Skip to content

Commit

Permalink
HAILCAST bug fixes: tiling, non-SI units (wrf-model#1877)
Browse files Browse the repository at this point in the history 
TYPE: bug fix

KEYWORDS: HAILCAST, tiling issues, undefined variables

SOURCE: Becky Adams-Selin

DESCRIPTION OF CHANGES:
Problem:
When running using tiling/quilted processes, the haildtacttime variable was being reset by every tile, thereby only activating HAILCAST to run on a single tile. Also, the RWA_new calculation can sometimes be undefined if the cloud base index is not set before the loop, and some SI/non-SI unit conversions were not handled correctly in the HEATBUD subroutine.

Solution:
The haildtacttime variable was changed to an gridded variable so could be set individually by gridpoint. Cloud base index definition added before the loop. SI/non-SI unit conversions fixed.

LIST OF MODIFIED FILES:
M Registry/Registry.EM_COMMON
M phys/module_diag_hailcast.F

TESTS CONDUCTED:
Modifications fixed problem, and are currently used in operational HRRR v4.0.
The Jenkins tests are all passing.

RELEASE NOTE: HAILCAST quilted processes/tiling and a few other bug fixes.
  • Loading branch information
@adams-selin @vlakshmanan-scala
adams-selin authored and vlakshmanan-scala committed Apr 4, 2024
1 parent 87b15aa commit 09db56e
Show file tree
Hide file tree
Showing 2 changed files with 113 additions and 53 deletions.
2 changes: 1 addition & 1 deletion Registry/Registry.EM_COMMON
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3275,7 +3275,7 @@ state real HAILCAST_DIAM_MEAN ij misc 1 - -
state real HAILCAST_DIAM_STD ij misc 1 - - "HAILCAST_DIAM_STD" "WRF-HAILCAST Stand. Dev. Hail Diameter" "mm"
state real HAILCAST_WUP_MASK ij misc 1 - r "HAILCAST_WUP_MASK" "Updraft mask, 1 if > 10m/s" ""
state real HAILCAST_WDUR ij misc 1 - r "HAILCAST_WDUR" "Updraft duration" "sec"
state real haildtacttime - - - - r "haildtacttime" "HAILDTACTTIME" "HAILCAST ACTIVATION TIME in s"
state real haildtacttime ij misc 1 - r "haildtacttime" "HAILDTACTTIME" "HAILCAST ACTIVATION TIME in s"


package hailcast hailcast_opt==1 - state:hailcast_diam_max,hailcast_diam_mean,hailcast_diam_std,hailcast_dhail1,hailcast_dhail2,hailcast_dhail3,hailcast_dhail4,hailcast_dhail5
Expand Down
164 changes: 112 additions & 52 deletions phys/module_diag_hailcast.F
View file
Original file line number Diff line number Diff line change
Expand Up @@ -50,7 +50,8 @@ SUBROUTINE hailcast_diagnostic_driver ( grid , config_flags &

REAL, INTENT(IN ),OPTIONAL :: haildt
REAL, INTENT(IN ),OPTIONAL :: curr_secs
REAL, INTENT(INOUT),OPTIONAL :: haildtacttime
REAL, DIMENSION( ims:ime, jms:jme ), &
INTENT(INOUT),OPTIONAL :: haildtacttime
INTEGER, INTENT(IN ) :: itimestep
REAL, INTENT(IN ) :: dt

Expand Down Expand Up @@ -209,10 +210,18 @@ SUBROUTINE hailcast_diagnostic_driver ( grid , config_flags &
( (.not. grid%nested) .or. &
( (grid%nested) .and. (abs(grid%dtbc) < 0.0001) ) ) )THEN
doing_adapt_dt = .TRUE.
IF ( haildtacttime .eq. 0. ) THEN
haildtacttime = CURR_SECS + haildt
END IF
END IF
!170519 - bug fix RAS
!Change haildtacttime to an array so can be set individually
! by gridpoint to account for quilted processes
DO j = j_start, j_end
DO i = i_start, i_end
IF ( haildtacttime(i,j) .eq. 0. ) THEN
haildtacttime(i,j) = CURR_SECS + haildt
END IF
ENDDO
ENDDO
END IF
! Calculate STEPHAIL
stephail = NINT(haildt / dt)
Expand Down Expand Up @@ -270,12 +279,19 @@ SUBROUTINE hailcast_diagnostic_driver ( grid , config_flags &
decided = .TRUE.
END IF
IF ( ( .NOT. decided ) .AND. &
( doing_adapt_dt ) .AND. &
( curr_secs .GE. haildtacttime ) ) THEN
run_param = .TRUE.
decided = .TRUE.
haildtacttime = curr_secs + haildt
!170519 - bug fix RAS
!Changed haildtacttime to an array so can be set individually
! by gridpoint to account for quilted processes
IF ( ( .NOT. decided ) .AND. &
( doing_adapt_dt ) .AND. &
( curr_secs .GE. haildtacttime(i_start, j_start) ) ) THEN
run_param = .TRUE.
decided = .TRUE.
DO j = j_start, j_end
DO i = i_start, i_end
haildtacttime(i,j) = curr_secs + haildt
ENDDO
ENDDO
END IF
write ( message, * ) 'timestep to run HAILCAST? ', run_param
Expand Down Expand Up @@ -442,6 +458,13 @@ END SUBROUTINE hailcast_diagnostic_driver
!!!! 06 May 2017...................................Becky Adams-Selin AER
!!!! Bug fixes for some memory issues in the adiabatic liquid
!!!! water profile code.
!!!! 23 Apr 2019...................................Becky Adams-Selin AER
!!!! Added check to make sure embryo spends at least some time in
!!!! cloud before returning a positive hail diameter.
!!!! 16 Dec 2022...................................Becky Adams-Selin AER
!!!! Consolodate a number of bug fixes, including fixing behavior
!!!! of haildt when tiling is used, and non-SI units errors in
!!!! HEATBUD subroutine.
!!!!
!!!! See Adams-Selin and Ziegler 2016, MWR for further documentation.
!!!!
Expand Down Expand Up @@ -609,7 +632,8 @@ SUBROUTINE hailstone_driver ( TCA, h1d, ht, PA, rho1d,&
!the difference between the saturation mixing ratio at each level
!and the cloud base water vapor mixing ratio
DO k=1,nz
IF (k.LT.KBAS) THEN
RWA_new(k) = RWA(k)
IF (k.LT.KBAS) THEN
RWA_adiabat(k) = 0.
CYCLE
ENDIF
Expand Down Expand Up @@ -640,19 +664,10 @@ SUBROUTINE hailstone_driver ( TCA, h1d, ht, PA, rho1d,&
ELSE IF ((k.ge.KBAS+1).AND.(RWA_adiabat(k).ge.1.E-12)) THEN
RWA_new(k) = RWA_adiabat(k)*(h1d(k)-h1d(k-1)) - RWA_new(k-1)
IF (RWA_new(k).LT.0) RWA_new(k) = 0.
ELSE
RWA_new(k) = RWA(k)
ENDIF
! - old code - !
!IF (k.eq.KBAS+1) THEN
! RWA_new(k) = RWA_adiabat(k)*(h1d(k)-h1d(k-1))
!ELSE IF ((k.ge.KBAS+2).AND.(RWA_adiabat(k).ge.1.E-12)) THEN
! RWA_new(k) = RWA_adiabat(k)*(h1d(k)-h1d(k-1)) - RWA_new(k-1)
! IF (RWA_new(k).LT.0) RWA_new(k) = 0.
!ENDIF
ENDDO
!remove the height factor from RWA_new
DO k=KBAS,nz
DO k=KBAS+1,nz !bug fix, ensure index start 1 higher
RWA_new(k) = RWA_new(k) / (h1d(k)-h1d(k-1))
ENDDO
Expand Down Expand Up @@ -696,6 +711,11 @@ SUBROUTINE hailstone_driver ( TCA, h1d, ht, PA, rho1d,&
CALL INTERP(VUU_pert, VUFZL, WFZLP, IFOUT, PA, nz)
CALL INTERP(rho1d, DENSAFZL, WFZLP, IFOUT, PA, nz)
!Check if height of cloud base is above embryo insertion point
!RAS-20190423-!
IF (ZFZL < ZBAS-1000) GOTO 400
!Begin hail simulation time (seconds)
sec = 0.
Expand Down Expand Up @@ -850,9 +870,15 @@ SUBROUTINE hailstone_driver ( TCA, h1d, ht, PA, rho1d,&
!so let's shed any excess water now.
D_ICE = ( (6*GM*(1.-FW)) / (3.141592654*DENSE) )**0.33333333
D = D_ICE
CALL MELT(D,TLAYER,PLAYER,RLAYER,LDEPTH,VT)
CALL MELT(D,TLAYER,PLAYER,RLAYER,LDEPTH,VT,TS,DENSE)

ENDIF !end check for melting call

!RAS - 20190423 - !
400 CONTINUE !Check for if embryo insertion point is below cloud base

!Require embryo to have stayed aloft for at least some time
IF (sec.LT.60) D = 0.

!Check to make sure something hasn't gone horribly wrong
IF (D.GT.0.254) D = 0. !just consider missing for now if > 10 in
Expand Down Expand Up @@ -993,7 +1019,7 @@ SUBROUTINE TERMINL(DENSA,DENSE,D,VT,TC)
RE=(GX/0.6)**0.5
!SELECT APPROPRIATE EQUATIONS FOR TERMINAL VELOCITY DEPENDING ON
!THE BEST NUMBER
!THE BEST NUMBER. Follows RH87 pg. 2762, but fixes an error in their (B1).
IF (GX.LT.550) THEN
W=LOG10(GX)
Y= -1.7095 + 1.33438*W - 0.11591*(W**2.0)
Expand Down Expand Up @@ -1115,17 +1141,18 @@ SUBROUTINE MASSAGR(D,GM,GM1,GMW,GMI,DGM,DGMW,DGMI,DGMV,DI,ANU,RE,AE,&
!experimentally derived, and are expecting cal-C-g units. Do some conversions.
DENSAC = DENSA * (1.E3) * (1.E-6)
!dynamic viscosity
ANU=1.717E-4*(393.0/(TC+120.0))*(TC/273.155)**1.5
ANU=1.717E-5*(393.0/(TC+120.0))*(TC/273.155)**1.5 !RAS units fix to agree with Roger and Yau ps. 102
!!! CALCULATE THE REYNOLDS NUMBER - unitless
RE=D*VT*DENSAC/ANU
E=(0.60)**(0.333333333)*(RE**0.50) !ventilation coefficient vapor (fv)
!!! SELECT APPROPRIATE VALUES OF AE ACCORDING TO RE
IF(RE.LT.6000.0)THEN
AE=0.78+0.308*E
ELSEIF(RE.GE.6000.0.AND.RE.LT.20000.0)THEN
AE=0.76*E
!RAS 190713
AE=0.5*0.76*E !bug fix to match RasmussenHeymsfield1987
ELSEIF(RE.GE.20000.0) THEN
AE=(0.57+9.0E-6*RE)*E
AE = 0.5*(0.57+9.0E-6*RE)*E !bug fix to match RasmussenHeymsfield1987
ENDIF
Expand Down Expand Up @@ -1159,6 +1186,9 @@ SUBROUTINE MASSAGR(D,GM,GM1,GMW,GMI,DGM,DGMW,DGMI,DGMV,DI,ANU,RE,AE,&

!!! CALCULATE THE TOTAL MASS CHANGE
DGM=DGMW+DGMI+DGMV
!Dummy arguments in the event of no water, vapor, or ice growth
DENSELW = 900.
DENSELI = 700.
!!! CALCULATE DENSITY OF NEW LAYER, DEPENDS ON FW AND ITYPE
IF (ITYPE.EQ.1) THEN !DRY GROWTH
!If hailstone encountered supercooled water, calculate new layer density
Expand All @@ -1169,7 +1199,12 @@ SUBROUTINE MASSAGR(D,GM,GM1,GMW,GMI,DGM,DGMW,DGMI,DGMV,DI,ANU,RE,AE,&
!!! FIND THE STOKES NUMBER (rasmussen heymsfield 1985)
NS = 2*VT*100.*(DC*1.E-4)**2. / (9*ANU*D*50) !need hail radius in cm
!!! FIND IMPACT VELOCITY (rasmussen heymsfield 1985)
W = LOG10(NS)
!W = LOG10(NS) - bug fix 20221216 - RAS !
IF (NS.LT.0.1)THEN
W=-1.
ELSE
W = LOG10(NS)
ENDIF
IF (RE.GT.200) THEN
IF (NS.LT.0.1) THEN
VIMP = 0.0
Expand Down Expand Up @@ -1213,7 +1248,7 @@ SUBROUTINE MASSAGR(D,GM,GM1,GMW,GMI,DGM,DGMW,DGMI,DGMV,DI,ANU,RE,AE,&
AFACTOR = -DC*VIMP/TSCELSIUS
IF ((TSCELSIUS.LE.-5.).OR.(AFACTOR.GE.-1.60)) THEN
DENSELW = 0.30*(AFACTOR)**0.44
ELSEIF (TSCELSIUS.GT.-5.) THEN
ELSE
DENSELW = EXP(-0.03115 - 1.7030*AFACTOR + &
0.9116*AFACTOR**2. - 0.1224*AFACTOR**3.)
ENDIF
Expand Down Expand Up @@ -1309,26 +1344,33 @@ SUBROUTINE HEATBUD(TS,TSm1,TSm2,FW,TC,VT,DELRW,D,DENSA,GM1,GM,DGM,DGMW, &
REAL RV, RD, G, PI, ALF, ALV, ALS, CI, CW, AK
REAL H, AH, TCC, TSC, DELRWC, DENSAC, TDIFF
REAL DCM, GMC, DGMWC, DGMIC, GM1C, DGMC
REAL DMLT
REAL TSCm1, TSCm2
DATA RV/461.48/,RD/287.04/,G/9.78956/
DATA PI/3.141592654/,ALF/79.7/,ALV/597.3/
DATA ALS/677.0/,CI/0.5/,CW/1./
!Convert values to non-SI units here
!Convert values to non-SI units here to match all the constants
TSC = TS - 273.155
TSCm1 = TSm1 - 273.155
TSCm2 = TSm2 - 273.155
TCC = TC - 273.155
DELRWC = DELRW * (1.E3) * (1.E-6)
DENSAC = DENSA * (1.E3) * (1.E-6)
!DI still in cm2/sec
DCM = D * 100. !m to cm
GMC = GM * 1000. !kg to g
GM1C = GM1 * 1000. !kg to g
DGMWC = DGMW * 1000. !kg to g
DGMIC = DGMI * 1000. !kg to g
DGMC = DGM * 1000. !kg to g
!!! CALCULATE THE CONSTANTS
AK=(5.8+0.0184*TCC)*1.E-5 !thermal conductivity - cal/(cm*sec*K)
!dynamic viscosity - calculated in MASSAGR
!ANU=1.717E-4*(393.0/(TC+120.0))*(TC/273.155)**1.5
!ANU=1.717E-5*(393.0/(TC+120.0))*(TC/273.155)**1.5
!!! CALCULATE THE REYNOLDS NUMBER - unitless
!RE=D*VT*DENSAC/ANU - calculated in MASSAGR
Expand All @@ -1341,10 +1383,11 @@ SUBROUTINE HEATBUD(TS,TSm1,TSm2,FW,TC,VT,DELRW,D,DENSA,GM1,GM,DGM,DGMW, &
AH=0.78+0.308*H
!AE=0.78+0.308*E
ELSEIF(RE.GE.6000.0.AND.RE.LT.20000.0)THEN
AH=0.76*H
!AE=0.76*E
!RAS 190713
AH=0.5*0.76*H !bug fix to match RasmussenHeymsfield1987
ELSEIF(RE.GE.20000.0) THEN
AH=(0.57+9.0E-6*RE)*H
!RAS 190713
AH=0.5*(0.57+9.0E-6*RE)*H !bug fix to match RasmussenHeymsfield1987
!AE=(0.57+9.0E-6*RE)*E calculated in MASSAGR
ENDIF
Expand All @@ -1358,11 +1401,12 @@ SUBROUTINE HEATBUD(TS,TSm1,TSm2,FW,TC,VT,DELRW,D,DENSA,GM1,GM,DGM,DGMW, &
!TSC=TSC-TSC*DGM/GM1+SEKDEL/(GM1*CI)* &
! (2.*PI*D*(AH*AK*(TCC-TSC)-AE*ALS*DI*DELRWC)+ &
! DGMW/SEKDEL*(ALF+CW*TCC)+DGMI/SEKDEL*CI*TCC)
TSC=0.6*(TSC-TSC*DGM/GM1+SEKDEL/(GM1*CI)* &
(2.*PI*D*(AH*AK*(TCC-TSC)-AE*ALS*DI*DELRWC)+ &
DGMW/SEKDEL*(ALF+CW*TCC)+DGMI/SEKDEL*CI*TCC)) + &
!units fix
TSC=0.6*(TSC-TSC*DGMC/GM1C+SEKDEL/(GM1C*CI)* &
(2.*PI*DCM*(AH*AK*(TCC-TSC)-AE*ALS*DI*DELRWC)+ &
DGMWC/SEKDEL*(ALF+CW*TCC)+DGMIC/SEKDEL*CI*TCC)) + &
0.2*TSCm1 + 0.2*TSCm2
TS = TSC+273.155
IF (TS.GE.273.155) THEN
TS=273.155
Expand All @@ -1375,14 +1419,21 @@ SUBROUTINE HEATBUD(TS,TSm1,TSm2,FW,TC,VT,DELRW,D,DENSA,GM1,GM,DGM,DGMW, &
IF (TCC.LT.0.) THEN
!Original Hailcast algorithm
FW=FW-FW*DGM/GM1+SEKDEL/(GM1*ALF)* &
(2.*PI*D*(AH*AK*TCC-AE*ALV*DI*DELRWC)+ &
DGMW/SEKDEL*(ALF+CW*TCC)+DGMI/SEKDEL*CI*TCC)
!FW=FW-FW*DGM/GM1+SEKDEL/(GM1*ALF)* &
! (2.*PI*D*(AH*AK*TCC-AE*ALV*DI*DELRWC)+ &
! DGMW/SEKDEL*(ALF+CW*TCC)+DGMI/SEKDEL*CI*TCC)
!units fixed
FW = FW-FW*DGMC/GMC+SEKDEL/(GMC*ALF)* &
(2.*PI*DCM*(AH*AK*TCC-AE*ALV*DI*DELRWC)+ &
DGMWC/SEKDEL*(ALF+CW*TCC)+DGMIC/SEKDEL*CI*TCC)
ELSE
!Calculate decrease in ice mass due to melting
DMLT = (2.*PI*D*AH*AK*TCC + 2.*PI*D*AE*ALV*DI*DELRWC + &
DGMW/SEKDEL*CW*TCC) / ALF
FW = (FW*GM + DMLT) / GM
!!Bug fix 28Jul2021 RAS - non-SI units
!More non-SI units - fixed 20220307 RAS
! Extra SEKDEL in the calculation - 20221102 RAS
DMLT = SEKDEL / ALF * (2.*PI*DCM*AH*AK*TCC + 2.*PI*DCM*AE*ALV*DI*DELRWC + &
DGMWC/SEKDEL*CW*TCC)
FW = (FW*GM1C + DMLT + DGMWC) / GMC
ENDIF
IF(FW.GT.1.)FW=1.
Expand Down Expand Up @@ -1436,13 +1487,19 @@ SUBROUTINE BREAKUP(DENSE,D,GM,FW,CRIT)
END SUBROUTINE BREAKUP


SUBROUTINE MELT(D,TLAYER,PLAYER,RLAYER,LDEPTH,VT)
SUBROUTINE MELT(D,TLAYER,PLAYER,RLAYER,LDEPTH,VT,TS,DENSE)
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!! This is a spherical hail melting estimate based on the Goyer
!!! et al. (1969) eqn (3). The depth of the warm layer, estimated
!!! terminal velocity, and mean temperature of the warm layer are
!!! used. DRB. 11/17/2003.
!!!
!!! Incorporated some known variables into the subroutine
!!! to use instead of constants (VT, TS, DENSE). RAS 10/2/2019
!!!
!!! Note - this subroutine assumes melted water is immediately
!!! shed. Possibly not the most accurate assumption.
!!!
!!! INPUT: TLAYER mean sub-cloud layer temperature (K)
!!! PLAYER mean sub-cloud layer pressure (Pa)
!!! RLAYER mean sub-cloud layer mixing ratio (kg/kg)
Expand All @@ -1453,7 +1510,7 @@ SUBROUTINE MELT(D,TLAYER,PLAYER,RLAYER,LDEPTH,VT)
IMPLICIT NONE

REAL*8 D
REAL TLAYER, PLAYER, RLAYER, LDEPTH, VT
REAL TLAYER, PLAYER, RLAYER, LDEPTH, VT, TS, DENSE
REAL eenv, delta, ewet, de, der, wetold, wetbulb, wetbulbk
REAL tdclayer, tclayer, eps, b, hplayer
REAL*8 a
Expand Down Expand Up @@ -1502,7 +1559,8 @@ SUBROUTINE MELT(D,TLAYER,PLAYER,RLAYER,LDEPTH,VT)
dv = 0.25e-4 ! diffusivity of water vapor (m2/s)
pi = 3.1415927
rv = 1004. - 287. ! gas constant for water vapor
rhoice = 917.0 ! density of ice (kg/m**3)
!rhoice = 917.0 ! density of ice (kg/m**3)
rhoice = DENSE ! we know the stone density, let's use it
r = D/2. ! radius of stone (m)
!Compute residence time in warm layer
Expand All @@ -1511,12 +1569,14 @@ SUBROUTINE MELT(D,TLAYER,PLAYER,RLAYER,LDEPTH,VT)
!Calculate dmdt based on eqn (3) of Goyer et al. (1969)
!Reynolds number...from pg 317 of Atmo Physics (Salby 1996)
!Just use the density of air at 850 mb...close enough.
rho = 85000./(287.*TLAYER)
re = rho*r*VT*.01/1.7e-5
!rho = 85000./(287.*TLAYER)
rho = player/(287.*TLAYER) !we have the layer pressure, just use it
!units fix - r is now in meters, not mm. Plus we need D, not r. RAS 20220308
re = rho*r*2*VT/1.7e-5
!Temperature difference between environment and hailstone surface
delt = wetbulb !- 0.0 !assume stone surface is at 0C
!wetbulb is in Celsius
!We know the stone surface temperature, let's use it
delt = wetbulb - (TS - 273.155) !again, wet bulb is in C

!Difference in vapor density of air stream and equil vapor
!density at the sfc of the hailstone
Expand All @@ -1531,7 +1591,7 @@ SUBROUTINE MELT(D,TLAYER,PLAYER,RLAYER,LDEPTH,VT)
!Calculate new mass growth
dmdt = (-1.7*pi*r*(re**0.5)/lf)*((ka*delt)+((lv-lf)*dv*dsig))
IF (dmdt.gt.0.) dmdt = 0
mass = dmdt*tres
mass = dmdt*tres ! < 0

!Find the new hailstone diameter
massorg = 1.33333333*pi*r*r*r*rhoice
Expand Down

0 comments on commit 09db56e

Please sign in to comment.