NMQ/Q2 System Updates

NOTICE:  7/30/2013

This log will no longer be updated.  NMQ/Q2 is now MRMS/Q3 and has a new update log.  If you wish to have access to this new log, please contact Carrie at Carrie.Langston@noaa.gov

05/30/2013 1550-1620 UTC - 3D Mosaic, Seamless HSR Mosaic

Three Dept of Defense WSR-88D radars have been added to 3D Mosaic and Seamless HSR Mosaic.  A table listing the radars and times applied is below.

Radar

3D Mosaic (UTC)

Seamless HSR Mosaic (UTC)

KEOX

1610

1620

KHPX

1610

1620

KMBX

1550

1620

Five more DoD radars will be added to the system as their Level II data become available.

04/10/2013 - Hybrid Scan Blockage, Radar Coverage

The Hybrid Scan Blockage and Radar Coverage products have been removed from the system.  They are no longer needed for evaluating radar QPE quality since similar information is contained in the Radar Quality Index product.

04/02/2013 1650 UTC - 3D Mosaic

Removed Canadian radar XSS until its electrical interference issues can be resolved.

04/01/2013 ~1510 UTC - QCNN for Canada

Most Canadian radars switched from winter to summer mode.  Updated QCNN to use summer clutter maps for Canada radars.

03/29/2013 2140 UTC - Seamless HSR

This Seamless HSR update deactivates an old approach for mitigating radar QPE errors in partially blocked areas on the hybrid scan by combining data from unblocked higher tilts with those from the lower but partially blocked hybrid scan tilt.  The old approach sometimes introduces overestimation in bright band areas and underestimation when the higher tilt overshoots the precipitation process.  A new approach based on QPE climatology is found to be more effective and is now replacing the old approach.

03/13/2013 - 3D Mosaic - Probability of Warm Rain

The Probability of Warm Rain (POWR) product computed in 3D Mosaic has been removed and replaced by a new POWR.   The older POWR was derived from an objective analysis of the 0/1 (no/yes) tropical rain identification at each radar site based on vertical profiles of reflectivity.  The new POWR is derived from model fields including 850 - 500 hPa lapse rate, freezing level height, the mean 1000 - 700 hPa relative humidity, etc.  This new POWR can be viewed on NMQ QVS under the “Model” tab.

03/05/2013 1700 UTC - QPEs in XMRG HRAP

The program used to convert Q2 QPEs to the XMRG HRAP format has been updated to remove a truncation error (a.k.a., low precip thresholding).  Previous products in this format were truncated at 1 mm.  The update is effective starting at 17Z for Q2 Radar-only products and 14Z for Q2 Gauge-corrected products.   The following images show the effects of removing the truncation.  

1H Q2 Radar-only QPE at 3/5/2013 12Z with (top) and without (bottom) truncation at 1mm

02/05/2013 1800 UTC - Q2 Gauge-correction

Updated Q2 Gauge-correction (a.k.a., LGC) to correct bug in data coverage values.  When applied, the CREF QC mask set no-coverage values to missing.  This bug is now fixed.  See the figures below for a before and after comparison.  Note that …

1. No-coverage values are determined by radar coverage.

2. Affected products are Q2 Gauge Adjusted Radar QPE and Q2 QC Gauge-only QPE.

Example LGC image before and after the bug fix.

02/04/2013 - NMQ/Q2 Products Removed

Real-time processing has stopped and all archive data deleted for the following products:

3D EDR

3D EDR Confidence

Hybrid Scan Blockage*  

Radar Coverage*

*The Radar Quality Index can be used instead.

01/22/2013 2145 UTC - Seamless HSR

SHSR mosaic scheme updated for bug fix.  The scheme originally used the grid cell corner latitude and longitude coordinates to mosaic data, which resulted in a slight offset.  This was corrected by using the grid cell center instead.

01/03/2013 2200, 2250 UTC - Seamless HSR

Updated nonstandard blockage mitigation (NSBM) table for KCAE (2200 UTC) and KLTX (2250 UTC).

12/13/2012 1935-1950 UTC - Bloom/AP QC

Updated algorithm configuration to apply Bloom/AP QC logic only during the peak bird migration seasons (Apr-Jun, Sep-Nov).  Also reduced region where Bloom/AP logic is applied from ConUS to the following lat/lon bounds:  

northwest corner:  40N, -105W        southeast corner:  20N, -70W

12/03/2012 1730-1735 UTC - QCNN for Canada

Updated Winter clutter maps for Canadian radars.  These new maps will more aggressively remove ground clutter.

11/02/2012 1440 UTC - QCNN for Canada

Most Canadian radars switched from summer to winter mode.  Updated QCNN to use winter clutter maps for Canada radars.

10/08/2012 1535 UTC - Seamless HSR Mosaic

Input for the SHSR Mosaic was switched from an older set of servers to a new set.  The older servers were running an outdated version of the single radar Seamless HSR algorithm.  This means that several SHSR updates discussed in previous log entries did not take affect as originally stated.  Specifically, posts dated 8/20/12, 7/20/12, 6/13/12, and 5/29/12 were not valid until now (10/8/12).

08/21/2012 2035-2055 UTC - Bloom/AP QC

Previous versions of Bloom/AP QC performed QC on the 3 lowest tilts of a volume only.  This new update applies QC to the entire volume.

08/20/2012 1915 UTC - Seamless HSR Mosaic

Increased the horizontal scale factor L from 50 to 100km in the seamless HSR mosaic weighting function (w = exp(-d^2/L^2)), where d is the distance from a give grid cell to a radar.  The change will help mitigate discontinuities across boundaries mid-way between neighboring radars.  See post from 5/29/2012

Correction:  See post from 10/8/2012

07/20/2012 1440 - 1500 UTC*, 2025 UTC - Seamless HSR

1440-1500 UTC*: Updated the Seamless HSR program to fix bugs, clean up code, and improve Non-Standard Beam Mitigation (NSBM) logics.  The NSBM cross-tilt smoothing schemes were modified to include beam heights in the weighting function.  The new weighting function will give a larger weight to the lower tilt than to the higher tilt at tilt boundaries in the hybrid scan.

2025 UTC: Updated NSBM reference tables for KATX,  KJAX, KGWX, and KMQT

*radars KABR - KBGM updated earlier at 7/19/2012 2140 UTC

Correction:  See post from 10/8/2012

06/21/2012 1350-1400 UTC - Bloom/AP QC

Updated Bloom/AP QC algorithm to use a no_precip flag.  If no precip is found in a volume, then no_precip = 1.  This flag is written as part of the algorithm’s netCDF output and is used by other programs (e.g. Seamless HSR) to speed up processing.

Also added logic to handle volume scans with missing tilts.

06/13/2012 1630-1715 UTC - Seamless HSR

Non-Standard Beam Mitigation (NSBM) tables have been added to Seamless HSR.  Tables are not available for ever radar/VCP, but a majority of radar/VCP pairs have an associated table.

Correction:  See post from 10/8/2012

05/29/2012 1510-1610 UTC - Seamless HSR

A small subset of radars were updated at 5/23/2012 1730 UTC for testing.  The large majority of radars were updated on 5/29/2012.  The following changes were made:

  1. Increased the horizontal scale factor L from 50 to 100km in the seamless HSR mosaic weighting function (w = exp(-d^2/L^2)), where d is the distance from a give grid cell to a radar.  The change will help mitigate discontinuities across boundaries mid-way between neighboring radars.

        Correction: The scale factor was increased for 3D Mosaic (not Seamless HSR mosaic)

  1. Modified the VPR correction scheme
  1. added constraints in calculations of the VPR slope in ice regions to reduce random errors in the slope.
  2. improved delineation of convective precipitation to include areas surrounding storm cores.
  3. limited VPR correction to bright band (BB) and ice regions in cool season and to BB areas in warm season.
  1. Added logic to more speedily produce products when no precip is observed.

Correction:  See post from 10/8/2012

05/02/2012 0100 UTC - NMQ transition from RUC to RAP

NCEP changed the Rapid Update Cycle (RUC) to Rapid Refresh (RAP) model.  This change occurred at 5/1/2012 12Z.  From 5/1/2012 12Z to 5/2/2012 01Z the NMQ real-times system did not have up-to-date model input.  NMQ was switched to the RAP in time for the 5/2/2012 01Z model run.  More information about the RUC to RAP transition at NCEP can be found here...  http://www.nws.noaa.gov/os/notification/tin11-53ruc_rapaae.htm

04/26/2012 - 04/30/2012 - Level III Dual Pol processing

At this time, NMQ processes 46 radars of Level III dual pol data.  Due to a server failure 26 of these radar were intermittently available.  Processing has been restored.  Affected radars are...

KAMA  KATX  KBOX  KDDC  KFFC  KGRR  KICT  KLGX  KMHX  KOKX  KPBZ  KRIW  KSOX

KAPX  KBLX  KCLE  KEMX  KGLD  KHTX  KIWA  KLOT  KNQA  KOTX  KPDT  KRTX  KVNX

02/15/2012 2100-2300 UTC - Bloom/AP QC

A new automated algorithm for identification and removal of bloom and anomalous propagation (AP) clutter echoes was implemented.   The new algorithm uses global image processing techniques and is based on two main characteristics of these echoes:

– The bloom and AP echoes are most prevalent in the lowest tilt.  The area of these echoes reduces significantly in size in the 2nd and higher tilts

– The bloom and AP echoes appear “noisier” overall than weather echoes.

Reference: Tang et al. 2011 

01/19/2012 1425 UTC - NMQ

KBBX fixed and added back to NMQ processing.

01/17/2012 1535 UTC - NMQ

Temporarily removed KBBX from NMQ processing.  Message from NWS:

000

NOUS66 KSTO 131545

FTMBBX

Message Date: Jan 13 2012 15:45:00


KBBX EXPERIENCING FAULTY REFLECTIVITY DATA...PART(S) ARE ON ORDER. COMMS LINES HAVE BEEN TERMINATED TO AVOID RELEASE OF BAD DATA. IT MAY BE MON OR TUE BEFORE PARTS ARRIVE.   JHM/STO  AFTER TRANSMISSION OF THIS FTM...LINES WILL BE DISCONNECT
ED.

Bad data continued to flow via Level II, however.  KBBX will be restored to NMQ once repairs are in place.

01/12/2012 2215 UTC - QCNN for Canada

Updated winter clutter maps for two Canadian radars: XPG and WHK.  

Previous clutter maps, derived from limited data samples, were not sufficient to remove the majority of persistent ground clutter.  The new clutter maps provide more effective clutter mitigation.

01/03/2012  1445 UTC - 3D Mosaic

KPIX added back to 3D Mosaic since the radar data, despite the noise, is providing useful coverage (in addition to WSR-88Ds) over the Napa and Russian River areas in California.

12/22/2011  0020 UTC - 3D Mosaic

Removed KPIX from 3D Mosaic due to extensive noise in the radar data.

11/01/2011 1340 UTC - QCNN for Canada

Most Canadian radars switched from summer to winter mode.  Updated QCNN to use winter clutter maps for Canada radars.

09/30/2011 1500-1510 UTC - NMQ

Starting at 1500 UTC, 30 Canadian radars were added to the NMQ product suites.  Integration of these radars will significantly improve data coverage in the northern part of the NMQ

domain. Programs affected are 3D Mosaic, Q2 Radar-only, Q2 Gauge-correction, and Radar Coverage,

09/09/2011 2000 UTC - Seamless HSR

 Adjusted the RQI to decrease more slowly above the freezing level.  The exponential function parameter (H_OVSHTNG_SCALE) was increased from 1500 to 2500 meters

09/03/2011 0240 UTC - 09/06/2011 1350 UTC - Tropical VPR ID & Q2 Radar-only

During this time period several changes were made to the NMQ tropical ID parameters to improve performance during TS Lee.  Changes included the following:

- Lowering the requirements for the ID of warm rain processes

- Changing the tropical Z-R used in Q2 Radar-only from standard to super tropical (9/3/2011 1450 UTC - 9/5/2011 2130 UTC)

- Switching off use of the tropical ID weight field in Q2 Radar-only (see post from 6/1/2011 1930 UTC)

The current status of NMQ is as follows:

- Tropical VPR ID

  1. maximum dBZ threshold is 25. When the maximum reflectivity in the instantaneous global VPR is less than this value, the tropical rain identification is prohibited.
  2. mean dBZ below brightband threshold is 20When the mean reflectivity below the bright band bottom in the instantaneous global VPR is less than this value, the tropical rain identification is prohibited.

- Q2 Radar-only

  1. tropical rain weight field switched off in Q2 Radar-only

The weight field was initially introduced in June 1, 2011 to minimize rainfall overestimation in the trailing stratiform precipitation of mesoscale convective systems.  Although the weighting scheme was found to cause rainfall underestimation in tropical systems.  Therefore the current tropical rain weight field is deactivated, and a new environmental data based tropical rainfall probability function will replace this weight field.

06/02/2011 2210 UTC - Q2 Radar-only

 Updated Q2 Radar-only to incorporate the tropical ID weight field (TRWGT) produced in 3D Mosaic.  When a grid cell is flagged as tropical, the precip rate is computed as a blend of tropical and stratiform Z-Rs.  The exact blend is determined by the grid cell's TRWGT value.

06/01/2011 1930 UTC - 3D Mosaic

 Updated the tropical ID (TRID) settings in 3D Mosaic.  Turned off TRID QC based on distance weighting function (see 5/27/2011 update).  Also changed related length scale from 0.5 to 2.0 km.  The original length scale resulted in a too steep tropical flag weight field.  Note the weight field will be used in Q2 Radar-only (update forthcoming).

05/31/2011 2350 UTC - Tropical VPR ID

 Updated VPR tropical ID (TRID) logic with the following changes:

  1. Changed VPR maximum dBZ threshold from 20 to 30
  2. Added logic to check bright band (BB) bottom height for each radar where a BB is identified.  If the BB bottom height is within 2.65km ARL of radar, then no TRID is allowed.
  3. Added logic to check mean reflectivity below BB bottom height.  If the mean dBZ below the BB bottom is less than 35 dBZ, then no TRID is allowed

All updates work to mitigate erroneous tropical IDs and reduce overestimation in NMQ QPEs.

05/27/2011 2155 UTC - 3D Mosaic

 Updated the tropical ID (TRID) logic in 3D Mosaic to mitigate the sometimes overaggressive identification of warm rain processes.  Propagation of the TRID is now restricted by the BB bottom and by a distance-based weighting function that also takes into consideration how many nearby radars have been flagged "tropical".

04/14/2011 1500-1550 UTC - Q2 Radar-only

 Update of Q2 Radar-only from 4/13/2011 introduced two problems... 

        1. a slight netCDF variable name change for any QPE longer than 1-hour

        2. incorrect threshold setting for QPE QC.

The update today restores the previous netCDF variable names and corrects the threshold setting which allowed more light precip into the 1-72 hour QPEs from 4/13/2011 2030 UTC - 4/14/2011 1550 UTC.  Update valid for US and US+Canada QPEs.

04/14/2011 1250-1430 UTC - 3D Mosaic

 Updated 3D Mosaic algorithm to the latest version.  The updated code included more support for CF-compliant netCDF (presently turned off), and the addition of 4 variables to the 2D derived multi-variable netCDF (i.e., mosaic2d_nc).  They are ...

    Layer Composite Reflectivity - Low [0-24 kft]

        Layer Composite Reflectivity - High [24-60 kft]

        Layer Composite Reflectivity - Super [33-60 kft]

        Storm Top [30 dBZ]

These variables were already available in HMRG binary format. 

And finally, the parameter controlling the history size of VPR Tropical IDs was decreased from 30 to 15-min.  Update valid for US and US+Canada 3D Mosaic.

04/13/2011 2030-2130 UTC - Q2 Radar-only

 Updated Q2 Radar-only algorithm to latest version.  The updated coded included support for CF-compliant netCDF (presently turned off), and a minor bug fix for QC'g the QPE fields.  Update valid for US and US+Canada QPEs.

02/23/2011 1930 UTC - SHSR Mosaic

 NW coordinates of the single radar SHSR products (algorithm input) were incorrectly referenced when being mosaicked.  This caused a shift in the resulting mosaic field of half a grid cell to the north and west.  Bug now fixed.

02/21/2011 - 02/22/2011 - Canada single radar processing

 Retooled Canada single radar processing to be more reliable.  Updated half of Canadian radars on 2/21/2011 ~2210 UTC.  Updated second half of Canadian radars on 2/22/2011 ~1440 UTC.

02/08/2011 2000 UTC - Seamless HSR

 Updated Seamless HSR to handle cases where 0 degC RUC height is missing or below ground.  Previously, the missing or below ground height caused an error in the RQI calculation (setting the RQI too high).  This is now fixed.

06/04/2010 2100-2200 UTC - Polar VPR & Seamless HSR

 Updated Polar VPR to correct minimum dBZ threshold bug.  Also changed minimum dBZ threshold for VPR input from 10 to 20 dBZ.

Updated Seamless HSR (SHSR) lower level blockage threshold from 0.0% to 10.0%

05/25/2010 0200 UTC - Polar VPR

Reverted the minimum dBZ threshold for VPR input back to 10 dBZ.  Found bug in Polar VPR algorithm that causes issue only when threshold not equal to 10.  Will debug and update Polar VPR at later date.

05/21/2010 2305 UTC - Polar VPR

Changed the minimum dBZ threshold for VPR input from 10 to 20 dBZ.

05/20/2010 1825 UTC - 3D Mosaic

The 3D Mosaic tropical ID has been modified to increase stability of tropical precip flags.  If a radar has 3 consecutive tropical IDs or more, only then are tropical flags are applied to the precip field based on that radar's observations.

05/03/2010 2300 UTC - 3D Mosaic

The 3D Mosaic tropical ID parameter for minimum dBZ changed from 30 to 35 dBZ.

04/06/2010 1850-1935 UTC - WDSS-II (QCNN)

Installed new WDSS-II build for single radar processes.  This QCNN update includes recent NN training.

03/22/2010 1420-1520 UTC - WDSS-II (QCNN)

 Installed new WDSS-II build for single radar processes.  The purpose of this is to update QCNN with improved bloom logic/NN training.  During the update, radar data dropped out for a short time but returned ~1515 UTC.

03/12/2010 1740-1750 UTC - QCNN

The clear air VCP check within the QCNN logic was turned off.  Previously, reflectivity data associated with a clear air VCP and warm surface temperatures (>4 degC) would be removed.

 

03/11/2010 17-1800 UTC - Polar Processing

The Polar QC Post-process (QCPP) algorithm was removed from the NMQ polar processing sequence between 1700 and 1800 UTC.  QCPP performed linear interpolation across small data gaps in both azimuthal and radial directions. 

03/08/2010 2355 UTC - 3D Mosaic

The VCP / RUC surface temperature check has been turned off for 3D Mosaic.  Previously, if a radar was in clear-air mode and the surface temperature was warm enough, 3D Mosaic would remove all reflectivity for that radar.  This is no longer occurring.  Change is effective starting 3/8/2010 2355 UTC.

09/23/2009 1205 UTC - 3D Mosaic

The 3D Mosaic tropical ID parameter for minimum dBZ changed from 40 to 30 dBZ.

08/28/2009 2130 UTC - 3D Mosaic

The 3D Mosaic tropical ID parameter for minimum dBZ changed from 35 to 40 dBZ.

08/21/2009 1400 UTC - New RUC Fields

Added RUC precipitable water (PWAT) field and precipitation efficiency (PE) to NMQ QVS.  Note PE is derived from RUC RH and PWAT fields.  PWAT and PE are available starting at 8/14/2009 0600 UTC.

08/10/2009 1450 UTC - 3D Mosaic

The minimum required dBZ for a grid cell to be flagged as tropical was increased from 30 to 35 dBZ.

08/10/2009 1430 UTC - Tropical VPR ID

Changed the Tropical VPR ID parameter to allow the lowest levels of the VPR to be used in the ID process (N_CUT_LOW_PTS:  0).  Previous settings excluded the lowest two VPR levels.  The VPR type used for tropical ID was made a configurable parameter.  Previously, the hourly averaged global VPR was used as input.  Now the instantaneous global VPR is used instead.

07/29/2009 - 08/07/2009 - Polar VPR

Significant changes were made to the Polar VPR program during this time period.

 

  1. Resampling of super-res tilts:  New logic resamples all super-res tilts to legacy (1 deg. x 1 km) resolution.  These tilts are converted from dBZ to power, resampled via averaging, and then converted back to dBZ for VPR processing.
  2. Sampling of volume:  Previous versions of Polar VPR assigned dBZ values to VPR levels using nearest neighbor (using the center height of a radar bin), which caused missed sampling.  New logic fixes this issue by linearly interpolating between tilts. 

Below is an example of Polar VPR output before and after the updates.

Before

After

Note that during this update period, Polar VPR was updated several times with bug fixes and new logic.  VPRs after 8/7/09 0000 UTC have all the changes and fixes applied.

06/08/2009 2115 UTC - Q2 Radar-only

The tropical Z-R cap is now turned on.  The max allowable reflectivity value associated with a tropical flag is now 50 dBZ.

06/02/2009 2140 UTC - Q2 Radar-only

The Tropical Z-R changed from super tropical to standard tropical.

06/02/2009 1445 UTC - 3D Mosaic

The Tropical ID parameter controlling the history size of VPR Tropical IDs was decreased from 1-hr to 30-min.

06/01/2009 0000 UTC - 3D Mosaic

The Tropical ID logic in 3D Mosaic was turned on at 6/1/2009 0000 UTC.  The VPR Tropical ID history is set to 1-hr, the influence of radius is 120 km, the dBZ threshold is 30, and the Tsfc threshold is 10 deg. C.

05/18/2009 1900 UTC - Stage 2 Remap

Bug fix.  A referencing error caused Stage 2 data (as remapped to the NMQ grid) to have an offset to the NE.  Stage 2 data starting on 5/16/2009 will have this fix. 

05/06/2009 1700 UTC - BB Analysis

Bug fix.  A timing bug in the single radar bright band heights allowed invalid heights to enter the analysis.  This caused artificial low spots in the BB top and bottom heights (see image below).

02/05/2009 2100 UTC - 3D Mosaic

Bug fix.  Portions of the 3D Mosaic precip flag logic mishandled bright band (BB) top/bottom data.  The BB heights were treated as MSL instead of AGL.  As a result, the logic to detect convective precip may have been too aggressive.  This is now corrected.

02/05/2009 2035 UTC - BB Analysis

Bug fix.  BB Analysis generates a top and bottom bright band (BB) height by combining a 2D 0ºC height field with BB top and bottom height point data (one point per radar).  In such instances where no BBs were detected in the point data, the program would incorrectly write out final BB heights in MSL.  Heights should be in AGL.  This is now corrected.  Below are examples of BB top heights in MSL (incorrect) and AGL (correct).

(Heights are km - MSL)

(Heights are km - AGL)

01/15/2009 2000 UTC - WDSS-II

Updated WDSS-II installation on single radar processing nodes processing *Canada* data.  Main update to is to QCNN.  The bloom removal algorithm was turned off, -t was tuned for more Winter weather (from 10 to 0), and the NN was improved via additional training.

01/14/2009 1640-2010 UTC - WDSS-II

Updated WDSS-II installation on single radar processing nodes (US radars only).  Main update to is to QCNN.  The bloom removal algorithm was turned off, -t was tuned for more Winter weather (from 10 to 0), and the NN was improved via additional training.

01/12/2009 2010 UTC - Bright Band Identification

Updated BBID code to account for instances where there is no BB identified anywhere in ConUS.  Previously, when this occurred, the BB top and bottom fields would be missing.  Now they are generated only without radar-based adjustments.

12/09/2008 2305 UTC - UnQC'd CREF

Added KPIX data to tile 1 and 5 unQC'd CREF product.

12/08/2008 2220 UTC - 3D Mosaic

Added new radar to tile 1 and 5.  This radar (KPIX) is a TV radar out of .  Its volume scan consists of a single tilt varying between 0.0 and 0.6 deg.  Note the data is *not* QC'd.

11/04/2008 2000 UTC - Model Remap

Fixed bug in that caused the temperature height index levels to be stretched in the x-direction.  These indicies are used by PCP_FLAG for identifying convective flags and HDA products (SHI, POSH, MEHS) in 3D Mosaic.  See image below that shows the -10 deg. C indices before (left) and after (right) the fix.  All other model data were unaffected.

10/27/2008 1945 UTC - 3D Mosaic

Turned off the tropical ID algorithm in 3D Mosaic.  The precip flag field will contain no tropical (warm rain) flags.  Thus, Q2 rates and accumulations will not include a tropical Z-R. 

10/24/2008 1745 UTC - 3D Mosaic & Q2 Radar-only

The 3D Mosaic tropical ID parameter for minimum dBZ changed from 1 to 30 dBZ.

The Q2 Radar-only tropical Z-R/dBZ cap has been turned off (i.e., set to a very high number).

10/20/2008 1435 UTC - 3D Mosaic

Changed tropical ID radius of influence from 180 to 120 km.

10/15/2008 2145 UTC - 3D Mosaic & Q2 Radar-only

The 3D Mosaic tropical ID parameter for minimum dBZ changed from 30 to 1 dBZ.

The Q2 Radar-only warm rain Z-R changed from standard to super-tropical.

10/03/2008 1645 UTC - Q2 Radar-only

Updated Q2 radar-only logic to include a minimum accumulation requirement for 1-hr QPE.  If the 1-hr QPE for a given grid point is less than a certain threshold, then that accumulation is reset to zero.  Current threshold is 1.27 mm (0.05 inches).

10/01/2008 2205 UTC - 3D Mosaic

Changed minimum reflectivity threshold for tropical ID from 35 to 30 dBZ.

09/12/2008 1535 UTC - 3D Mosaic

Changed tropical ID radius of influence from 120 to 180 km.

07/31/2008 1930 UTC - Polar & 3D Mosaic

The polar processes and 3D Mosaic were updated today starting at 1600 UTC and ending at 1930 UTC.  The changes made during this updated are listed below

   -- Polar --

   - Modified SRC file format to allow for sparse grids (compression).

   - Updated QCNN:  Previously, polar data was being aligned/resampled to 1deg x 1km resolution before passing into QCNN.  This allowed QCNN to run on the data resolution for which it was trained.  Output from QCNN was in 1deg x 1km resolution.  With the update, the need to align/resample the data before QCNN is gone.  Instead, QCNN now does the following:

1. Reads polar field in its native resolution (legacy or super-res)

2. Remaps field to 1deg x 1km, if necessary.  Remapping is done via "recombination" (taking average).

3. Performs pre-processing, NN, and post-processing in 1deg x 1km resolution.

4. Sub-samples the 1deg x 1km precip confidence field to super-res if the input data was super-res (0.5deg x 0.25km)

5. Apply precip confidence field to reflectivity data in native resolution.

6. Write out QC'd reflectivity field in native resolution

   -- 3D Mosaic --

   - Modified algorithm to handle new SRC file format

07/10/2008 1530 UTC - Q2 Gauge-correction

Updated executable to correct referencing bug.  Radar-only QPE values in some parts of the program log file and in the Bad_Gauge_record products ("Radar V") were being incorrectly referenced.  The problem is now fixed.  Below is an example Bad_Gauge_record entry...

20080710.1200  Gauge: MTNW1  Lat: 47.07  Lon: -121.08  Gauge V:  3.05  Radar V: 0  % Pixcel over criteria  99.2188  Criteria:  2

No other products were affected by this bug.  Bad_Gauge_record products starting at 7/10/08 1200 UTC have the fix.

07/09/2008 1710 UTC - Tropical VPR ID   

Changed the Tropical VPR ID process to allow the lowest three points in the VPR to be used in the ID process.  The old logic excludes the lowest three points and could potentially cause some false positives (and rainfall overestimation).  The new logic makes the excluded number of the lowest points in VPR a configurable parameter.  ( Current setting= N_CUT_LOW_PTS:  0 )

03/28/2008 2030 UTC - 3D Mosaic

-  Updated 3D Mosaic with new surface precip type (rain/snow), tropical flag logic, and RUC data resolution

-NOTE: Between 2030 and 2120 UTC the new version of 3D Mosaic had a small bug in its new PCP TYPE logic.  This was corrected at 2125 UTC.

-  Surface Precip Type (PCP TYPE):  Previous logic determined rain/snow using the HSR and RUC surface wet bulb temperature.  New logic continues to use these field but also uses the RUC surface temperature field.  If the wet bulb temperature <= 0 deg C  AND  surface temperature <= 2 deg C  AND  HSR >= 5 dBZ, then flag as snow.  Otherwise, if HSR >= 10 dBZ, flag as rain.  See image below of a before/after comparison of the new logic.

-  Tropical Flag:  Previous logic used the PCP TYPE to restrict tropical flags in areas believed to be too cold (i.e., if type snow, then don't flag as tropical).  New logic uses the RUC surface temperature.  If surface temperature > 10 deg C, then tropical flags are allowed.

-  RUC fields:  3D Mosaic used a set of RUC data fields at differing resolutions (and domains).  All RUC fields used in 3D Mosaic have a 10-km resolution and cover the ConUS.

03/21/2008 1900 UTC - System / CA-US RRCT

- ~1900 UTC: Switched US-Canada RRCT input from old polar processing nodes to new processing farm.

03/20/2008 - System

- Canada prototype tile 3 input was switched to the new processing farm (nmqxrt-27).

- The hybrid scan blockage input was switched to the new processing farm (nmqxrt-26).

03/12/2008 1435, 1525, 1615, 1710 - UTC - System

- Updated 3D Mosaic input configuration.  Setup configured on 3/7/2008 was less efficient than this new one.  No longer using makeNetcdfIndex to build code_index.lb files.  Instead, we're using a LDM script to create/write the LB notification.  Update times were....  1435 UTC for tile 1 & 2, 1525 UTC for tile 3 & 4, 1615 UTC for tile 5 & 6, and 1710 UTC for tile 7 & 8.

03/11/2008 2250 UTC - Polar VPR

- Changed VPR the minimum number of observations needed for a valid VPR value in a layer from 10 to 200.

03/11/2008 2100, 2110, 2115  UTC - System

- 2100 UTC: Switched US-only RRCT input from old polar processing nodes to new processing farm (see 03/07/2008 1430 UTC).

- 2115 UTC: Switched unQC'd CREF input from old polar processing nodes to new processing farm.  Also updated WDSS-II build to use newest w2merger lookup tables and radar locations.

03/07/2008 2340 UTC - 3D Mosaic

- Changed tropical dBZ threshold from 30 to 35 dBZ

 - Changed tropical influence radius from 150 to 120 km

- Question: Do we allow tropical flags to overwrite flags that indicate brightband contamination?  Answer: No, good convective and good stratiform flags are overwritten.

03/07/2008 1430-1630 UTC - System

- Switched 3D Mosaic input (single radar Cartesian grids, SRCs) source from old polar processing nodes to new processing farm.  The Farm is using the latest and greatest WDSS-II build (w2merger, w2qcnn), has updated radar location information, and uses new (improved and debugged) lookup tables for remapping from polar to Cartesian coordinates.  The updated radar locations consists mostly of changes related to elevation but also included a few changes in lat/lon (~0.001 deg difference). 

- Changed Q2 Radar-only timing.  Changes to 3D Mosaic input forces the delay of Q2 Radar-only data by 5 minutes.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~