Operational Flood Forecasting Research Articles

The use of MOGREPS ensemble rainfall forecasts in operational flood forecasting systems across England and Wales

Abstract. Operational flood forecasting systems share a fundamental challenge: forecast uncertainty which needs to be considered when making a flood warning decision. One way of representing this uncertainty is through employing an ensemble approach. This paper presents research funded by the Environment Agency in which ensemble rainfall forecasts are utilised and tested for operational use. The form of ensemble rainfall forecast used is the Met Office short-range product called MOGREPS. It is tested for operational use within the Environment Agency's National Flood Forecasting System (NFFS) for England and Wales. Currently, the NFFS uses deterministic forecasts only. The operational configuration of the NFFS for Thames Region is extended to trial the use of the new ensemble rainfall forecasts in support of probabilistic flood forecasting. Evaluation includes considering issues of model performance, configuration (how to fit the ensemble forecasts within the current configurations), data volumes, run times and options for displaying probabilistic forecasts. Although ensemble rainfall forecasts available from MOGREPS are not extensive enough to fully verify product performance, it is concluded that their use within current Environment Agency regional flood forecasting systems can provide better information to the forecaster than use of the deterministic forecasts alone. Of note are the small number of false alarms of river flow exceedance generated when using MOGREPS as input and that small flow events are also forecasted rather well, notwithstanding the rather coarse resolution of the MOGREPS grid (24 km) compared to the studied catchments. In addition, it is concluded that, with careful configuration in NFFS, MOGREPS can be used in existing systems without a significant increase in system load.

Validation of satellite rainfall products for operational flood forecasting: the case of the Evros catchment

An extensive validation of two of the most popular and recently upgraded satellite rainfall products, 3B42 and 3B42RT, was performed over the Evros catchment in southeastern Europe using data recorded from January 2000 to April 2009. For conducting this validation study, the Climate Prediction Center's (CPC) ground data were used. The satellite data products were aggregated to daily time series, remapped to spatial resolution of 0.5°, validated against CPC, and intercompared using a variety of statistical indices and coefficients. After the validation process, all three data sets (CPC, 3B42, and 3B42RT) were separately fed in a statistical rainfall–runoff model, in order to predict the five major recorded flood events which occurred in the Evros catchment during the last decade. It has been found that post-calibration with ground data, which is present only in 3B42 product, is a necessity for operational flood forecasting and similar studies conducted in areas at mid-latitudes. Knowledge of rainfall events with small intensities is crucial for estimating the total rainfall height and drastically improves the skill of the satellite product.

Challenges in communicating and using ensembles in operational flood forecasting

AbstractFollowing trends in operational weather forecasting, where ensemble prediction systems (EPS) are now increasingly the norm, flood forecasters are beginning to experiment with using similar ensemble methods. Most of the effort to date has focused on the substantial technical challenges of developing coupled rainfall‐runoff systems to represent the full cascade of uncertainties involved in predicting future flooding. As a consequence much less attention has been given to the communication and eventual use of EPS flood forecasts. Drawing on interviews and other research with operational flood forecasters from across Europe, this paper highlights a number of challenges to communicating and using ensemble flood forecasts operationally. It is shown that operational flood forecasters understand the skill, operational limitations, and informational value of EPS products in a variety of different and sometimes contradictory ways. Despite the efforts of forecasting agencies to design effective ways to communicate EPS forecasts to non‐experts, operational flood forecasters were often skeptical about the ability of forecast recipients to understand or use them appropriately. It is argued that better training and closer contacts between operational flood forecasters and EPS system designers can help ensure the uncertainty represented by EPS forecasts is represented in ways that are most appropriate and meaningful for their intended consumers, but some fundamental political and institutional challenges to using ensembles, such as differing attitudes to false alarms and to responsibility for management of blame in the event of poor or mistaken forecasts are also highlighted. Copyright © 2010 Royal Meteorological Society

Communicating uncertainty in hydro‐meteorological forecasts: mission impossible?

AbstractThere is a common agreement in the scientific community that communicating uncertain hydro‐meteorological forecasts to water managers, civil protection authorities and other stakeholders is far from being a resolved issue. This paper focuses on the communication of uncertain hydrological forecasts to decision‐makers such as operational hydrologists and water managers in charge of flood warning and scenario‐based reservoir operation. Results from case studies conducted together with flood forecasting experts in Europe and operational forecasters from the hydroelectric sector in France are presented. They illustrate some key issues on dealing with probabilistic hydro‐meteorological forecasts and communicating uncertainty in operational flood forecasting. Copyright © 2010 Royal Meteorological Society

Informing operational flood management with ensemble predictions: lessons from Sweden

AbstractThis paper highlights some communicative and institutional challenges to using ensemble prediction systems (EPS) in operational flood forecasting, warning, and civil protection. Focusing in particular on the Swedish experience, as part of the PREVIEW FP6 project, of applying EPS to operational flood forecasting, the paper draws on a wider set of site visits, interviews, and participant observation with flood forecasting centres and civil protection authorities (CPAs) in Sweden and 15 other European states to reflect on the comparative success of Sweden in enabling CPAs to make operational use of EPS for flood risk management. From that experience, the paper identifies four broader lessons for other countries interested in developing the operational capacity to make, communicate, and use EPS for flood forecasting and civil protection. We conclude that effective training and clear communication of EPS, while clearly necessary, are by no means sufficient to ensure effective use of EPS. Attention must also be given to overcoming the institutional obstacles to their use and to identifying operational choices for which EPS is seen to add value rather than uncertainty to operational decision making by CPAs.

Application of generic data assimilation tools (DATools) for flood forecasting purposes

This paper describes the generic data assimilation software tool DATools. DATools can be used as standalone or within Delft-FEWS. DATools is completely configurable via XML configuration. DATools is built up of three components: a Filter, a Stochastic Modeler, and a Stochastic Observer. Configuration of all these three parts is explained in detail. At the moment two data assimilation filters are available within DATools: (1) ensemble Kalman Filter and (2) the residual resampling filter. Results of a twin experiment with both filters with DATtools show similar results as a previous study performed with custom implementations. It is also shown that DATools can function inside Delft-FEWS software used for operational flood forecasting. Applying EnKF to a 1D hydrodynamic SOBEK-RE model of the river Rhine within the operational system FEWS-NL Rhine and Meuse improves the forecasts at the Lobith gaugin station and downstream of Lobith. DATools has been coupled with the HBV-96, SOBEK, and REW models and will be coupled to MODFLOW, Delft-3D, and the geotechnical model MSetlle in the near future. Uncertainty analysis with this tool is also possible and calibration will be added later this year.

Operational flood-forecasting in the Piemonte region – development and verification of a fully distributed physically-oriented hydrological model

Abstract. A hydrological model for real time flood forecasting to Civil Protection services requires reliability and rapidity. At present, computational capabilities overcome the rapidity needs even when a fully distributed hydrological model is adopted for a large river catchment as the Upper Po river basin closed at Ponte Becca (nearly 40 000 km2). This approach allows simulating the whole domain and obtaining the responses of large as well as of medium and little sized sub-catchments. The FEST-WB hydrological model (Mancini, 1990; Montaldo et al., 2007; Rabuffetti et al., 2008) is implemented. The calibration and verification activities are based on more than 100 flood events, occurred along the main tributaries of the Po river in the period 2000–2003. More than 300 meteorological stations are used to obtain the forcing fields, 10 cross sections with continuous and reliable discharge time series are used for calibration while verification is performed on about 40 monitored cross sections. Furthermore meteorological forecasting models are used to force the hydrological model with Quantitative Precipitation Forecasts (QPFs) for 36 h horizon in "operational setting" experiments. Particular care is devoted to understanding how QPF affects the accuracy of the Quantitative Discharge Forecasts (QDFs) and to assessing the QDF uncertainty impact on the warning system reliability. Results are presented either in terms of QDF and of warning issues highlighting the importance of an "operational based" verification approach.

A data based mechanistic approach to nonlinear flood routing and adaptive flood level forecasting

Operational flood forecasting requires accurate forecasts with a suitable lead time, in order to be able to issue appropriate warnings and take appropriate emergency actions. Recent improvements in both flood plain characterization and computational capabilities have made the use of distributed flood inundation models more common. However, problems remain with the application of such models. There are still uncertainties associated with the identifiability of parameters; with the computational burden of calculating distributed estimates of predictive uncertainty; and with the adaptive use of such models for operational, real-time flood inundation forecasting. Moreover, the application of distributed models is complex, costly and requires high degrees of skill. This paper presents an alternative to distributed inundation models for real-time flood forecasting that provides fast and accurate, medium to short-term forecasts. The Data Based Mechanistic (DBM) methodology exploits a State Dependent Parameter (SDP) modelling approach to derive a nonlinear dependence between the water levels measured at gauging stations along the river. The transformation of water levels depends on the relative geometry of the channel cross-sections, without the need to apply rating curve transformations to the discharge. The relationship obtained is used to transform water levels as an input to a linear, on-line, real-time and adaptive stochastic DBM model. The approach provides an estimate of the prediction uncertainties, including allowing for heterescadasticity of the multi-step-ahead forecasting errors. The approach is illustrated using an 80 km reach of the River Severn, in the UK.

Propagation of uncertainty from rainfall to runoff: A case study with a stochastic rainfall generator

We explore the impact of uncertainties in the spatial–temporal distribution of rainfall on the prediction of peak discharge in a typical mountain basin. To this end, we use a stochastic generator previously developed for rainfall downscaling, and we estimate the basin response by adopting a semi-distributed hydrological model. The results of the analysis provide information on the minimum rainfall resolution needed for operational flood forecasting, and confirm the sensitivity of peak discharge estimates to errors in the determination of the power spectrum of the precipitation field.

Discharge assimilation in a distributed flood forecasting model

Abstract. In the field of operational flood forecasting, uncertainties linked to hydrological forecast are often crucial. In this work, data assimilation techniques are employed to improve hydrological variable estimates coming from numerical simulations using all the available real-time water level measurements. The proposed assimilation scheme, a classical Kalman filter extension to non-linear systems, is applied in a rainfall-runoff distributed model based on the SCS-CN approach. The complex hydrological system of the Toce river basin is studied, a mountainous catchment of about 1500 km2 in the Italian alps, through the development of a prototype available for operational use. For the considered flood event, the assimilation scheme is stable, even when available observations show gaps or outliers. It allows significant improvements in the simulation results, in particular when the focus is addressed to the peak.

Flood forecasting using medium-range probabilistic weather prediction

Abstract. Following the developments in short- and medium-range weather forecasting over the last decade, operational flood forecasting also appears to show a shift from a so-called single solution or 'best guess' deterministic approach towards a probabilistic approach based on ensemble techniques. While this probabilistic approach is now more or less common practice and well established in the meteorological community, operational flood forecasters have only started to look for ways to interpret and mitigate for end-users the prediction products obtained by combining so-called Ensemble Prediction Systems (EPS) of Numerical Weather Prediction (NWP) models with rainfall-runoff models. This paper presents initial results obtained by combining deterministic and EPS hindcasts of the global NWP model of the European Centre for Medium-Range Weather Forecasts (ECMWF) with the large-scale hydrological model LISFLOOD for two historic flood events: the river Meuse flood in January 1995 and the river Odra flood in July 1997. In addition, a possible way to interpret the obtained ensemble based stream flow prediction is proposed.

NOAA'S ADVANCED HYDROLOGIC PREDICTION SERVICE

Abstract The National Oceanic and Atmospheric Administration (NOAA) National Weather Service (NWS) Advanced Hydrologic Prediction Service (AHPS) program was established to meet our nation's need for more precise flash-flood forecast information. AHPS uses NOAA investments in remote sensing, precipitation forecasts, climate predictions, data automation, hydrologic science, and operational forecast system technologies. AHPS establishes a pathway for the infusion of new verified science and technology, and expands the use of NWS climate, weather, and water analyses and information products. State-of-the-art science is used for improved operational forecasting of floods, and drought conditions. The objective is to deliver more precise forecast information over greater temporal scales (hours, days, and months) and to depict the magnitude and certainty of occurrence for events ranging from droughts to floods. The AHPS program improves flash-flood forecasts, and provides ensemble streamflow forecasting and flood...

Prediction of a Flash Flood in Complex Terrain. Part II: A Comparison of Flood Discharge Simulations Using Rainfall Input from Radar, a Dynamic Model, and an Automated Algorithmic System

Three techniques were employed for the estimation and prediction of precipitation from a thunderstorm that produced a flash flood in the Buffalo Creek watershed located in the mountainous Front Range near Denver, Colorado, on 12 July 1996. The techniques included 1) quantitative precipitation estimation using the National Weather Service’s Weather Surveillance Radar-1988 Doppler and the National Center for Atmospheric Research’s S-band, dual-polarization radars, 2) quantitative precipitation forecasting utilizing a dynamic model, and 3) quantitative precipitation forecasting using an automated algorithmic system for tracking thunderstorms. Rainfall data provided by these various techniques at short timescales (6 min) and at fine spatial resolutions (150 m to 2 km) served as input to a distributed-parameter hydrologic model for analysis of the flash flood. The quantitative precipitation estimates from the weather radar demonstrated their ability to aid in simulating a watershed’s response to precipitation forcing from small-scale, convective weather in complex terrain. That is, with the radar-based quantitative precipitation estimates employed as input, the simulated peak discharge was similar to that estimated. The dynamic model showed the most promise in providing a significant forecast lead time for this flash-flood event. The algorithmic system did not show as much skill in comparison with the dynamic model in providing precipitation forcing to the hydrologic model. The discharge forecasts based on the dynamic-model and algorithmic-system inputs point to the need to improve the ability to forecast convective storms, especially if models such as these eventually are to be used in operational flood forecasting.

Applying soft computing approaches to river level forecasting

This paper assesses one of many potential enhancements to conventional flood forecasting that can be achieved through the use of soft computing technologies. A methodology is outlined in which the forecasting data set is split into subsets before training with a series of neural networks. These networks are then recombined via a rule-based fuzzy logic model that has been optimized using a genetic algorithm. The methodology is demonstrated using historical time series data from the Ouse River catchment in northern England. The model forecasts are assessed on global performance statistics and on a more specific flood-related evaluation measure, and they are compared to benchmarks from a statistical model and naive predictions. The overall results indicate that this methodology may provide a well performing, low-cost solution, which may be readily integrated into existing operational flood forecasting and warning systems.

Statistical and hydrodynamic models for the operational forecasting of floods in the Venice Lagoon

The risk of flooding in Venice has increased strongly since the beginning of the century. To reduce the damage to the city and the negative impact on the activities in the lagoon, an accurate flood warning system is necessary. This system will also be fundamental during the construction and for the efficient operation of storm surge barriers covering the three existing inlets of the lagoon. In this context new operational statistical and hydrodynamic models have been developed. Forecast winds and pressure fields which constitute basic information for the warning system have been obtained through an ad hoc Limited Area Meteorological model. It has been demonstrated that, provided that this information is available on an operational basis, the implementation of a flood warning system for Venice using the models developed is feasible. The statistical model, which is based on a multiple regression technique, extends the forecasting range of the model presently in operation at the Centro Previsioni e Segnalazioni Maree del Comune di Venezia, from 3 hrs up to 24 hrs, and presents good accuracy (estimated mean absolute errors smaller than 10 cm) for short-term forecasts up to 9 hrs. The hydrodynamic model includes all the physical processes important for the simulation of water levels and currents in coastal and marine environments. The model set-up adopted covers the entire Adriatic Sea, with a grid spacing of 6 km. Special attention has been given to the positioning of the open boundary and to the correct reproduction of the main free oscillation of the Adriatic, which is responsible for the possible recurrence of flooding after the main storm has passed. The inclusion of this model in a flood warning system is mainly intended for long-term forecasts (> 24 hrs), and can typically be used to forecast up to 3–4 days ahead, with an estimated mean absolute error smaller than 20 cm.

Using the DPFT approach to improve flash flood forecasting models

DPFT is a lumped approach for operational flash flood forecasting, based on the unit hydrograph. Using a multi-event alternating iterative algorithm, it identifies a robust and stable average transfer function and a consistent set of effective rainfall series associated with each event at the same time. This key ingredient allows an objective calibration of different loss functions, relating gauged precipitation and effective rainfall. A case study based on an operational French basin (545 km2) is presented. Three lumped production functions have been calibrated and compared. The results show that more elaborate models of loss functions must be proposed, and some possible directions for this are pointed out.

Rainfall measurements by weather radar: Applications to hydrology

Surface precipitation measurements are of prime importance in studies of hydrology, climatology, and meteorology. These measurements can be improved by using both weather radar and conventional rain and snow gages. Recent advances in digital radar data management make possible improved precipitation estimates for hydrological applications such as flood forecasting. A project undertaken in Alberta provides areal precipitation data for operational flood forecasting. Maps depicting rainfall intensities are generated at specified time intervals from the Alberta Research Council's S band weather radar. Transmitted 150 km by telecopier to the Flow Forecasting Branch, Alberta Environment, the real‐time radar‐derived areal maps of precipitation are used with point surface observations to specify more precisely the precipitation events that lead to floods. During precipitation events, weather radar data provide an indication of the applicability of point surface observations, leading to better operational flow forecasts. In one example the radar showed that high rainfall amounts observed at a conventional surface rain gage fell over a localized area. Using only surface gage information, a flood forecast likely would have been issued. However, radar intelligence suggested otherwise, so the forecast was not issued. The potential benefit of weather radar to hydrologists is assessed using rainfall within a representative subbasin. The precipitation was estimated from radar observations and from an independent gage network used by operational hydrologists. Throughout a stormy 6‐day period, the accumulated precipitation for a subbasin estimated by the two methods agreed to within 15%. Shortcomings of typical rain gage networks in assessment of areal rainfall are evident. In one case, weather radar data revealed that although rain fell on the network gages, the subbasin itself received little rain, so the gage estimate was high. In another example, the weather radar demonstrated that rain fell in the subbasin but missed the network gages. The spatial variability of precipitation, evident in weather radar observations, is generally not observed by gage networks. Therefore radar measurements of rainfall intensity and rainfall amount can be of considerable benefit to operational and applied hydrologists.