1. NWS River Forecast System Evolves to CHPS Scott Lindsey Development and Operations Hydrologist, Alaska-Pacific River Forecast Center NOAA/National Weather Service 30 October 2009

2. NWSRFS Evolves to CHPS2 Topics  National Weather Service River Forecast System (NWSRFS) Introduction  Snow modeling within NWSRFS  Rainfall Runoff modeling within NWSRFS  Transition to Community Hydrologic Prediction System (CHPS) – What is it and what does it mean for the research community?

3. 30 October 2009NWSRFS Evolves to CHPS3 National Weather Service River Forecast System (NWSRFS)  Is the heart of today’s river forecasting mission  Composed of subsystems for  Deterministic operational forecasting  Ensemble streamflow prediction  Model calibration  Forecast verification  Contains  Hydrologic and hydraulic models  Data preparation and manipulation modules  Interactive display for river forecasters  Model set-up and calibration  Uses basin-based computational domain

4. 30 October 2009NWSRFS Evolves to CHPS4 Functions of NWSRFS Operational Forecast system Hydrologic Operations Calibration System Ensemble Streamflow Prediction System Verification System

5. 30 October 2009NWSRFS Evolves to CHPS5  System was cutting edge architecture in the 1970s  Developed for use on mainframe computers  Computer real memory constrained  Performance required highly customized database  Implemented in Fortran IV  Modular in the sense of having being able to put together different models to simulate different hydrologic situations  Difficult to add new models and techniques  Collaboration and research to operations was inhibited 1970s 1980s 1990s 2000s 2003 today 1970s 1980s 1990s 2000s 2003 today NWSRFS Introduction

6. 30 October 2009NWSRFS Evolves to CHPS6 Calibration System Historical Data Real-time Data Observed & Projected Hydromet Analysis Observed & Predicted Values Historical Data Analysis Areal Time Series Model Calibration Parameters Hydrologic/Hydraulic Models Short Term Forecast Current States Hydrologic/ Hydraulic Models now time window flow Probabalistic Products Deterministic Products Verification Software Verification Data/Graphics Archived Data Observed and Forecast Verification System NWSRFS Schematic Operational System Ensemble Streamflow Prediction

7. 30 October 2009NWSRFS Evolves to CHPS7 Hydrologic Modeling in NWSRFS  Snow and Glaciers  Rainfall/Runoff  Soil Moisture Accounting  Routing River Forecasting

8. 30 October 2009NWSRFS Evolves to CHPS8 Snow Accumulation/Ablation modeling  APRFC uses the Snow-17 model in NWSRFS  Conceptual model using air temperature as the sole index to determine the energy exchange across the snow-air interface  Precipitation is the only other required input variable  SNOW-17 was primarily designed for use in river forecasting. This means that the model needs to use data that are readily available everywhere, both historical climatological data for calibration and real time data for operational applications.  Calibrated by adjusting model parameters such as the melt factors, areal depletion curves, and the snow correction factor to tune the volume and timing of the snowmelt.  Good documentation available at http://earth.boisestate.edu/home/jmcnamar/hydro09/Readings/snow17.pdf http://earth.boisestate.edu/home/jmcnamar/hydro09/Readings/snow17.pdf

9. 30 October 2009NWSRFS Evolves to CHPS9 Anderson, 2006

10. 30 October 2009NWSRFS Evolves to CHPS10 Anderson, 2006

11. 30 October 2009NWSRFS Evolves to CHPS11 Summary of Snow Model Parameters The SNOW-17 model has 12 parameters. This counts the areal depletion curve as one parameter though it is input as a series of 9 values used to define the shape of the curve. Some of the parameters have more influence on the simulation results than others. The most influential, or major parameters, are those that typically have to be determined through calibration even though some guidelines are available to obtain initial estimates [Anderson (2002)]. The others, the minor parameters, typically can be assigned values based on the climatological conditions at the location being modeled. These parameters have a much smaller effect, in general, on the results and seldom need to be altered from their initially assigned value. There are no parameters associated with the snow depth computations in SNOW-17. There are several potential parameters, but all of these have been set to a fixed value based on comparisons of simulated versus observed snow depth and water equivalent at selected locations that had quality observations of both quantities.

12. 30 October 2009NWSRFS Evolves to CHPS12 1. SCF - The multiplying factor which adjusts precipitation that is determined to be in the form of snow. SCF primarily accounts for gage catch deficiencies, but also implicitly includes the net effect of vapor transfer (sublimation and condensation, including from intercepted and blowing snow) and transfers across areal divides. 2. MFMAX - Maximum melt factor during non-rain periods – assumed to occur on June 21st (mm·°C-1· 6 hr-1). 3. MFMIN - Minimum melt factor during non-rain periods – assumed to occur on December 21st (mm·°C-1· 6 hr-1). 4. UADJ - The average wind function during rain-on-snow periods. UADJ is only a major parameter when there are fairly frequent rain-on-snow events with relatively warm temperatures. 5. SI - The mean areal water equivalent above which there is always 100 percent areal snow cover (mm). SI is not a major parameter when the model is applied at a point location or when significant bare ground appears soon after melt begins no matter the magnitude of the snow cover. 6. Areal Depletion Curve – Curve that defines the areal extent of the snow cover as a function of how much of the original snow cover remains after significant bare ground shows up. The areal depletion curve also implicitly accounts for the reduction in the mean areal melt rate that occurs as less of the area is covered by snow. Generally not needed for a point location. The major parameters for the SNOW-17 model are:

13. 30 October 2009NWSRFS Evolves to CHPS13 1. NMF - Maximum negative melt factor (mm·°C-1·6 hr-1). The negative melt factor has the same seasonal variation as the non-rain melt factor, thus the maximum value is assumed to occur on June 21st. 2. TIPM - Antecedent temperature index parameter (real – range is 0.01 to 1.0). Controls how much weight is put on temperatures from previous time intervals when computing ATI. The smaller the value of TIPM, the more previous time intervals are weighted. 3. PXTEMP - The temperature that separates rain from snow (°C). If the air temperature is less than or equal to PXTEMP, the precipitation is assumed to be in the form of snow. The PXTEMP parameter, as defined for SNOW-17, is not used if a rain-snow elevation time series is used to determine the form of precipitation. 4. MBASE - Base temperature for snowmelt computations during non-rain periods (°C). Typically a value of 0°C is used. 5. PLWHC - Percent liquid water holding capacity (decimal fraction). Indicates the maximum amount of liquid water, as a fraction of the ice portion of the snow, that can be held against gravity drainage (maximum allowed value is 0.4). 6. DAYGM - Constant daily amount of melt which takes place at the snow-soil interface whenever there is a snow cover (mm·day-1). The minor parameters for the SNOW-17 model are:

14. 30 October 2009NWSRFS Evolves to CHPS14 Anderson, 2006

15. 30 October 2009NWSRFS Evolves to CHPS15 Rainfall-Runoff modeling  APRFC uses the Sacramento Soil Moisture Accounting (SAC-SMA) model  Spatially-lumped continuous soil moisture accounting model  Ideal model for the simulation of large-scale (>1000 km2) basins  Takes mean precipitation, evaporation and temperature as input  Calibrated by adjusting baseflow, tension water capacities and runoff simulation parameters  Calibration input includes point or areal estimates of historical precipitation, temperature, and potential evaporation  Documentation at http://www.nws.noaa.gov/oh/hrl/nwsrfs/users_manual/part2/_pdf/23sacsma.pdf http://www.nws.noaa.gov/oh/hrl/nwsrfs/users_manual/part2/_pdf/23sacsma.pdf

17. How the Sacramento Model Works Each basin is represented vertically by two zones:  An upper zone (short- term storage capacity)  A lower zone (bulk of the soil moisture and longer groundwater storage)  Tension water elements (water bound by adhesion and cohesion, extracted only by evapotranspiration)  Free water elements (free to move under gravitational forces, may be depleted by evapotranspiration, percolation, horizontal flow)

18. Sacramento Model Parameters http://www.crh.noaa.gov/ncrfc/doc/calibration/flowing.html

19. 30 October 2009NWSRFS Evolves to CHPS19 Interactive Hydrologic Tools

20. 30 October 2009NWSRFS Evolves to CHPS20 Modeling the Tanana River Basin

21. 30 October 2009NWSRFS Evolves to CHPS21 Flow Simulation

22. 30 October 2009NWSRFS Evolves to CHPS22 Snowpack Simulation

23. 30 October 2009NWSRFS Evolves to CHPS23 Shortcomings of NWSRFS  Current snow model is a temperature index model, not an energy balance.  Scale of inputs: Lumped inputs from point data applied over basins ranging from 300 to 25000 km 2  Ability to incorporate modeling advances (distributed models, new snow and soil moisture models) is inhibited by system design (originally for a main frame).

24. 30 October 2009NWSRFS Evolves to CHPS24 Stimulus for Modernization  New NOAA Hydrology Program initiatives  Nationally consistent gridded water resources forecasts delivered to customers for critical decisions  Probabilistic river forecasts at all time scales  Verification extended to probabilistic forecasts  Better research-to-operations pathway  Easier, faster, cheaper  Increase collaboration with federal, academic, public, private, and international partners  Share data and models more effectively  Increase capabilities to modify/expand services

25. 30 October 2009NWSRFS Evolves to CHPS25 CHPS Project  CHPS is the NWS’s Community Hydrologic Prediction System  CHPS will replace NWSRFS  CHPS is built using framework delivered by the product, “Delft-FEWS”, from Deltares in The Netherlands  CHPS  Highly modular and configurable  Based on Service Oriented Architecture (SOA) principles  Allows easy addition of new models and techniques  Promotes streamlined research to operations  Allows greater collaboration with NOAA’s hydrology partners (e.g., USACE, USGS, etc.)

26. 1. What’s the difference between CHPS and FEWS?  Flood Early Warning System (FEWS) is a suite of infrastructure software maintained and supported by Deltares. Only when FEWS is configured for the user’s specific domain does it transform into a functioning system. A user would supply the necessary modeling operations, or acquire them from a source which shares (open source) or sells FEWS-compatible models.  Community Hydrologic Prediction System (CHPS) is NOAA’s customized application of FEWS. CHPS runs models that are compatible with FEWS - including those migrated from NWSRFS – providing extra user capabilities not available via FEWS, such as model calibration. In the future, NOAA will make CHPS models available to other FEWS users. 30 October 2009NWSRFS Evolves to CHPS26

27. 30 October 2009NWSRFS Evolves to CHPS27 CHPS Terminology FEWS FEWS features have potential application to the entire FEWS user community (generic time- series storage, displays, workflows, etc.) FEWS is a suite of configurable modules which can store, manipulate, and display time series data using your own and community applications. CHPS is a uniquely configured realization of FEWS, using RFC-specific data and applications. Other FEWS-based systems (e.g., NFFS) are different realizations of FEWS. Community applications shared with other FEWS users Applications, configurations, data, unique to CHPS

28. 30 October 2009NWSRFS Evolves to CHPS28 CHPS Implementation Strategy  Port NWSRFS models that require calibration  Hydrologic/Hydraulic Models  Soil Moisture Models  Snow Models  Reservoir Models (BASEFLOW, SARROUTE, CONS_USE, LAG/K, LAY-COEF, TATUM, MUSKROUT, RES-J, RSNWELEV, SNOW-17, CHANLOSS, STAGE-Q, SSARRESV, UNIT-HG, RES-SNGL, SAC-SMA)  Create FEWS adapters for external models (e.g., HEC- RAS and ResSim from USACE) (2003) 2006 2007 2008 2009 2010 2011

29. 30 October 2009NWSRFS Evolves to CHPS29  Develop forecaster “MODs” capability in FEWS  “MODs” give forecaster ability to adjust model inputs and parameters in during forecast runs  Deltares building new “MODs” capability and interface into FEWS infrastructure for all customers by working closely with RFC personnel  Can modify model states, input data, & parameters (see NWSRFS list: (IGNORETS, FMAP, SSARREG, MFC, RRICHNG, SWITCHTS, TSCHNG, CHGBLEND, WECHNG, RAINSNOW, RRIMULT, WEADD, TSADD, SACCO, AESCHNG, ROMULT, SETMSNG, UADJ, ROCHNG, UHGCHNG, SETQMEAN, UHGDATE, QCSHIFT, QPSHIFT, HECRAS, etc.) (2003) 2006 2007 2008 2009 2010 2011 CHPS Implementation Strategy

30. 30 October 2009NWSRFS Evolves to CHPS30 The FEWS Adapter  A key feature of DELFT-FEWS is its ability to run external modules to provide essential forecasting functionality. The General Adapter is the part of the DELFT- FEWS system that implements this feature. It is responsible for the data exchange with these modules and for executing the modules and their adapters.

31. 30 October 2009NWSRFS Evolves to CHPS31 Development of FEWS Adapter

32. 30 October 2009NWSRFS Evolves to CHPS32 Models linked to Delft-FEWS Models linked to Delft-FEWS  The table below gives an overview of the models linked via the Published Interface to the Delft-Fews system. For these models, a model adapter is available. Please note that adapters that have not been developed by Deltares cannot be used without permission of the owner. All models indicated in bold typeface are running in operational systems.Published Interface Model Type Supplier/Owner Country  ISIS Hydrodynamic HR/Halcrow UK ISISHRHalcrow  PDM Rainfall-Runoff CEH UK PDMCEH  TCM Rainfall-Runoff CEH UKCEH  KW Routing (kinematic wave) CEH UKCEH  PACK Snow Melt CEH UKCEH  ARMA Error Correction CEH UKCEH  PRTF Event Based RR PlanB UK  PCRASTER Dynamic Modelling Software Univ. Utrecht NetherlandsUniv. Utrecht  TRITON Surge propagation/Overtopping PlanB UK  TWAM 2D Hydrodynamics PlanBUK  STF Transfer functions EA UK  DODO Routing (layered Muskingum) EAUK  MCRM Rainfall-Runoff EAUK  Modflow96/VKD 3D groundwater Deltares/Adam Netherlands/UK Modflow96/VKD Taylor  Mike11 Hydrodynamics DHI DenmarkDHI

33. 30 October 2009NWSRFS Evolves to CHPS33 Models linked to Delft-Fews (cont.) Model Type Supplier/Owner Country  NAM Rainfall-Runoff DHI DenmarkDHI  TOPKAPI Rainfall-Runoff Univ. of Bologna Italy  HBV Rainfall-Runoff (inc SHMI Sweden snowmelt)  Vflo Distributed Rainfall-Runoff Vieux & Associates USA  SWMM Urban Rainfall-Runoff USGS USA  HEC-RAS Hydrodynamic USACE USA  HEC-ResSim Reservoir Simulation USACE USA  Snow17 Snow Melt NWS USA  SAC-SMA Rainfall-Runoff NWS USA  Unit-HG Unit-Hydrograph NWS USA  LAG/K Routing (hydrological) NWS USA  SARROUTE Routing (hydrological) NWS USA  SSARRESV Reservoir Simulation NWS USA  RESSNGL Reservoir Simulation NWS USA  BASEFLOW Baseflow Simulation NWS USA

34. 30 October 2009NWSRFS Evolves to CHPS34 Models linked to Delft-Fews (cont.) Model Type Supplier/Owner Country  CHANLOSS Channel loss Simulation NWS USA  APICONT Rainfall-Runoff NWS USA  CONSUSE Consumptive use of River NWS USA Simulation  GLACIER Glacier simulation NWS USA  LAYCOEF Routing Model NWS USA  MUSKROUTRouting Model NWS USA  RSNELEV Rain Snow Elevation NWS USA Simulation  SACSMA-HT Rainfall-Runoff (Heat NWS USA Transfer)  LAYCOEF Routing Model NWS USA  TATUM Routing Model NWS USA  rtcModule Reservoir Simulation Deltares Netherlands rtcModuleDeltares  PRMS Rainfall-Runoff Univ. of Karlsruhre Germany  SynHP Hydrodynamics BfGGermany  SOBEK Hydrodynamics, Water Deltares Netherlands SOBEKDeltares Quality, RR

35. 30 October 2009NWSRFS Evolves to CHPS35 Models linked to Delft-Fews (cont.) Model Type Supplier/Owner Country  SOBEK-2d Linked 1d/2d inundation Deltares Netherlands SOBEK-2dDeltares modeling  DELFT-3D 2D-3D Hydrodynamics Deltares NetherlandsDELFT-3DDeltares  Sacramento Rainfall-Runoff Deltares NetherlandsDeltares  RIBASIM Water distribution Deltares NetherlandsDeltares + Reservoir  REW Distributed Rainfall-Runoff Deltares NetherlandsDeltares  DELFT3D 2/3D Hydrodynamics/ Deltares Netherlands DELFT3DDeltares Water quality  Flux 1D Hydrodynamics Scietec AustriaScietec  URBS rainfall-runoff and Don Caroll Australia hydrological routing  Grid2GridDistributed Hydrologic CEH UKCEH Model  Het Wageningen Rainfall-Runoff HaskoningNetherlands Model

36. 30 October 2009NWSRFS Evolves to CHPS36 Summary  Hydrologic Modeling at the River Forecast Center has been in a “remodeling” phase for many years (porting NWSRFS from Mainframe to IBM RISC to HP UNIX to LINUX).  Now we are in a “new construction” phase that should allow us to take much better advantage of the advances in research, technology and modeling.  We plan to keep UAF modelers in the loop and hope to apply your expertise to the real-time hydrologic forecasting challenges here in Alaska

37. 30 October 2009NWSRFS Evolves to CHPS37 Questions???