1. The 13km RUC - Nearing Implementation at NCEP Changes, testing, plan Aviation Weather Center – Monday 16 May 2005 Stan Benjamin John Brown Kevin Brundage Dezso Devenyi Georg Grell Bill Moninger Steven Peckham Barry Schwartz Tanya Smirnova Tracy Lorraine Smith Steve Weygandt Geoff Manikin – NCEP/EMC RUC web site - http://ruc.fsl.noaa.gov

2. 2 2005- Implementation of 13km RUC in operations at NCEP Assimilation of new observations - GPS-precipitable water  improved moisture forecasts - METAR cloud/vis/current weather  improved ceiling and vis fcsts - Mesonet 915 MHz boundary-layer profilers - 915 MHz boundary-layer profilers, RASS temperatures - Soil moisture/temp nudging Model changes – New versions of - mixed-phase cloud microphysics (NCAR-FSL) - Grell-Devenyi convective parameterization - Revised radiation – cloud effects - Corrected treatment of frost formation  improved icing and convection forecasts, cloud/sfc temp, vis forecasts Improved post-processing – visibility, precip type, 20km/40km look-alike Hourly forecasts to be extended to 9h from 3h duration (at 01z, 02z, 04z, 05z, …)

3. 3 What is different qualitatively in precip/surface behavior with the RUC13? - Definitely a ‘wetter’ model than the previous RUCs - Surface dewpoint, moisture aloft - CAPE - Precipitation - But ‘drier’ over warm oceans AREA FORECAST DISCUSSION...UPDATE NATIONAL WEATHER SERVICE GOODLAND KS 1155 AM MDT SUN MAY 8 2005.UPDATE... A SHORTWAVE TROF WAS EVIDENT THIS MORNING AT BOTH 700 AND 500 MB OVER SOUTHEAST WYOMING...SPREADING TOWARD THE TRI STATE AREA IN NORTHWESTERLY FLOW ALOFT. THE ETA/NAM WAS HANDLING THIS FEATURE WELL. THE RUC IS A BIT TOO DRY.

4. 4 2005 - 13km RUC implementation plan February-May 2005 - Real-time parallel testing – EMC - Intercomparison RUC13 vs. RUC20 by EMC, FSL - daily inspection w/ graphics, statistics (raob, surface) - Transfer afterward to NCEP/NCO May-June 2005 - Real-time parallel testing – NCO - 4-week field test evaluation – NWS regions, AWC, SPC… *** STARTING NEXT WEEK - Retrospective testing (at NCEP) for cold season - Report on testing to EMC, NCEP directors ~ June-July 2005 - Operational implementation

5. 5 13km RUC Improvements expected from 13km RUC - Improved near-surface forecasts - Improved precipitation forecasts - Better cloud/icing depiction - Improved frontal/turbulence forecasts 20km RUC Terrain elevation - 200 m interval NCEP computer upgrade allows RUC13 to run in same time as current RUC20

6. 6 RUC History – NCEP (NMC) implementations 1994 - First operational implementation of RUC - 60km resolution, 3-h cycle 1998 – 40km resolution, 1-h cycle, - cloud physics, land-sfc model 2002 – 20km resolution - addition of GOES cloud data in assimilation 2003 – Change to 3dVAR analysis from previous OI (April) 2005 – 13km resolution, new obs, new model physics (estimated ~June-July) 2007 – WRF-based Rapid Refresh to replace RUC

7. 7 More detailed coastline with 13km resolution 13km RUC20km RUC Soil moisture – 22z - 21 Feb 2005 Dark blue = water

8. RUC 3DVAR development/testing RUC Analysis Goals Initial fields that yield very accurate short-range forecasts (down to 1-h): - upper-level wind forecasts - aviation hazards (turbulence, icing, ceiling/visibility) - surface sensible weather forecasts Hourly assimilation of all conventional observations, including surface observations Close analysis fit to all observations, especially METAR and rawinsonde

9. 13-km 3DVAR running at FSL since summer 2004, NCEP testing in progress 5 aspects of RUC13 3DVAR analysis: 1. New observations assimilated 2. Cloud analysis (GOES, METAR ceiling/vis/curr-wx) 3. Use of surface obs throughout PBL (NCEP Fall 2004) 4. Revised control variable for moisture (pseudo-RH) 5. Soil moisture/temp nudging

10. Observations used in RUC Data Type ~Number Freq. -------------------------------------------------- Rawinsonde 80 /12h NOAA profilers 30 / 1h VAD winds 110-130 / 1h Aircraft (V,temp) 1400-4500 / 1h Surface/METAR 1500-1700 / 1h Buoy/ship 100-150 / 1h GOES precip water 1500-3000 / 1h GOES cloud winds 1000-2500 / 1h GOES cloud-top pres 10 km res / 1h SSM/I precip water 1000-4000 / 6h -------------------------------------------------- GPS precip water ~300 / 1h Mesonet ~5000 / 1h PBL – prof/RASS ~25 / 1h METAR-cloud-vis-wx ~1500 / 1h -------------------------------------------------- NCEP RUC20 operational RUC13 (at NCEP June 2005) Cloud analysis variables

11. 11 Obs quality issues with RUC13 Mesonet data Winds of questionable quality – widespread siting, anemometer height problems Will be later added from approved mesonet providers (e.g. OK Mesonet) Assigned larger obs errors Boundary-layer profilers Implementation deferred (decision – 6 May 05) Quality appears to be poorer than NOAA-network profilers – better monitoring needed Rawinsonde moisture profiles Intermittently shows incorrect saturation in lowest 200 mb Detected in RUC13 processing using GPS PW neighbors

12. Use multiple data types to modify cloud, hydrometeor, and moisture fields: -- GOES cloud-top pressure/temp (implemented in 2002 w/ RUC20) -- Surface METAR (clouds, weather, visibility) (2005 w/ RUC13) Construct 3-d logical arrays (YES/NO/UNKNOWN) for clouds and precipitation from all info Clear/build moisture, cloud, precipitation fields Safeguards for known problems (marine stratus, convective clouds, snow, nocturnal inversion) RUC Cloud Analysis

13. -Nearest station up to 100 km distance - Assigned in ASL, includes terrain intersection for low clouds - Maps info to 3-d cloud, precip. Y/N/U arrays - Change cloud water, cloud ice, water vapor fields as follows: Build for BKN / OVC / Vertical Visibility - 40 mb thick layer (2+ model levels) - 150 mb thick for precip. + GOES clouds - Can build multiple broken layers Clear - Up to cloud base (if needed) - to 12,000 feet for CLR report Assimilation of METAR cloud/wx/vis Better analysis, prediction of ceiling and visibility

14. analysis – with METAR cloud/ visibility obs Cloud water mixing ratio Sample modification of cloud water from METAR cloud/weather/ visibility observations Cloud water mixing ratio 400 Pres (mb) -500- -600- -700- -800- -900- 1000 1700 UTC 27 Jan 2004 Background (previous 1h fcst) Relative Humidity

15. 15 IFRLIFR VFR CLR MVFR Sample ceiling analysis impact Ceiling from RUC hydrometeors Analysis WITH cld/wx/ vis obs Analysis NO cld/wx/ vis obs Aviation Flight Rules cloud ceiling height (meters) 1800 UTC 17 Nov 2003 Observations

16. 16 GPS precipitable water observations ~200-300/h- All-weather obs

17. 17 Analysis fit to GPS precipitable water obs over CONUS RUC20 RUC13 RMS difference Bias

18. 18 3h forecast fit to GPS precipitable water obs over CONUS RUC20 RUC13 RMS difference Bias

19. 19 RUC enhancements: 1.Use of METAR obs through boundary-layer depth (Sept 04) 2.Assimilation of GPS precipitable water observations (RUC13 - ~Jun05) CAPE impact from two RUC enhancements 3h fcst WITH enhancements 3h fcst OPERATIONAL 0000 UTC 21 Apr 2004 Severe reports NWS SPC Norman, OK Tornadoes

20. 20 September 2004 – Change to oper RUC20 analysis – use boundary-layer depth in assimilation of surface/METAR obs  better temp/dewpoint/CAPE/convection forecasts Spring 2005 – Will be first convective season to see effect of PBL-based assimilation. Also will be evident in RUC13 results.

21. 21 Problems with model PBL structure in RUC analysis Sounding structure is sometimes not realistic and does not account for boundary-layer depth No accounting in current RUC analysis for vertical correlation of forecast error near surface based on estimated boundary- layer depth

22. 22 PBL-based METAR assimilation Problem: Representativeness of sfc data in PBL not used Design: Use METAR data through PBL depth from 1h fcst Create extra pseudo-residuals (obs-bkg) in PBL RUC oper analysis 18z 3 Apr 02 IAD Original slides from April 2004

23. 23 PBL-based METAR assimilation Use METAR data through PBL depth from 1h fcst RUC oper analysis 18z 3 Apr 02 IAD x x x x Effect of PBL-based METAR assimilation

24. 24 EMC/Geoff Manikin RUC20/13 comparison web page http://www.emc.noaa.gov/mmb/ruc2/para

25. 25 EMC/Geoff Manikin RUC20/13 comparison web page http://www.emc.noaa.gov/mmb/ruc2/para

26. 26 Revised moisture analysis design in RUC13 – pseudo-RH Current RUC20 Separate analyses of in-situ moisture observations and precipitable water obs with intervening recalculation of observation innovations Use of log (water vapor mixing ratio-qv) as analysis control variable for moisture. Some good characteristics (continuous in 3-d space, will not go negative) but not ‘responsive’ to increase RH (e.g., from 10% to 50%) Related to summer 2004 RUC analysis fixes to avoid CAPE noise problems Revised RUC13 Fully integrated moisture analysis with both in-situ and precipitable water obs treated together Uses pseudo-RH (RH relative to background sat qv) as analysis control variable

27. 27 RUC13-exp RUC20-ops RUC13 matches moisture profiles much better than OperRUC20 Use of pseudo-RH moisture control variable instead of Log (water vapor mixing ratio) used in RUC20. TUS soundings – 12z 18 April 2005

28. 28 RUC13-exp RUC20-ops RUC13 forecast soundings at TUS also better than RUC20 - Reasons Assimilation of GPS PW Improved soil moisture in RUC13 Better physics/assimilation, soil nudging TUS soundings – 12z 18 April 2005 RUC 6h forecasts valid 12z

29. 29 36h after restoration of oper RUC20 soil moisture Oper RUC20 Para RUC13 6h fcst valid 00z Analysis valid 00z 2m dewpoint - 00z 22 Apr 05 - excessive dryness in RUC20

30. 30 RUC13 analysis – includes nudging of soil moisture and temperature Dependent on: 2m T/Td 1h forecast errors Only in daytime (zen angle > 0.3) with no clouds or precipitation (defined after METAR/GOES cloud assimilation) Developed as part of NOAA New England High-Resolution Temperature (NEHRT) Program - FSL, ETL, NSSL, NCEP/EMC Oper RUC20 Para RUC13

31. 31 Oper RUC20 Para RUC13

32. 32 Typical CAPE effect from improved soil moisture and moisture analysis in RUC13 12h forecasts Valid Thur evening 00z- 5 May 05 Oper RUC20 Para RUC13

33. 33 Typical CAPE effect from improved soil moisture and moisture analysis in RUC13 Para RUC13 Oper RUC20 More convective precip in RUC13 over land, less over ocean CAPE 0-12h prec

34. 34 Typical CAPE effect from improved soil moisture and moisture analysis in RUC13

35. 35 MODEL PHYSICS CHANGES FOR RUC13 -Land-surface model: When ground temp < 0°C, vapor deposition on ground now based on ice saturation, not water * Diminishes unrealistic widespread nighttime fog formation, especially over snow cover Evident in real-time comparisons between RUC13 and oper RUC20 for visibility forecasts, especially at night.

36. 36 Test of Ice Saturation for Surface Latent-heat Flux 13 Apr 2004 - 18h fcst valid 06z Oper RUC20 Revised RUC20 Cloud-water mixing ratio at lowest model level Reduction in fog, especially over snow. Much improved visibility forecasts, avoids high bias

37. 37 RUC13 Model Physics Changes (cont) -Convection (Grell-Devenyi scheme) * Empirically estimated ensemble weights to improve quantitative precipitation forecasts RUC20- dCAPE/dt - Kain-Fritsch – CAPE relaxation - low-level vertical velocity - moisture convergence RUC13 - adds Arak-Schu scheme for cloud work fn - no KF over ocean, reweighted all closures Addition of convection-inhibition ensemble members CAPE dp – 25, 75, 125 mb – fn ( TKE) Does not produce significant outflows (slightly more in RUC13) Still no shallow convection in RUC13 Grell-Dev scheme Still much less sounding modification than NAM/BMJ Eliminated extreme surface drying showing up in certain situations

38. 38 RUC13 Model Physics Changes (cont) - Microphysics [NCAR and FSL] Overall goal: To incorporate best understanding of “warm-rain” and mixed-phase processes important for cold-season aviation operations (inflight icing, pre-takeoff deicing requirements, low ceiling, visibility) into operational NWP models. Major changes for RUC13 Dropsize distribution now transitions between drizzle and rain Replace Kessler with Barry-Reinhardt autoconversion (collision-coalescence cloud droplets to rain)  Correct ice-particle accretion (ice  snow more readily)  Ice particle fall speed no longer set to 1 m/s

39. 39 RUC13 Model Physics Changes (cont) * - Snow - Diagnosis of snow-particle size distribution - Operational RUC20: - Depends on snow mixing ratio - RUC13: - Depends on temperature  - Graupel - Growth/depletion via deposition/sublimation: Change size distribution from gamma to inverse exponential - Results in less graupel

40. 40 EAST COAST BOMB 22-23 JAN 2005 RUC13 9h Forecast for 0900 UTC 23 Jan Sfc wind, temp, 3-h pcpn Precipitation type Snow Rain X-sec

41. 41 HYDROMETEOR CROSS SECTIONS RUC13RUC20 - OPER CLOUD-WATER MIXING RATIO (qc) Atlantic MA ME RUC 9h Forecasts for 0900 UTC 23 Jan

42. 42 Hydrometeor Mixing Ratios (Cont) RAIN WATER (qr) RUC13 Atlantic MA ME Oper - RUC20 More rain in RUC13 – different rain processes RUC 9h Forecasts for 0900 UTC 23 Jan

43. 43 Hydrometeor Mixing Ratios (Cont) CLOUD ICE (qi) RUC13 Oper RUC20 Atlantic MA ME Less ice in RUC13 – modified ice-particle accretion RUC 9h Forecasts for 0900 UTC 23 Jan

44. 44 Hydrometeor Mixing Ratios (Cont) SNOW (qs) RUC13 Oper RUC20 Atlantic MA ME More snow in RUC13 – modified ice-particle accretion RUC 9h Forecasts for 0900 UTC 23 Jan

45. 45 Hydrometeor Mixing Ratios (Cont) GRAUPEL (qg) RUC13 Oper RUC20 Atlantic MA ME Less graupel in RUC13 More supercooled liquid water (important for icing/aviation) RUC 9h Forecasts for 0900 UTC 23 Jan

46. 46 RUC13 Model Physics Changes (cont) - Coupling between physics processes * Reduction of cloud attenuation coefficients for shortwave and longwave transmission to match WRF/MM5 * Effects of microphysics changes (more ice fallout - results in less Cirrus, Cirrostratus) on radiation budget and surface temperature forecasts * Detrainment of hydrometeors from convection to the grid scale – More latent heating due to freezing on explicit scale in convection

47. 47 2004/05 - 13km RUC testing Fall 2003 - 13km RUC model real-time testing started - Initialized from 20km analysis, run 4x daily May 2004 – Full 1-h cycle testing started Evaluation - FSL verification – against raobs, sfc, precipitation obs - 13km RUC vs. backup RUC – raob - NOAA New England High-Resolution Temp Project - ETL, NSSL evaluation - June-Sept 2004 - Experimental use in NWS Central and Eastern Region WFOs - July 2004 – current - Available via LDM from FSL (contact Tracy Smith) - NCEP pre-implementation testing started – Jan 05

48. 48 RUC20 RUC13 6h precipitation forecast Valid 18z Thurs 30 Sept 2004 RUC20 (oper) vs. RUC13

49. 49 Obs radar - 0235z - 31 Mar 05 RUC13 9h fcst - valid 03z- 31Mar05 RUC20 9h fcst - valid 03z - 31 Mar 05

50. 50 Obs radar - 0115z - 12 May 05 RUC13 12h fcst - valid 03z- 12May05 RUC20 12h fcst - valid 03z- 12May05

51. 51 CAPE Oper RUC20 Exp RUC13 Improved moisture analysis in RUC13 – pseudo-RH instead of ln-qv - use of GPS PW Thursday 21 April 2005 – 12z

52. 52 Thursday 21 April 2005 – 12z Oper RUC20 Exp RUC13 Improved moisture analysis in RUC13 – pseudo-RH, GPS PW Radar summary – 1400z

53. 53 RUC20 vs. RUC13 precipitation forecast (12h) verification Winter –Jan-Feb 05 (verified w/ NCEP 24h precip analysis) RUC13 -higher equitable threat score - bias closer to 1 Why? - Improved initial conditions - Improved precip physics - Higher horizontal resolution

54. 54 RUC20 vs. RUC13 precip forecast – 24h period ending 12z - 12 Feb 2005 (verified w/ NCEP 24h precip analysis) RUC20 RUC13 obs (inches)

55. 55 Ceiling analysis: comparison of Operational RUC (top) with the NCEP RUC13 (bottom) Scale (AGL): x 1000 ft Analyses at 1800 UTC on 9 May

56. 56 Observations: cigs <= 5000 ft only Observations: cigs <= 500 ft only Point observations at 1800 UTC 9 May, for different ceiling heights, to compare to previous figure. 4 areas of lower cigs are apparent. -New England: Better coverage of lower cigs by RUC13 looks good. - Northern Plains/Midwest: more coverage of lower cigs in RUC13, tho similar coverage for very lowest categories. Again, obs for aob 5000 ft cigs support RUC13. - West Coast: both have lower cigs in LA area. More coverage on RUC13 in intermountain west, not sure if this is overdone or not. -Northeast TX: RUC13 looks better here. Observations: cigs <= 3000 ft only

57. 57 Ceiling 1h forecasts: comparison of Operational RUC (top) with the NCEP RUC13 (bottom) Scale (AGL): x 1000 ft Forecasts at 1900 UTC on 9 May

58. 58 Ceiling 3h forecasts: comparison of Operational RUC (top) with the NCEP RUC13 (bottom) Scale (AGL): x 1000 ft Forecasts at 2100 UTC on 9 May

59. 59 Focus on the Midwest and Northern Plains. Quite a few obs have cigs < 3000 ft, but only a few lower vis reports. It does appear that low clouds lurk JUST east of the eastern WI shoreline, like what was found in the RUC Operational run.

60. 60 Focus on eastern Texas. Cigs < 3000 ft are relatively widespread.

61. 61 VariableRUC20RUC13 Temp –s.d.East domain2.82.6 Temp – s.d.West domain3.63.3 Dewpoint – s.d.East domain2.92.7 Dewpoint – s.d.West domain3.53.2 12-h surface forecasts verified vs. METAR obs 25 Nov 2004 – 18 Jan 2005 - every 3 h (00z,03z,…21z) 20-km RUC (oper) vs. 13-km RUC - METARs in RUC domain (East/West of 100° W) - RUC13 uses 13km mini topography field derived from 3.3km topo

62. 62 Rawinsonde verification - wind - RUC13 vs. RUC20 -RUC13 provides improvement in 150-300 mb winds 20km 13km

63. 63 Rawinsonde verification - RH - RUC13 vs. RUC20 -RUC13 provides much better RH forecasts - With RUC13, 3h RH forecasts are more accurate than 12h RH forecasts (unlike RUC20)  GOES and GPS used (more) effectively, physics improved also. 20km 13km

64. 64 2005 - 13km RUC implementation plan February-May 2005 - Real-time parallel testing – EMC - Intercomparison RUC13 vs. RUC20 by EMC, FSL - daily inspection w/ graphics, statistics (raob, surface) - Transfer afterward to NCEP/NCO May-June 2005 - Real-time parallel testing – NCO - 4-week field test evaluation – NWS regions, AWC, SPC… - Retrospective testing (at NCEP) for cold season - Report on testing to EMC and NCEP directors ~ June/July 2005 - Operational implementation

65. 65 RUC13 output grids ~Same products/fields as for RUC20 and RUC40 –Add sfc-based CAPE to best-300mb CAPE 20km and 40km look-alike grids –Native (Note: 50 levels instead of 40) –25-mb isobaric and 88 2-d grids –25 2-d variables BUFR station time series output –nearly identical to Eta (Exception: 5 soil levels with RUC (out of actual 6) compared to 4 with Eta/NAM model) –RUC13 – station files with all output time (as with RUC20) -RUC13 – improved land/water matching (important for coastal stations) Continue RUC web forum for questions (under RUC web site) Continue RUC hot backup – transition to 13km

66. 66 2005- Implementation of 13km RUC in operations at NCEP Assimilation of new observations - GPS-precipitable water  improved moisture forecasts - METAR cloud/vis/current weather  improved ceiling and vis fcsts - Mesonet, RASS temperatures Soil moisture/temp nudging Improved moisture analysis – pseudo-RH instead of log q Model changes – New versions of - mixed-phase cloud microphysics (NCAR-FSL) - Grell-Devenyi convective parameterization - Revised radiation – cloud effects - Corrected treatment of frost formation  improved icing and convection forecasts, cloud/sfc temp, vis forecasts Improved post-processing – visibility, precip type, 20km/40km look-alike Hourly forecasts to be extended to 9h from 3h duration (at 01z, 02z, 04z, 05z, …) (12h forecasts to be continued at 00z, 03, 06… init times)