1. Turn on the surface obs in file nam_regional_convinfo.txt_cld
2. Turn off type 250 which was on by mistake when updating satwnd subtypes.
3. Put back gross error for surface winds to 5 m/s as in current operational NDAS; 6 m/s is used in the global.

1. 192 - sfc reports from synop land missing station and msl pressure
2. 193 - sfc reports from metar missing station and msl pressure
3. 194 - sfc reports from ship/cman/buoy missing station and msl pressure
4. 195 - sfc reports from mesonet missing station and msl pressure

1. The original jan19 version can result in NSImax=0 if there is no ice present, and that in turn could incorrectly lead to no ice present with NInuclei=0.
2. Fix made that set NSImax to be no less than NSI_max=250.e3 (250 per liter), so that NInuclei would be set to that value. This should lead to cloud ice now appearing at very cold temperatures. Note that NSImax can be >>250.e3 when there is a lot of ice present (like at the tops of convection); the value for NSI_max(=250.e3) just establishes a lower limit.

1. Turn on the diurnal reject lists and wbinuselist from RTMA
2. Turn on cloud analysis and diabatic digital filter initialization with radar-derived temperature tendencies during the NDAS tm09, tm06, and tm03 analysis/forecast

1. Turn on separate microphysics species advection
2. The maximum number concentration of ice (NLImax) is a function of the rime factor (RF) and temperature.
   1. For conditions when ice densities are high (RF>=10), NLImax=1.e3, which allows for mean ice diameters to exceed 1 mm in size. This so-called "hail mode", which allows the mean ice diameters to exceed 1 mm in size, only occurs in conditions associated with intense convection.
   2. For conditions when ice densities are not high (RF<10), NLImax is 10 L-1 at 0C and decreases to 5 L-1 at <=-40C. When the mean diameters of ice particles reach 1 mm in size, then the criterion is relaxed and the number concentrations of ice (NLICE) can exceed NLImax. This change helps with promoting the development of stratiform precipitation.
3. Never adjust RF values when the number concentrations of ice (NLICE) exceed the maximum value (NLImax). The previous code would artificially increase the rime factor (ice densities) so that NLICE=NLImax, but this is no longer justified when the mass-weighted rime factor is being advected. This change also helps with promoting the development of stratiform precipitation.
4. Ice deposition does not change rime factor values (RF) of ice when RF>10 at temperatures >-40C. This change preserves helps with preserving the presence of high-density ice particles in intense convective cores.
5. Limit the increase in ice fall speeds due to reduced air resistance at low pressures (high altitudes) to not exceed a factor of 1.5 times the fall speeds at a reference pressure of 1000 hPa. This change prevents ice particles from falling too quickly in the upper troposphere.
6. At cold temperatures (<-30C), the maximum number concentrations of small ice crystals (NSImax) is calculated based on assuming 10% of the total ice content is due to cloud ice. This change allows for higher amounts of cloud ice to exist at the tops of intense onvective cores, which can then gradually be converted into snow as parcels move away from areas of intense convection.

defc=0.005*(3-l+1.)/3.

to

defc=0.015*(3-l+1.)/3.

where l=the model level. This change eliminated the instability in the top model layers that caused the failures in the 00z 1/3 and 00z 1/4 NAMX runs in the Alaska nest.

1. NOAA-19 AMSU-A Ch 7
2. NOAA-19 MHS Ch 3

1. Minimum effective radius for cloud droplets varies from 10 microns at 20C to 15 microns at 0C
2. Number concentration of cloud droplets varies from 200 cm^-3 at 20C to 50 cm^-3 at 0C.
3. Treat all ice as cloud ice in RRTM's progcld2. The effective radius for cloud ice is parameterized following Wyser (1998 J. Climate), in which the smaller cloud ice that is assumed should lead to slightly warmer (improved) temperatures aloft
4. Build in tiny amounts of cloud within RRTM driver when no cloud is present and RH>95%.
5. Used the "sep17" version of the modified fer_hires microphysics with RHgrd=98%.
6. Changes to the JSFC surface layer, increasing ZTmax1 & ZTmax2 from 1.0 to 9.0.
7. Added the snow cover array (SNOWC, the fraction of the grid covered by snow) that is calculated in the Noah LSM. The maximum fraction is capped at 0.98 (98% coverage) rather than 1.00 (100%) as in the repository code. The SNOWC array is provided as input to the RRTM radiation for the snow cover, where it is calculated following Marshall et al. (1994; =SNOW/[SNOW+70] for SNOW depth in mm).
8. The fraction of frozen precipitation is partitioned into liquid and frozen contributions for accumulation in the land surface model. The liquid portion goes into the soil, the frozen portion contributes to the snow pack. Rime factor is passed into the Noah LSM and used to increase the density of the incoming frozen precipitation in order to reduce the depth of snow accumulations in marginal winter conditions with complex precipitation types. Also, the upper limit of snow density is increased from 0.4 g/cm^3 to 0.9 g/cm^3.

1. Turn on new satellite bias correction.
2. Upgrade GSI to include GFS spectral IO modification which will improve wall time and reduce memory requirement; needed for global ensemble with higher resolution.
3. Monitor GOES hourly visible winds

1. Increase long wave cloud emissivities by 1.5x
2. Increase minimum cloud droplet radius from 5 to to 15 microns

1. Redo the use of radiosonde significant level data
2. Changes to use CRTM2.1.3
3. Assorted bug fixes
4. Add endian-independent open mechanism

TEM = MAX(VELCO(I,K)*VELCO(I,K), .0001)

to

TEM = MAX(VELCO(I,K)*VELCO(I,K), .1)

where

SCR = BNV2(I,K) / TEM

where BNV2 is the Brunt-Vaisala frequency. This change was made to make this expression consistent with the GFS version.

FR = BNV * ULOI(I) * min(HPRIME(J),hpmax)

where hprime is the standard deviation of the terrain height, BNV is Brunt-Vaisala frequency, and ULOI is 1/Umean, the mean low-level flow.

1. Evergreen Needleleaf Forest
2. Deciduous Broadleaf Forest
3. Mixed Forests
4. Croplands
5. Cropland/natural vegetation mosaic

1. Decreased NLImax (the maximum number of large ice particles) from 20.e3 (20 per liter) to 10.e3 (20 per liter).

1. An error in the LCL layer in the triggering part of the code was fixed
2. CAPE is checked instead of entropy in the triggering part of the code
3. Redundant constant definitions through argument list and use are removed
4. Weak momentum transport by shallow convection
5. Combination of old and new shallow convection

1. Turn on variational QC scheme in GSI code
2. Assimilate Meteosat 10 wind subtypes with different data thinning

1. Replace refractivity with bending angle from satellites GRACE A, COSMIC 1-6, METOP A
2. Turn on bending angle obs from TSX, SAC-C, C/NOFS
3. Set cut off height to 30 km (12mb)

1. Use global monthly CO2 concentrations that are used in the GFS
2. Removed diurnal variation of albedo
3. Code changed to add in small amounts of convective clouds into the input cloud calculations, but only if all of the following conditions are met:
   1. The maximum condensate mixing ratio is < QWmax.
   2. Only shallow convection is present, do not apply to deep convection.
   3. Only apply if the depth of shallow convection is between CU_DEEP_MIN (50 hPa) and CU_DEEP_MAX (200 hPa).
   4. Convective precipitation rate must be <0.01 mm/h.

1. When GFS spectral files are used (as is the case in starting the NDAS), the specific humidity rather than the RH field now is used as the moisture source.
2. The RH processing is modified to prevent setting a floor value of 1% in the RH --> mixing ratio conversion routine. Now a floor value of 0.001% is utilized.

1. Use the dynamic albedos from the land surface model
2. Smoothing of vertical profiles of heating/cooling rates
3. Uses cloud fractions implemented in October 2011 in the NAM (see here for a description of the change)
4. Correctly separate cloud ice and snow mixing ratios from the input total ice mixing ratios
5. Calculates cloud-layer (e.g., low/middle/high) and broadband radiation from the prognostic cloud code rather than from the old NAM/GFDL driver (19 Mar version of radiation_clouds.f)

1. Decreased NLImax (the maximum number of large ice particles) from 50.e3 (50 per liter) to 20.e3 (20 per liter).
2. Fixed a bug in the algorithm that calculates the number of large ice particles and the mean ice-particle diameter. It will also alter the rime factor (ice particle density) under special, rare circumstances when the mean diameter and number concentration of the large ice are set to maximum values.

1. Run with experimental BMJ scheme ("bmj_dev")
2. Reduce resolution factor for DSP's ("fres') from 1.0 to 0.5