WRF hydrometeor lateral-boundary advection and nudging

[wang1202]

WRF hydrometeor lateral-boundary advection and nudging

User-facing configuration

Enable WRF real lateral boundaries and the WRF hydrometeor fields in the
erf input section:

erf.init_type            = "WRFInput"
erf.use_real_bcs         = true
erf.nc_init_file_0       = "wrfinput_d01"
erf.nc_bdy_file          = "wrfbdy_d01"
erf.use_wrf_bdy_qc_qi    = true
erf.bdy_moist_nudge_type = 3

erf.use_wrf_bdy_qc_qi = true enables WRF hydrometeor boundary data. It
requires WRFInput real boundaries and a wrfbdy file containing the relevant
fields. Set erf.bdy_moist_nudge_type = 3 to directly nudge the active moist
components toward the time-interpolated WRF boundary values. The usual
erf.real_width and erf.bdy_nudge_factor settings control the boundary-zone
width and nudging strength.

Hydrometeor behavior

ERF determines the behavior from the active erf.moisture_model state layout.
If the moisture scheme has separate qc, qi, qr, qs, and qg
components, all of those categories are advected and filled at the lateral
boundaries separately. With moisture nudging type 3, they are also nudged
separately using the corresponding WRF fields:

ERF component	WRF field
qc	QCLOUD
qi	QICE
qr	QRAIN
qs	QSNOW
qg	QGRAUP

This separate-component path applies to schemes that provide all five
hydrometeor components, such as Morrison, WSM6, SAM, WDM6, and SuperDroplets.

For schemes that do not provide all five components, ERF uses the aggregate
fallback:

• QCLOUD + QRAIN supplies qc.
• QICE + QSNOW + QGRAUP supplies qi when an ice component is active.

The precipitating categories that are not active in the selected moisture
scheme are not added as ERF state components.

Notes

• The WRF graupel variable is named QGRAUP.
• erf.bdy_moist_nudge_type = 3 requires
  erf.use_wrf_bdy_qc_qi = true.
• Regenerate any erfbdy cache after switching between aggregate and
  separate-component moisture layouts.

Validation

• CUDA production targets built successfully.
• The focused WRF boundary moisture unit-test source compiled successfully.

[asalmgren]

@wang1202 -- I'm wondering if the ice etc are available in wrfbdy and are used in the moisture model, should we force the nudging to always use nudging type 3? This wouldn't break any existing cases and would make the logic more straightforward...

[wang1202]

@wang1202 -- I'm wondering if the ice etc are available in wrfbdy and are used in the moisture model, should we force the nudging to always use nudging type 3? This wouldn't break any existing cases and would make the logic more straightforward...

I'm not sure whether all sources have those hydrometeors (ice, snow, rain...), but it's in the wrfbdy generated from HRRR. Maybe after more testings to see that the runs are stable we can make this the default case?

In my testings, I did see that the results are improved (although it seems not to be the reason of the missing patch of cloud above Nantucket.)