1 / 48

Applications in hydrology and real-time flood forecasting

Applications in hydrology and real-time flood forecasting. Florian Pappenberger ( florian.pappenberger@ecmwf.int ) www.ecmwf.int/staff/florian_pappenberger European Centre for Medium-Range Weather Forecasts. Flooding – a global challenge. Flooding – a global challenge.

twyla
Download Presentation

Applications in hydrology and real-time flood forecasting

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Applications in hydrology and real-time flood forecasting Florian Pappenberger (florian.pappenberger@ecmwf.int) www.ecmwf.int/staff/florian_pappenberger European Centre for Medium-Range Weather Forecasts

  2. Flooding – a global challenge

  3. Flooding – a global challenge

  4. Flooding – an individual disasters

  5. Causes of flooding • snowmelt runoff; • rainfall; • ice jams and other obstructions; • catastrophic outbursts; • coastal storms (tsunamis, cyclones, hurricanes); • urban stormwater runoff; • dam failure (or the failure of some other hydraulic structure). Etc …

  6. Mr. Flood Catchment Characteristics Event Characteristics Aim & Purpose Communication & Warning Skill & Resources Fit for purpose In flood forecasting there is no one size fits all. Integration of different systems and methods is a major challenge. Any system does not have to be perfect but suitable.

  7. Forecasting chain using Ensemble Prediction Systems of Numerical Weather Predictions Hydrology EPS Hydraulics Pre-Processor Warning Post-Processor For a full summary see Ensemble Flood Forecasting: A Review, Cloke H.L., and Pappenberger, F., forthcoming, Journal of Flood Risk Management

  8. EPS in hydrology – who uses it? ECMWF Technical Memoranda 574, Cloke, H.L. and F. Pappenberger, 2008, “Operational flood forecasting: a review of ensemble techniques” Also published in Journal of Hydrology (details see my for webpage)

  9. EPS in hydrology – who uses it? • ~14 centres use EPS • Majority in Europe • Majority uses ECMWF inputs (or derivates) + UK

  10. EPS in hydrology – who uses it? Hungary Finland Bangladesh JRC Sweden France

  11. EPS - Why ensembles??? Why EPS in flood forecasting??? • Allows to take account of uncertainty in this boundary condition and eventually for a ‘better’ forecast (see other presentations) • “the use of meteorological ensembles to produce sets of hydrological predictions increased the capability to issue flood warnings” (Balint et al., 2006, p.67) • “The hydrological ensemble predictions have greater skills than deterministic ones”. (Roulin, 2007) • “The use of EPS in hydrological forecasting proved to be of great added value to a flood early warning system, as the EPS-based forecasts showed in general higher skill than the deterministic-based ones”. (Bartholmes et al., 2008) • Cloke and Pappenberger (2009, Journal of Hudrology) list a large number of case studies and long term evaluations showing the added value of EPS

  12. EPS in hydrology - A global review • Over 35 case studies • 3 long term studies • 2 case studies using multiple Ensemble forecasts

  13. EPS in hydrology – who uses it? • Most case studies indicate that there is added value in using EPS in comparison to deterministic forecasts • A few are convinced of the potential, but are cautious about the added value – mostly quoting the inaccuracy of precipitation predictions as reasons • Most case studies have severe weaknesses in the analysis: • No report of false alarm • Qualitative statements only (sometimes only loosely linked to the displayed figures) • Comparison only done against proxy observations • decision support or communication of these forecasts to end-users is not adequately considered

  14. EPS how is it used? Severe Flood Warning Flood Warning Flood Watch

  15. EPS – Example of Systems 232 weather forecasts in grand ensemble Pappenberger et al., 2008, GRL

  16. Steps in a forecast chain Hydrology EPS Hydraulics Pre-Processor Warning Post-Processor For a full summary see Ensemble Flood Forecasting: A Review, Cloke H.L., and Pappenberger, F., forthcoming, Journal of Flood Risk Management

  17. Do I have to do all???? The steps in the forecast chain will compensate for each other. 3 Hydrology EPS Hydraulics 4 Pre-Processor 5 Warning ? Suggested order of importance for medium range, plain dominated flood forecasting 1 2 Post-Processor For a full summary see Ensemble Flood Forecasting: A Review, Cloke H.L., and Pappenberger, F., forthcoming, Journal of Flood Risk Management

  18. Warning & Decisions • Bartholmes et al. (2008) investigated several options for a warning system based on EPS: • Number of Ensembles above threshold • Persistency • Combining different forecasts to derive warning decisions • The results indicate that it is possible to derive binary decisions. The quality of such a system can be enhanced by using multiple EPS (see TIGGE case study later)

  19. Warning and Decision • On average ~40% of people affected in Europe receive a warning. Only a small proportion acts on the warning Scientific Study by Sebastien Norbert (King’s College London) investigates the communication of uncertainties by EPS, some quotes: “But these people simply don’t understand, they don’t need this information. I don’t care what is the probability. [They say] give me exact figure!! It really doesn’t operate on uncertainties. I said there is uncertainty of 10%. What does it, what do you mean, 10% uncertainty? Give me the figure. I want exact forecast!” (24/10/08). “Because we can't predict exact . . . but the problem is how to show this uncertainty. In between this range at this time with 75%. This would be the next step. But this is too much information I think that the public cannot use.”

  20. Cost of Damage 1000-3500 m3/s Flood Depth EPS based forecasts can provide ranges that become meaningless for a decision maker In practice: Decision making with uncertainty? & velocity

  21. The Energy gained through hydropower is directly • proportional to the height of the water. Lowering the water level for flood protection needs to be done several days in advance and represents an important economic loss for the company. Cost/loss based decisions… … difficult to apply in decision making • In many countries firefighters are volunteers that are called from regular jobs to help with flood protection. They can only be called when flooding is certain.

  22. EPS in hydrology – key challenges? TM574 postulates 6 key challenges • Key challenge 1: improve current NWPs • Key challenge 2: Understand the total uncertainties in the system • Key challenge 3: Analyse more case studies • Key challenge 4: Install more enough computer power • Key challenge 5: Learning how to use it in an operational setting • Key challenge 6: Communicating uncertainty and probabilistic forecasts

  23. Case studies • European Flood Alert System • TIGGE

  24. Support for EFAS • EFAS was launched 2003 at the Joint Research Centre (IT) • Financial support from different DG’s in the European Commission and the European Parliament. • 5 Member States detached experts to the JRC for 4 years (AT, CZ, DE, HU, SK) • EFAS team consists of 10-12 hydrologists, meteorologists, GIS experts, Web-development, and Programmers

  25. EFAS main objectives Added value Novel information • Catchment based information • Lead times up to 10-15 days • Probabilistic information • Operationally targeted research • -Comparable information across Europe • Tool for international aid assistance during crisis International Civil Protection National water authorities

  26. EFAS 2 - Model 1- Data (obs and NWP) LISFLOOD EFAS user interface 3 – Products Real-time data (EU-FLOOD-GIS/ETN-R) Historical Data Static Data DATA EFAS partner network Alert email NHS parter netwok Europ. Data Layers Meteo -Data 4 – Alerts Expert Knowledge of Member States 5 – Possible actions

  27. EFAS - Data – Weather forecasts • Deterministic • DWD - global model, 40 km, 7 days) • DWD - EU, 7 km, 3 days) • ECMWF -global, 25 km, 15 days) • Ensembles • ECMWF VAREPS (global, staggered time and spatial resolution, [40 km, 1-10 days], [80km, 11-15 days], 51 members) • COSMO-LEPS (EU, 7 km, 5 days, 16 members) DWD ECMWF

  28. EFAS- Ensembles - let’s calculate • 51 ECMWF members • 16 Cosmo-LEPS (only 12:00) • 1 ECMWF deterministic • 1 DWD deterministic • Two forecasts a day (00:00 & 12:00) 122 discharge curves to analyse per day 69 ? 69 stream flow forecasts for 1 forecast 138

  29. Thresholds Q20 Q5 Q2 Q1.3 EFAS Technical Scheme Meteorological observations LISFLOOD Discharge time series N years Return period statistics • Thresholds are derived from simulated time series. • The same model set-up and parameterisations are used in the forecasts to remain model consistent

  30. EFAS - Visualisingthreshold exceedance Probability of exceeding precipitation thresholds Highest EFAS threshold exceeded with ECMWF NWP

  31. EFAS - Exceedanceof EPS Nr of EPS above High threshold Nr of EPS above Severe threshold

  32. EFAS - Time series

  33. EFAS - Time series simplied Single deterministic forecasts EPS forecasts

  34. EFAS - Condensing information Nr of EPS exceeding thresholds

  35. Event forecast Previous forecasts EFAS - Looking back in time Today’s forecast Evaluation of persistence in time and consistence between forecasts are important

  36. EFAS forecasting page

  37. EFAS statistics

  38. Case studies • European Flood Alert System • TIGGE

  39. TIGGE – case study Romania 7 different forecasts for the October 2007 floods in Romania

  40. TIGGE – case study Romania Warning maps: Some individual centres clearly over predict others significantly under predict

  41. TIGGE – case study Romania

  42. TIGGE – case study Romania Increased warning time

  43. TIGGE – 18 month study Severn (UK) Meteo Input corrected by QQ-mapping. Discharge corrected by QQ-mapping and logistic regression.

  44. TIGGE – 18 month study Severn (UK)

  45. TIGGE – 18 month study Severn (UK)

  46. TIGGE – 18 month study Severn (UK) Key results: • Lead times which are longer than the concentration time of the catchment (2.5 days in our case) are dominated by meteorological uncertainties. However, it is notable that parameter set and initial condition uncertainty is present • the entire Grand Ensemble predictions are not in general the best. Instead it is a combination of centres from the grand ensemble that performs best. • Detrmining which centre -centre combination will perform best at any particular time step a priori is not currently possible. • In general centre combinations perform better than single centres.

  47. Summary • EPS are increasingly tested and applied for operational flood forecasting for early warning (LEPS, EPS, seasonal) • EPS based forecasts allow earlier detection of floods and provide early warning. Decision making for Civil Protection based on EPS remains difficult • Uncertainty of EPS based flood forecasts can be reduced significantly through the use of threshold exceedance, persistency criterion and post-processing

  48. Thanks for listening! • References can be provided on request, just email me florian.pappenberger@ecmwf.int “ In case of flooding “

More Related