Summary of RUC13 code changes

Updated 28 June 2005

(Powerpoint at http://ruc.fsl.noaa.gov/ruc13_docs/RUC13ppt.htm )

`

      Date - NCEP operational implementation occurred at 12z 28 June 2005

      Higher resolution – from 20 km to 13km.

o      Full cycling continuing, now at 13km resolution, of all variables, including all atmospheric variables, including all hydrometeor types, and land-surface fields including multi-level soil temperature, soil moisture, and snow water equivalent and snow temperature.

o      Domain

       AWIPS grid number for RUC13 13km grid is 130.

       Dimensions – now 451 x 337 (previously 301 x 225 for RUC20)

       Grid length at 35N is 13.545km, exactly 2/3 of previous 20.317km at same latitude in RUC20.  Therefore, the RUC13 has 3 grid lengths for every 2 grid lengths in the RUC20.  RUC13 actual grid lengths are about 12.4 km at 45N, and about 10 km at northern domain.

       Domain area same as for RUC20.

       RUC13 still has same 50 isentropic-sigma hybrid vertical levels as in RUC20.

       Domain parameters available in http://maps.fsl.noaa.gov/fslparms/13km/MAPSCON

o      Fixed fields

       Topography and mini-topography (for 2-m temp/dewpoint) calculated from WRF-SI 30-second data.

       Land-use determined from predominant 1km USGS 24-category data in each13km RUC grid-point area.  A smoothing technique is applied to minimize land-use variations over land, but not to land-water coastlines.

       Soil texture from 1km USGS data

       Roughness length calculated from 1km USGS data

       All fixed fields and text land-use and soil type descriptions are available via http://ruc.fsl.noaa.gov/fslparms/13km . 

      Forecast duration

o      Extended to at least 9h (instead of current 3h) every hour, continuing out to12h forecasts every 3rd hour (00z, 03z, 06z). 

      Analysis

o      New observation types added with RUC13

       GPS precipitable water

       METAR cloud/visibility/current weather

       Mesonet (temperature, dewpoint, surface pressure)

       RASS virtual temperatures

       Deferred until later pending resolution of quality issues

      Mesonet winds

      915 MHz profilers, mostly near west and east coasts)

 

o      Use of surface data

       Includes PBL-based assimilation of surface observations.  This was implemented into the operational RUC20 in late Sept 2004, but its positive effects in the warm season will become evident for the first time in spring and summer 2005, including in the RUC13.

o      Cloud/hydrometeor assimilation

       Use of METAR cloud/visibility/current weather observations.  (See Benjamin et al. 2004 – AMS ARAM paper)

      METAR indicators of BKN, OVC, and VV assimilated as a cloud base.

      METAR indicators of FG and BR assimilated as cloud water near surface. 

      Depth of assumed cloud layer above cloud base increased if current weather indicates precipitation.

       Consistency with terrain: Cloud base from METAR is assumed to be locally constant with height and can intersect with nearby higher terrain.

       GOES cloud top pressure/temp treatment

      More aggressive use of GOES cloud data (including absence of cloud) for building and clearing down to surface if zenith angle cos > 0.30 or if over water.

      Continue to avoid use of GOES cloud data for small convective clouds, possible level assignment for marine stratus, and if model skin temperature is close to or colder than the cloud-top temperature.

      Assign GOES CTP values to –2/+2 (previously –1/+1) 13km grid points to ensure more consistent GOES treatment and despeckling.  Median values of nearby GOES cloud values around each grid point are used. 

      Clearing above cloud top now done correctly – bug fixed.

o      Soil moisture/temperature nudging – Uses analysis-background difference (driven largely by 1h error in 2m temp and dewpoint) to allow a slight change in soil temperature or moisture at the top 2 soil levels.   This is only allowed in daytime with no precipitation or clouds.  This nudging has been effective in avoiding soil moisture drift in RUC tests at FSL and NCEP.

o      QC hierarchy of different observation types set so that mesonet obs cannot corroborate each other.   Also, use higher obs errors for mesonet obs than for METARs.

o      Moisture analysis control variable changed from ln (qv) to pseudo-RH (RH relative to background saturation/temperature field). 

      Unified variational moisture treatment, with in situ and precipitable water (PW) observations used simultaneously to calculate 3-d moisture analysis increment.

o      Includes raob PW values implicitly through raob moisture profile

o      Includes GPS PW obs (in addition to GOES)

o      Includes new moisture QC techniques

       Compares rawinsonde PW with nearby GPS PW values and flags entire raob profile if raob PW varies substantially from nearby GPS stations  (Gutman et al. 2004 IOS AMS conf)

       Also checks GPS PW obs to see if more than 25% are found to exceed background value by at least 25%, in which case all GPS PW obs are flagged.   This presumes that some problem occurred in the GPS PW processing.

o      PW innovations calculated using new PW-RH variable developed for RUC13 analysis, which exhibits much better spatial coherence than PW itself over regional variations of terrain elevation.

      Model

o      Digital filter initialization (DFI)

       Uses more time steps, consistent with same time period previously used in DFI period in 20km version.

       Modifies 3-d qv (water vapor mixing ratio) at end of DFI to ensure that the 3-d RH analysis field is maintained with modified temperature (actually qv) post-DFI. This fix avoids evaporation of hydrometeors due to DFI-produced warming and results in an improved 1-h precipitation forecast

o      Increased diffusion (0.0015 to 0.0025) in stratosphere in top 5 levels (385-500K) of model domain

o      More efficient version of positive definite horizontal advection scheme (Smolarkiewicz scheme).

o      Diagnosis of vertical motion is corrected, avoiding exaggerated effect of latent heating in convective and stratiform precipitation areas.  This problem was only in the diagnosed output field but not in the actual vertical motion in the RUC model.

      Model physics

o      Convective scheme (Grell Devenyi) – revised version in RUC13

       Added ensemble-based version of capping convective inhibition term for 25/75/125mb depths.  Also added 5th ensemble closure to previous 4  closures, using a new modified Arakawa-Schubert formulation.

       Improved quantitative precipitation forecasts via revised ensemble closures, empirically estimated weights, addition of convection-inhibition members.

       Eliminated extreme surface drying showing up intermittently in RUC20.

o      Bulk mixed-phase cloud microphysics (NCAR/RAP and FSL)

       Overall results – averages 10-15% more precipitation, less ice, more drizzle.

       Drizzle approximation added by using lower fall speed when rain water mixing ratio is small.

       Collision-coalescence from cloud water to rain changed (Barry-Reinhardt replacing Kessler formulation)

       Correction ice-to-snow accretion, resulting in less ice and more snow.

       Snow-particle size distribution now dependent on temperature in RUC13, rather than on snow mixing ratio (RUC20).

o      Shortwave and longwave radiation

       Change to cloud attenuation for shortwave and longwave transmission.  Main effect is lower attenuation of short-wave radiation through clouds (now using Dudhia revised coefficients from late 1990s, not previously used in RUC).

o      Land-surface processes

       Test for ice saturation added for surface latent-heat flux.  Allows proper frost deposition and gets rid of problem with excessive areal fog cover (cloud-water mixing ratio) prevalent over snow cover at night in RUC20.

      Post-processing

o      20-km and 40-km look-alike GRIB files produced from RUC13 (for native, isobaric, and surface files).

o      New variables for isobaric and surface files

       Surface-based CAPE – added to previous best CAPE

       Surface-based CIN – added to previous CIN corresponding to best CAPE

       Skin temperature – based on 0-2mm soil temperature

o      Visibility

       Vertical averaging of hydrometeor mixing ratios in lowest 2 levels rather than maximum value in lowest 4 levels –  reduces excessive areal coverage by very low visibility.

       Includes day/night visibility coefficients from Roy Rasmussen (NCAR).   A moderate snow in the daytime will be interpolated as a light snow at night, even though the snowfall is the same.

       Effect of wind shear to reduce attenuation from RH term if near-surface vertical wind shear (between levels 1 and 5, about 22 mb) exceeds 4 m/s.  (from Evan Kuchera M.S. thesis)

       Affected strongly by correction of frost deposition in RUC model.

o      Tropwnd.f (calculates tropopause level products)– Fix to possible (but extremely rare) divide by zero.

o      Revised smoothing factors for isobaric output, especially at 40km resolution.

o      Use of 13km topomini.f (for calculation of 2-m temp/dewpoint) from 3.3km USGS/WRF data set.  Topomini field allows reduction of 2m T/Td to improve accuracy at METAR/surface stations.

o      Fix to ensure hydrostatic consistency between heights and virtual potential temperature (qv ) to work around inconsistency from output point in model time step

o      Precipitation type algorithm changed

       Rain/drizzle - indicated now for 1.e-3 instead of 1.e-2 cm/h rate. This increases areas of drizzle and freezing drizzle (FZDZ) or FZFG . Was too limited before, and definitely inconsistent with snow coverage.
Graupel vs. RA/FZ types - Implemented a choice based on predominant p-type. Now use graupel instead of rain if graupel rate > rain rate. Choose rain (or FZ) if rain rate > 4x graupel rate. Was previously getting almost exactly the same coverage for FZ and IP, so this change produces now a reasonable overlap but also a better distinction.

o      BUFR station time-series data continued, but with improved land-water match

       RUC BUFR soundings matches format used for Eta/NAM soundings except that the RUC has 5 soil levels (out of actual 6) compared to only 4 levels for Eta/NAM.

o      Interpolation of RUC13 analysis fields (2m T, 2m Td, 10m u/v/wind gust, visibility, sfc pressure) to 5-km NDFD grid/topography to produce background for Real-Time Mesoscale Analysis

      Previous documentation on RUC20

o      RUC20 Technical Procedure Bulletin – RUC20 TPB link at http://ruc.fsl.noaa.gov

o      2004 Monthly Weather Review articles on RUC – available from RUC/MAPS pubs link at http://ruc.fsl.noaa.gov

       An Hourly Assimilation-Forecast Cycle:  The RUC

       RUC model – Mesoscale Weather Prediction with the RUC Hybrid Isentropic-Terrain-Following Coordinate

o      Recent RUC-related conference papers – also available from RUC/MAPS pubs link at http://ruc.fsl.noaa.gov