Regione Umbria Dept of Environment, Territory and Infrastructures Division of Land Defence, Quarries, Mines and Mineral

1. Regione UmbriaDept of Environment, Territory and InfrastructuresDivision of Land Defence, Quarries, Mines and Mineral Waters Rainfall-runoff models for the activities of the regional warning centre (first applications) Nicola Berni, Angelo Viterbo, Claudia Pandolfo, Marco Stelluti, Luca Brocca Workshop on Rainfall – Runoff and regional models Sofia, Bulgaria 9 - 11 July 2007 FLOODMED Monitoring, forecasting and best practices for FLOOD Mitigation and prevEntion in the CADSES region

2. PP6 (Umbria region – Italy) role on FLOODMED project: • Provide hydrometeorological data from the regional real time operating network for model calibration (Actions 2.2, 2.3, 2.5, 3.1) • In this action we just would like to give a contribution about our first applications about rainfall – runoff models implemented for warning purposes in a sub-basin of the Tiber river Floodmed pilot area

3. Ø OBSERVATION Ø NOW-CASTING Prevenzione and control management Monitoring Gestione della Previsional emergency prevention del rischio emergenza risk PHASE 1 PHASE 2 PHASE 3 PHASE 4 FUNCTIONAL CENTERS NATIONAL WARNING SYSTEM CIVIL PROTECTION REAL TIME Regional “Centri Funzionali” (Functional Centers) and National Warning System

4. LSPP control Meteo predictions Predicted Ground Effects for the following day Precipitation observed in previous days Evaluation tables

5. Italian National network of Functional Centers 25 FC (central, regional) 2 CFC – “Central” (DPC , APAT – Roma) 21 CFD – “Regional” 2 CFS – “Service” (CAE – Bologna; ACROTEC - Savona)

6. Italian National network of Functional Centers fully operating * 9 • Fully operating for DPC of Rome • CFD Province Autonome Trento e Bolzano • CFD regionali: • Piemonte, Lombardia, Liguria, Emilia Romagna, Toscana, Marche, Campania Not yet fully operating * Update at june 2007

7. Umbrian Functional Center • December 2006: conclusion of new office civil and plants works, logistics, personnel definition; • March 2007: beginning of activities related to the definition of the procedures, scientific and technical work. • June 2007:political inauguration and official start-up.

8. Data acquisition and collection • Real time acquisition of: • LAM models from CF Emilia Romagna and Toscana; • Hydrometeorological data from the whole central Italy networks (Arno and Tiber rivers); • Meteosat second generation images; • Meteo Radar data (this autumn); • Quasi static data collection of: • GIS themes on hydrogeologic risk matters; • Cartography at different scales; • Planning studies related to flooding areas mapping; • Flood lamination plan (dam managing during severe events); • Hydrologic and hydraulic studies in general.

9. Rainfall – runoff models: Main issues For the activities of our regional warning centre we need simple models, robust, with few parameters but able to forecast quickly mainly time to peak and peak discharges Dilemma: Is it better to develop “home made” codes or adopt commercial or free to use tools continuously updated and tested?

10. Available forecasting tools operating nowadays Mathematical models running: • X_Nash; • Mike DRiFt + Mike 11 HD mod; • Hec-HMS + Hec-RAS; • Mobidic; • STAFOM (CNR-IRPI); On-line with real time data V V RR models Not yet V routing model Not yet

11. Why we have chosen these models? • DRiFt and HMS are common event-bases models quite easy to use; • Mobidic is a distributed physically-based continuous model developed by Florence University and usefully adopted by Tuscan Functional Center from 2002.

12. Hec - HMS • Most of the rainfall-runoff and routing models included in HEC-HMS are: • mainly event models; • mainly lumped; • deterministic; • both empirical and conceptual models.

13. For precipitation-runoff-routing simulation, HEC-HMS provides the following components: • Precipitation-specification options which can describe an observed (historical) precipitation event, a frequency-based hypothetical precipitation event, or a event that represents the upper limit of precipitation possible at a given location. • Loss models which can estimate the volume of runoff, given the precipitation and properties of the watershed. • Direct runoff models that can account for overland flow, storage and energy losses as water runs off a watershed and into the stream channels. • Hydrologic routing models that account for storage and energy flux as water moves through stream channels. • Models of naturally occurring confluences and bifurcations. Models of water-control measures, including diversions and storage facilities.

14. HEC-HMS also includes: • A distributed runoff model for use with distributed precipitation data, such as the data available from weather radar. • A continuous soil-moisture-accounting model used to simulate the long-term response of a watershed to wetting and drying. • An automatic calibration package that can estimate certain model parameters and initial conditions, given observations of hydrometeorological conditions. • Links to a database management system that permits data storage, retrieval and connectivity with other analysis tools available from HEC and other sources.

15. Mike DRiFt • DRiFt is a linear, semi-distributed model with constant parameters entirely developed by an Italian research institute (CIMA of Genova); • It’s implemented by DHI among the available hydrologic models of MIKE 11 from rel. 2005; • Uses a Time Variant Unit Hydrograph (TVUH) approach; • It can be easily coupled with HD Mike 11 module.

16. 3 modules: Dreainage network estimation Concentration times for each cell Effective rainfall

17. 5 parameters: k, ASk vh, vc SCS-CN

18. Concentration time distribution:

20. Mobidic The model MOBIDIC (MOdello di Bilancio Idrologico DIstribuito e Continuo – Distributed and Continuous Model for the Hydrological Balance) is used for the hydrological forecast and monitoring at the Tuscan Functional Center. The model was developed in cooperation with the University of Florence. MOBIDIC is a physically-based model that allows the estimation of the components of the hydrologic balance in the subsurface layer, the soil-vegetation system and surface water bodies. The hydrological balance can then be run in realtime or deferred mode with the desired spatial and temporal resolution.

21. In MOBIDIC, the spatial domain for the computation of the hydrological processes is represented with an horizontal discretization of the basin in square cell with arbitrary size and a vertical discretization into 5 layers: 1) vegetation, 2) surface reservoirs (rivers, lakes, artificial basins) , 3) gravitational soil 4) capillary soil 5)groundwater In the representation of physical processes, the main innovations concern the coupling of the water balance in soil and vegetation with surface energy balance (to the benefit of evapotranspiration computation and use of remotely sensed maps of Land Surface Temperature for calibration and validation) and the detailed interaction between ground water and surface water bodies.

22. The states computed by the model are soil water content, hydrological and energy balance components (evapotranspiration, soil temperature) and discharge in each branch of the river network (including minor branches).

23. At the Functional Center, MOBIDIC is run in realtime mode, fed by the data from the hydrometeorological network (every 30 minutes) and Quantitative Precipitation Forecasts from 4 different meteorological models. Also through a web interface, the user can query the main discharge stations and view the predicted discharges assuming different predicted rainfall.

24. Valore massimo previsto e ora prevista di raggiungimento 2° threshold value 6.43 (02:48 gg/mm) 1° threshold value Predicted levels Observed levels Predicted rainfall with LAMI Observed rainfall DWD WRF

25. Internet experimental web page (Umbria region FC)

26. Case study:upper Tiber catchment at P.Felcino 3 stream gauge (along the main channel); 8 for tributaries; 11 raingauges GOOD amount of data (GIS layers, DEM, meteo stations, rating curves, etc.)

27. Flooding events considered 6 calibrating events 1 test events (the most important one of November 2005)

28. HMS calibration 6 calibrating events Parameters: Lag CN Ia

29. HMS calibration • Good estimation of the dynamic aspects (Lag); • Critical aspect (usually for the majority of Mediterranean catchments): AMC correct assessment (API doesn’t work well!); DRifT calibration • Good estimation of the dynamic aspects (more difficulties related to the number of parameters to calibrate with no automatic calibration routine); • same critical aspect about AMC assessment.

30. Mobidic “validation” • Parameters NOT CALIBRATED yet; • (Just fed with real time data with quite short time series);

31. Test event: November 2005 MAIN STATISTICS

32. Conclusions 1/2 • All the shown models, thus after first quick applications, gave quite good results for warning purposes; • HMS and DRiFt produced similar results on simulating observed hydrographs in terms of time to peak and peak discharge value; • Very good results were linked to the estimation of time to peak value; • Better results (discharge peak value) depend very strongly on initial moisure conditions (AMC): it will be useful to find other indicators (BFI, etc.);

33. Conclusions 2/2 • Continuous models could solve these problems; • Calibration of Mobidic continuous model can produce for Umbrian Tiber river, such as for Tuscan Arno river, good operative results; • HEC-HMS, for the simplicity of use and calibration of parameters, for the fact that it’s free of charge and worldwide used, seems to be a good “parallel” tool for warning purposes.

34. Work to do • Implementation of the models within a distributed- semidistributed framework, and link with hydraulic models for the main Tiber river; • Feed all available RR models automatically from real time data (on-line continuous working); • Implement continuous HEC-HMS for AMC good estimates; • Continue the comparison between HEC-HMS and MIKE-DRifT; • Calibration of Mobidic; • (and much more…)

35. Thank you