2. RUC Development - Applications to National C/V John Brown NOAA Forecast Systems Lab 19 April 2001

3. 2 Key goals for 20km RUC Take advantage of increased computer power  higher spatial resolution Improve RUC performance for QPF, especially for convective precipitation Improve RUC initial conditions ------------------------------------------------------------ Incorporate new advances in model and assimilation Eliminate bugs

4. 3 1999-2000 OperationsSpring 2001 Operations Resolution40 km, 40  levels20 km, 50  levels AnalysisOptimal interpolation on3-d variational technique on generalized on generalized  surfaces  surfaces, hydrometeor analysis w/ GOES…, use raw instead of interp. obs AssimilationIntermittent 1-h cycleIntermittent 1-h cycle Stable cloudsMixed-phase cloud microphysics MM5),New version of MM5/RUC / precipitation explicit fcst of cloud water, rain water, microphysics (FSL and NCAR), snow, ice, graupel, much shorter microphysics time step no. concentration of ice particles Sub-grid-scaleGrell (1993)New Grell scheme w/ ensemble cloud, precipitation shallow convection, detrainment of cloud water to microphysics TurbulenceBurk-Thompson explicit TKE schemeBurk-Thompson, e-  parameterization under development RadiationMM5 LW/SW scheme, f(hydrometeors)MM5 scheme with fix to SW lag error Land-sfc processes6-level soil/veg model (Smirnova, 2-layer snow model, 1997, 2000) w/ frozen soil, 1-layer snow improved cold season processes, improved diurnal cycle Sfc conditionsDaily 50km SST/14 km LST,Add high-res USGS land-use/soil type, 0.14  monthly NDVI veg frac, cycled soil albedo moisture/temp, snow depth/temp Boundary condsEta model initialized every 12hEta model initialized every 6h Forecast durationHourly output to  3h, 12h fcst every 3hHourly output to  6 h, 12h fcst every 3h Rapid Update Cycle – Present and Next Version

5. 4 Subset of full domain 20km RUC/MAPS topography

6. 5 40km RUC 40 levels 20km RUC 50 levels 10 new levels 7 – upper levels – 330 – 500 K 3 – lower levels – 270 – 290 K RUC native coordinate levels

7. 6 1999-2000 OperationsSpring 2001 Operations Resolution40 km, 40  levels20 km, 50  levels AnalysisOptimal interpolation on3-d variational technique on generalized on generalized  surfaces  surfaces, hydrometeor analysis w/ GOES…, use raw instead of interp. obs AssimilationIntermittent 1-h cycleIntermittent 1-h cycle Stable cloudsMixed-phase cloud microphysics MM5),New version of MM5/RUC / precipitation explicit fcst of cloud water, rain water, microphysics (FSL and NCAR), snow, ice, graupel, much shorter microphysics time step no. concentration of ice particles Sub-grid-scaleGrell (1993)New Grell scheme w/ ensemble cloud, precipitation shallow convection, detrainment of cloud water to microphysics TurbulenceBurk-Thompson explicit TKE schemeBurk-Thompson RadiationMM5 LW/SW scheme, f(hydrometeors)MM5 scheme with fix to SW lag error Land-sfc processes6-level soil/veg model (Smirnova, 2-layer snow model, 1997, 2000) w/ frozen soil, 1-layer snow improved cold season processes, improved diurnal cycle Sfc conditionsDaily 50km SST/14 km LST,Add high-res USGS land-use/soil type, 0.14  monthly NDVI veg frac, cycled soil albedo moisture/temp, snow depth/temp Boundary condsEta model initialized every 12hEta model initialized every 6h Forecast durationHourly output to  3h, 12h fcst every 3hHourly output to  6 h, 12h fcst every 3h Rapid Update Cycle – Present and Next Version

8. 7 400 500 600 700 800 900 1000 400 500 600 700 800 900 1000 400 500 600 700 800 900 1000 OI 3DVAR OBS Riverton, WY sounding 12 UTC 18 Aug 2000

9. 8 Data for 20-km RUC 3dVAR at NCEP Data Type~NumberFreq. Rawinsonde (inc. special obs) 80/12h NOAA 405 MHz profilers 31/ 1h (better use) VAD winds (WSR-88D radars) 110-130/ 1h (3h currently) Aircraft (ACARS) (V,temp) 1400-4500/ 1h Surface/METAR - land (V,p sfc,T,T d )1500-1700/ 1h Buoy 80-150/ 1h GOES precipitable water1500-3000/ 1h (closer fit) GOES cloud drift winds 1000-2500/ 1h (closer fit) GOES cloud-top pressure ~40km res / 1h SSM/I precipitable water1000-4000/2-6h GPS precipitable water ~65  / 1h Boundary-layer (915 MHz) profilers ~24/ 1h RASS (WPDN and PBL) 15/ 1h Ship reports 10s/ 3h Reconnaissance dropwinsonde a few/ variable New for 20km RUC at NCEP

10. 9 Initial RUC cloud analysis technique Uses GOES/NESDIS cloud-top pressure (sounder-based) Uses RUC 1-h hydrometeor fcst (cloud water, ice, snow/rain/graupel) as first guess Performs cloud clearing and cloud building Uses observations to build 3-d cloud yes/no/unknown field surface cloud obs, radar reflectivity assimilable with same code structure/logic Goal – Improve RUC precipitation, cloud, icing forecasts

11. 10 3h 20km RUC cloud-top fcst w/ GOES cloud assim Verification Cloud-top pressure based on NESDIS product Effect of GOES data on 3-h RUC cloud forecasts 2100 UTC Wed 21 March 2001 3h 40km RUC cloud-top fcst No GOES cloud assim

12. 11 Cloud water analysis – cross-section Without GOES With GOES Ice mixing ratio analysis – cross-section Without GOES With GOES

13. 12 RUC cloud analysis status Technique developed combining GOES sounder-based cloud-top data with RUC explicit cloud forecasts Current version – impact on forecasts –Cloud top – more improvement in 1-h forecasts, smaller but consistent improvement in 12-h cloud-top forecasts –Slight improvement in RH bias and standard deviation, especially in mid-troposphere –Slight improvement in QPF at larger thresholds

14. 13 1999-2000 OperationsSpring 2001 Operations Resolution40 km, 40  levels20 km, 50  levels AnalysisOptimal interpolation on3-d variational technique on generalized on generalized  surfaces  surfaces, hydrometeor analysis w/ GOES…, use raw instead of interp. obs AssimilationIntermittent 1-h cycleIntermittent 1-h cycle Stable cloudsMixed-phase cloud microphysics MM5),New version of MM5/RUC / precipitation explicit fcst of cloud water, rain water, microphysics (FSL and NCAR), snow, ice, graupel, much shorter microphysics time step no. concentration of ice particles Sub-grid-scaleGrell (1993)New Grell scheme w/ ensemble cloud, precipitation shallow convection, detrainment of cloud water to microphysics TurbulenceBurk-Thompson explicit TKE schemeBurk-Thompson, e-  parameterization under development RadiationMM5 LW/SW scheme, f(hydrometeors)MM5 scheme with fix to SW lag error Land-sfc processes6-level soil/veg model (Smirnova, 2-layer snow model, 1997, 2000) w/ frozen soil, 1-layer snow improved cold season processes, improved diurnal cycle Sfc conditionsDaily 50km SST/14 km LST,Add high-res USGS land-use/soil type, 0.14  monthly NDVI veg frac, cycled soil albedo moisture/temp, snow depth/temp Boundary condsEta model initialized every 12hEta model initialized every 6h Forecast durationHourly output to  3h, 12h fcst every 3hHourly output to  6 h, 12h fcst every 3h Rapid Update Cycle – Present and Next Version

15. 14 Key changes in 20km RUC forecast model for QPF Fixes to bugs in vertical advection of moisture Fix to bug for effect of vertical velocity on convective precipitation 20km/50 level resolution – resolved precipitation improved – grid volumes are now ~5 times smaller Improved land-surface parameterization New ensemble cumulus parameterization – Grell Revised version of MM5/RUC microphysics

16. 15 Deep convection parameterization – new Grell scheme Inclusion of cloud water/ice detrainment (interaction with microphysics scheme) Ensemble of closure and feedback assumptions

17. 16 Cloud and precipitation microphysics New version of Reisner et al (1998, Quart. J. Roy. Meteor. Soc.) "option 4" mixed-phase, bulk microphysics scheme ("EXMOISG") Initially developed at NCAR for MM5, and implemented with RUC-2 at NCEP in April 1998 New version removes the most egregious faults with version of EXMOISG currently in the operational RUC-2 at NCEP

18. 17 MM5/RUC Cloud and precipitation microphysics scheme Main aspects of EXMOISG scheme Predicted quantities: - Mixing ratios of water vapor cloud liquid water rain water cloud ice snow graupel - Number concentration of cloud ice Water and ice permitted to coexist for -40  C < T  0  C. Inverse exponential (Marshall-Palmer) particle size distribution assumed.

19. 18 St. Lawrence valley/ New England ice storm - 9h RUC2 fcst valid 2100z 9 Jan 98 N-S cross sections of RUC2 microphysics | YUL/Montreal Water vapor mixing ratio /  Cloud water mixing ratio Graupel mixing ratio Rain water mixing ratio Excessive graupel production

20. 19 PRINCIPAL CHANGES TO EXMOISG Bug fixes (including one major one involving graupel at cold temps) Reduce interval between calls to EXMOISG Formation of new ice crystals ("nucleation") constrained via "Cooper curve" rather than by "Fletcher Curve" (both empirical) Improve consistency of ice particle number calculations. Different criteria for formation of graupel as result of collisions between snowflakes and drops of cloud liquid water (temp < 0  C) Allow larger concentrations of cloud water to be present before raindrops are formed Very small rainwater contents are regarded as being composed of drizzle-sized drops

21. 20 Changes to MM5/RUC microphysics with 20km RUC SW-NE vertical cross-section across WA /Olympic Peninsula into BC and Alberta - 12h forecast valid 0300z 5 January 2001 40km operational RUC at NCEP 20km test RUC w/ microphysics fixes Bug fixes Changes for formation of ice and graupel - result – less “ice friendly” Change in time step from 10 min to 2 min Result: More realistic supercooled liquid water Improved precip type

22. 21 Outlook for RUC microphysics NCEP implementation -- Ongoing validation of real-time QPF, precipitation-type forecasts Longer term -- Continued work to improve model guidance for icing prediction will be needed well into the future. -- Possible further upgrade to EXMOISG in RUC within ~ 12 months -- Use of EXMOISG in RUC cloud assimilation -- Continued collaboration with NCAR for MM5, RUC and WRF application -- Further improvements for air quality and regional climate applications (e.g., air chemistry, radiation, land surface) NOAA-NCAR collaboration has been vital to improving EXMOISG.

23. 22 Purpose – Improve near-sfc, precip, cloud fcsts Ongoing cycle of soil moisture, soil temp, snow cover/depth/temp) 2-layer snow model RUC/MAPS Land-surface Process Parameterization New in 20km - change in thermal conductivity – better diurnal cycle - frozen soil physics, 2-layer snow model

24. 23 RUC/MAPS cycling of soil/snow fields - soil temperature, soil moisture - snow water equivalent, snow temperature MAPS snow water equivalent depth (mm) 9 December 1999 1800 UTC NESDIS snow cover field 9 December 1999 2200 UTC

25. 24 Snow water equivalent depth (RUC-20) Snow depth (USAF) Snow coverage (NESDIS) 28 November 2000 2000 UTC RUC-20 NESDIS USAF

26. 25 Further modifications of RUC land-surface model Johansen (1975) parameterization of thermal conductivity in soil – improvements against the previous parameterization (McCumber 1980) especially for saturated soils Refreezing of 13% of melted water inside the snow pack (Koren et al. 1999)

27. 26 Improved 1-d (PILPS 2d – Valdai, Russia) total runoff and snow water equivalent forecasts with improved snow and soil physics in MAPS land- surface model Total runoff Snow water equivalent

28. 27 CHANGES TO SURFACE INPUT FIELDS IN RUC-20 Soil type 1-km CONUS-SOIL, 16 classes ( 1-degree Zobler soil types in current RUC) Vegetation type 1-km data, 24 USGS classes (1-degree vegetation types in current RUC) Albedo 0.14 0 monthly data, NESDIS (1-degree seasonal climatology in current RUC) Sea-surface temperatures daily 50km analyses Vegetation fraction 0.14 0 data – NESDIS – NDVI-based

29. 28 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Soil types in 40 km RUC Soil types in 20 km RUC

30. 29 Vegetation types in 40km RUC (12 classes) High-resolution vegetation types in 20km RUC (24 USGS classes)

31. 30 Albedo in 40 km RUC Albedo in 20 km RUC Albedo in 20 km RUC -monthly NESDIS data 18 May 2000

32. 31 RUC Cloud-Base Height Height above surface where q c + q i first exceeds 10 -3 g/kg

33. 32 RUC derived visibility Relative humidity and hydrometeors are considered separately: Hydrometeors: Stoelinga and Warner (1999) with ad hoc addition of graupel Relative humidity: > 95%  visibility of 8 km < 15%  visibility of 60 km. Lowest visibility value is used.

34. 33 40 km RUC 3h fcst 20 km RUC 3h fcst 28 August 2000 15 UTC GOES visible image Vis  1 mi Vis – 3-5 mi RUC visibility forecasts

35. 34 Summary - the 20km RUC Schedule for implementation –April 2001 – Get new code/scripts running at NCEP – test mode Available for 20km RUC seminars at NWS Regions, NCEP Centers –May – Real-time testing at NCEP – evaluation in field, retrospective runs –June – Consideration for CAFTI approval –Late June – early July - planned operational implementation 20 km/50 level 1 hr version –with model improvements including cloud microphysics, convection, land-surface, hourly fcsts out to  6h –3-d variational analysis, cloud/hydrometeor analysis using satellite combined with explicit cloud fcsts in RUC –Improvements in warm- and cold-season precipitation and cloud/icing forecasts, also in surface forecasts RUC web site - http://ruc.fsl.noaa.gov - 20km test RUC products

36. 35 RUC C&V Future Work * Understand and correct deficiencies in model physics (particularly PBL, microphysics). [ongoing] * Translation algorithms– development, evaluation [ongoing] * Addition of aerosols, simple chemistry. [2002 and later] * Use of ensembles => probability forecasts of restrictive conditions [future] * Shift to WRF as rapid-update medium [future] … All activities will involve interactions with other groups.