a Delft-FEWS system are often producing data in their native format. Once a model run is ready, the timeseries are converted to XML or NetCDF before being imported back into Delft-FEWS for further (post) processing and visualization. Once the model output is in the Delft-FEWS central database, it can be exported just like any other timeseries. See Platform Output TimeSeries and Data Download Format.
Platform Output Time-Series: Apart from many visualization options within the platform, output can be generated in the form of web reports containing graphs, tables and maps. Scalar and gridded time series can be exported in a large variety of formats (including ensemble traces and statistics). Delft-FEWS web services can be used for direct interaction with the timeseries data. Locations, levels and timing of threshold crossings are part of the visualization and output options of the platform and so are scores of forecast performance.
Data Download Format: Able to export data in more than 60 export formats. Forecast results can be disseminated through configurable HTML formatted reports, and variety of time series formats includingCSV, XML and NetCDF-CF.
Optimization: The Delft-FEWS data validation and transformation libraries offers extensive performance optimisations to handle heterogeneous datasets. Calibration display is available in standalone mode. OpenDA data assimilation toolbox which includes a variety of algorithms for parameter calibration such as Dud, sparse dud, simplex, powell, etc. (openda.org)
A performance module is also provided to assess the accuracy of forecasting models used and provides a powerful set of analysis tools to assist post event analysis.
Data Preparation: Delft-FEWS supports quality checking and pre-processing of the imported data using extensive data validation and transformation libraries. Data validation includes checks on extreme values, rate of change, spatial homogeneity and trend detection. Serial interpolation (gap filling) for example is available to complete data series where required. Data hierarchy options allow alternative data sources to be used as a fall-back, ensuring continuity of the forecasting process, even if available data is incomplete or inconsistent. Another set of utilities is available for transforming data with disparate spatial and temporal scales. This includes for example spatial interpolation to derive areal-weighted precipitation from spatially-distributed point sources, or from gridded spatial data such as radar data and numerical weather prediction models. Data transformation utilities include typical hydrological functions such as stage-discharge relationships and evaporation calculations, but also allow the user to define their own mathematical equations or even write small data transformation scripts that include an extensive library of GIS operations. Data validation and transformation modules can be automated, but can also be operated interactively while reviewing data in dedicated displays.
Data Assimilation: Delft-FEWS provides a number of advanced forecasting tools that can be used in assessing and improving the quality of forecasts. This includes generic data assimilation methods such as an ARMA (Auto Regressive Moving Average) based error correction module and the openDA data assimilation toolbox which includes a variety of algorithms for state updating and parameter calibration such as Ensemble Kalman Filtering and Particle Filtering. Manual data assimilation and general user interaction with model parameters, states and time series is offered through the Delft-FEWS displays. A performance module is also provided to assess the accuracy of forecasting models used and provides a powerful set of analysis tools to assist post event analysis.
Ensemble: The import of external data also supports ensemble weather predictions now routinely made available by meteorological forecasting agencies.
Uncertainity: OpenDA toolbox, which includes uncertainty analysis methods; Bayesian Model Averaging; Quantile Regression. Delft-FEWS can produce probabilistic forecasts and communicate it as ensemble traces, statistical summaries or as a summary of information on the probability of exceeding selected threashold.
Model Output Statistics: The interface can be configured to support additional functionality, such as on-the-fly statistics of time series, or manual editing.
Model Calibration/Validation: Calibration Display available in stand alone version. Using calibration and archive tools such as graphs of observed and simulated data, goodness-of-fit indices. Archive of historical data, simulations, forecast products and other products can be used for post-event analysis, model calibration, reviews, and performance analysis.
Inventory Models Availability: Delft-FEWS can integrate any kind of model as long as there is a so-called model adapter available. Numerous specialized adapters are already freely available for a wide range of hydraulic and hydrological models, including HEC-RAS, HEC-HMS, ISIS, Mike11, OpenDA, OpenStreams, RTC Tools, SOBEK, DELFT-3D, Flood Modeller Pro and HBV. The full list can be found here:
https://publicwiki.deltares.nl/display/FEWSDOC/Models+linked+to+Delft-FEWS
Additional Models Availability: See Inventory Models Availability
Installation Resources: Delft-FEWS components for client-server forecasting systems, web services and long term archive can be installed on premise (local machines), on virtual machines in remote data centers, or in cloud environments. Components can all be installed on one the mentioned options or in any hybrid set-up.
Installation: Automatic installation packages available (RPMs, Powershell install scripts) or well documented step-by-step instructions for manual deployment. Basic System Admin knowledge required.
Access: All installed Delft-FEWS platforms are owned by the client. They can allow remote access (e.g. for their employees). From Deltares, no Platform-as-a-Service is offered (yet). Forecasts-as-a-Service is being considered and piloted. The Delft-FEWS software is enabling these services.
Platform Hosting: No hosting (of complete production forecasting systems for a specific client) is currently offered by Deltares.
User Education: BSc
Degree of Difficulty: 3
GIS Expertise: No, basic GIS knowledge is convenient for navigating maps, interpreting spatial plots (legends) and configuring spatial data transformation. Familiarity with common GIS fils types (shape, GeoJSON) and Web Mapping Services is considered basic GIS knowledge.
IT Expertise: Yes, average IT skills are required, supplemented with the Delft-FEWS system admin course.
Hardware Requirements: Normal PC, Minimum 1 GB RAM, 2GB recommended. PC, Client/Server, Virtual Server, Cloud. Delft-FEWS is a fully scalable system. It can be run as a standalone manually driven forecasting system operating on a laptop, but can also be deployed as a fully automated distributed Client-Server application, that can be extended to the cloud. The Client-Server platform supports operational forecasting with job scheduling, remote administration, email/text message alerts, automated failover, multiple remote clients, and continuous data ingest and product export.
https://publicwiki.deltares.nl/display/FEWSDOC/Delft-FEWS+Hardware+and+software+requirements
Operating System: MS Windows, Linux.
Supported Browser: Delft-FEWS has an embedded, JCEF/Chromium based browser available for displaying (white-listed) external URLs or internal web pages. When not whitelisted, the default, external web browser will be used for opening the page.
Language of Core Code: JavaTM technology, and is fully configurable by the user through open XML and CSV formatted configuration files. By configuring and automating your imports, data handling routines, model runs and displays, optimal structure and control over forecast processes can be obtained.
Open Source: Closed source. Delft-FEWS can be considered Open Software. Customizable with a closed-core. Open Software statement can be found here: https://oss.deltares.nl/documents/2355249/2574220/Delft-FEWS_softwarestrategy_v3.pdf/5111d3b9-c36b-2b97-1a02-486ecccf4771?t=1602491682874
Last Update and Version: Delft-FEWS 2023.01
Next Update and Version: A new version of Delft-FEWS is released twice a year (Spring, *-01, Autumn *-02)
Active Development Community: Yes. Developments are driven by Delft-FEWS user community and led by Deltares.
Download URL: https://oss.deltares.nl/web/delft-fews/downloads
Free to Download and Use: Yes
Language of Software Interface: English, Dutch, German, Spanish, Chinese, Indonesian, French and Vietnamese
Online Support URL: https://publicwiki.deltares.nl/display/FEWSDOC/Home
Training Material URL (including example data sets): https://oss.deltares.nl/web/delft-fews/downloads
Language of Trainings: English
Guidance Material URL (including case studies and benchmarking of performance/speed): https://oss.deltares.nl/web/delft-fews/community
Language of Guidance: English
References: https://www.sciencedirect.com/science/article/pii/S1364815212002083
Owner:
Deltares
https://www.deltares.nl/en/
General address Deltares
P.O. Box 177
2600 MH Delft
The Netherlands
tel: +31 (0)88 335 8273 (08:00 and 17:30 hours)
fax: +31 (0)88 335 8582
Developer
Delft-FEWS Product Management (fews-pm@deltares.nl)