HRRR Frequently Asked Questions
Q: How is the HRRR initialized?
A: The HRRR is initialized from the RAP one hour prior to the forecast start time. Then the model is advanced for one hour, assimilating radar reflectivity observations every 15mins. This is referred to as a one hour pre-forecast. Following the pre-forecast step, final data assimilation and hydrometeor analysis occurs and a free forecast is started.
Q: When are the radiosonde observations assimilated into the HRRR?
A: Most of the 00/12 UTC radiosonde observations are assimilated into the 00/12 UTC HRRR. The 00/12 UTC HRRR uses an "early" data dump of the available observations, which includes approximately 2/3 of the of the radiosonde data (mostly lower-to-mid levels). The 01/13 UTC HRRR runs benefit from the complete radiosonde observations via the initial conditions from the 00/12 UTC RAP.
Q: What are the native (sigma) model levels?
A: eta_levels = 1.0000, 0.9980, 0.9940, 0.9870, 0.9750, 0.9590, 0.9390, 0.9160, 0.8920, 0.8650, 0.8350, 0.8020, 0.7660, 0.7270, 0.6850, 0.6400, 0.5920, 0.5420, 0.4970, 0.4565, 0.4205, 0.3877, 0.3582, 0.3317, 0.3078, 0.2863, 0.2670, 0.2496, 0.2329, 0.2188, 0.2047, 0.1906, 0.1765, 0.1624, 0.1483, 0.1342, 0.1201, 0.1060, 0.0919, 0.0778, 0.0657, 0.0568, 0.0486, 0.0409, 0.0337, 0.0271, 0.0209, 0.0151, 0.0097, 0.0047, 0.0000,
Q: Why do "circles" appear in the HRRR ceiling and cloud base height analysis (0 hour forecast)?
A: Circular shapes in the ceiling or cloud base height analysis are caused by assimilation of observations from METAR ceilometers.
We are adding clouds within a 60-km radius of a METAR site (somewhat arbitrary and admittedly a bit crude) in the absence of adjacent METAR sites within 120-km, if the HRRR model does not already contain those clouds. We have plans to improve this approach in the very near future (first in our experimental HRRR in November 2014 and then in an update to the operational HRRR next year) where we will rely upon satellite cloud top observations as the final determination (as opposed to METARs) whether clouds should be added where the model does not have them. This enhancement will help to better define the presence and horizontal extent of these clouds and reduce these circular patterns.
Q: Do you have FORTRAN code for converting the u and v wind components aligned to the HRRR grid to components with respect to true north (meteorological standard)?
Here is a code snippet that should do the job.
PARAMETER ( ROTCON_P = 0.622515 )
PARAMETER ( LON_XX_P = -97.5 )
PARAMETER ( LAT_TAN_P = 38.5 )
C** ROTCON_P R WIND ROTATION CONSTANT, = 1 FOR POLAR STEREO
C** AND SIN(LAT_TAN_P) FOR LAMBERT CONFORMAL
C** LON_XX_P R MERIDIAN ALIGNED WITH CARTESIAN X-AXIS(DEG)
C** LAT_TAN_P R LATITUDE AT LAMBERT CONFORMAL PROJECTION
C** IS TRUE (DEG)
angle2 = rotcon_p*(olon(i,j)-lon_xx_p)*0.017453
sinx2 = sin(angle2)
cosx2 = cos(angle2)
ut = u(i,j,k)
vt = v(i,j,k)
un(i,j,k) = cosx2*ut+sinx2*vt