source: trunk/LMDZ.MARS/libf/phymars/updaterad.F90 @ 1307

module updaterad



! This module intents to group together all ice and dust radius computations done in the GCM,
! so that it stays coherent through the code and make (numerous) radius bugs easier to debug.
! All different thresholds values are defined below.

! Radius computation do not always occur on the whole grid (cf. improvedcloud).
! So, subroutines are designed for scalar values instead of tables


! T. Navarro, June 2012



! For instance, if R^3 is lower than r3icemin, then R is set to ricemin.
! So, ricemin should be the cubic root of r3icemin, but not necessarily ...

real, parameter :: r3icemin = 1.e-30   ! ie ricemin  = 0.0001 microns
real, parameter :: ricemin  = 1.e-10

real, parameter :: r3icemax = 125.e-12 ! ie ricemax  = 500 microns
real, parameter :: ricemax  = 500.e-6 


real, parameter :: nccn_threshold  =  1.
real, parameter :: qccn_threshold  =  1.e-20

real, parameter :: r3ccnmin = 1.e-21    ! ie rccnmin = 0.1 microns
real, parameter :: rccnmin  = 0.1e-6

real, parameter :: r3ccnmax = 125.e-12  ! ie rccnmax  = 500 microns
real, parameter :: rccnmax  = 500.e-6  



real, parameter :: ndust_threshold  = 1.
real, parameter :: qdust_threshold  = 1.e-20

real, parameter :: r3dustmin = 1.e-24  ! ie rdustmin = 0.1 microns
real, parameter :: rdustmin  = 0.01e-6

real, parameter :: r3dustmax = 125.e-12 ! ie rdustmax  = 500 microns
real, parameter :: rdustmax  = 500.e-6  


real, parameter :: rdust0 = 0.8e-6


      

contains


!============================================================================
!============================================================================
!============================================================================
! Update ice radius if microphys == true 
subroutine updaterice_micro(qice,qccn,nccn,coeff,rice,rhocloud)
use tracer_mod, only: rho_dust, rho_ice
USE comcstfi_h
implicit none

real, intent(in)  :: qice,qccn,nccn
real, intent(in)  :: coeff         ! this coeff is tauscaling if microphy = T (possibly ccn_factor^-1 otherwise)
real, intent(out) :: rice,rhocloud ! rhocloud is needed for sedimentation and is also a good diagnostic variable
real nccn_true,qccn_true ! radius of the ice crystal core to the power of 3

    
nccn_true = max(nccn * coeff, 1.e-30)
qccn_true = max(qccn * coeff, 1.e-30)


!! Nota: It is very dangerous to apply a threshold on qccn or nccn to force rice to be ricemin.
!! Indeed, one can obtain ricemin for small but non negligible qice values, and therefore hugely opaque clouds.
 

rhocloud = (qice*rho_ice + qccn_true*rho_dust) / (qice + qccn_true)
rhocloud = min(max(rhocloud,rho_ice),rho_dust)

rice = (qice + qccn_true) * 0.75 / pi / rhocloud / nccn_true
  
if (rice .le. r3icemin) then
  rice = ricemin
else if (rice .ge. r3icemax) then
  rice = ricemax
else
  rice = rice**(1./3.) ! here rice is always positive
endif


end subroutine updaterice_micro
!============================================================================
!============================================================================
!============================================================================





!============================================================================
!============================================================================
!============================================================================
! Update ice radius from a typical profile if microphys == false
subroutine updaterice_typ(qice,tau,pzlay,rice)
use tracer_mod, only: rho_ice
USE comcstfi_h
implicit none

real, intent(in)  :: qice
real, intent(in)  :: tau   ! tau for dust
real, intent(in)  :: pzlay ! altitude at the middle of the layers
real, intent(out) :: rice
real rccn,nccn ! radius  and number of ice crystals

!   Typical CCN profile following Montmessin et al. 2004
!   (N0=2e6 m-3 has been converted into N0=1.3e8 kg-1, otherwise the equation for rice is not homogeneous...)
     nccn  = 1.3e+8*max(tau,0.001)/0.1*exp(-pzlay/10000.)
!   The previously used profile was not correct:
!   Nccn=( epaisseur/masse ) * 2.e+6/0.1*max(tau,0.001)*exp(-pzlay/10000.)
     
if (nccn .le. 1) then

  rice = ricemin

else

! Typical dust radius profile:
  rccn = max(rdust0*exp(-pzlay/18000.),1.e-9)
  rice  = qice * 0.75 / pi / rho_ice / nccn + rccn*rccn*rccn
       
  if (rice .le. r3icemin) then
    rice = ricemin
  else if (rice .ge. r3icemax) then
    rice = ricemax
  else
    rice = rice**(1./3.) ! here rice is always positive
  endif

endif
 
end subroutine updaterice_typ
!============================================================================
!============================================================================
!============================================================================






!============================================================================
!============================================================================
!============================================================================
! This subroutine computes the geometric mean radius(or number median radius)
! For a lognormal distribution :
! geometric mean radius = mass mean radius x exp(-1.5 sigma0^2)
! To be used with doubleq == true. otherwise, rdust is constant !!!
subroutine updaterdust(qdust,ndust,rdust,tauscaling)
use tracer_mod, only: r3n_q
USE comcstfi_h
implicit none

real, intent(in) :: qdust,ndust ! needed if doubleq
real, intent(in), optional :: tauscaling ! useful for realistic thresholds
real, intent(out) :: rdust
real coeff

if (present(tauscaling)) then
  coeff = tauscaling ! thresholds on realistic values
else
  coeff = 1. ! thresholds on virtual values
endif

if ((ndust .le. ndust_threshold/coeff) .or. (qdust .le. qdust_threshold/coeff)) then

   rdust = rdustmin

else

   rdust  = r3n_q * qdust / ndust

   if (rdust .le. r3dustmin) then
       rdust = rdustmin
   else if (rdust .ge. r3dustmax) then
       rdust = rdustmax
   else
      rdust = rdust**(1./3.)
   endif

endif
     

end subroutine updaterdust
!============================================================================
!============================================================================
!============================================================================






!============================================================================
!============================================================================
!============================================================================
! This subroutine computes the mass mean radius, 
! used for heterogenous nucleation on CCNs in microphysics.
! For a lognormal distribution :
! geometric mean radius = mass mean radius x exp(-1.5 sigma0^2)
subroutine updaterccn(qccn,nccn,rccn,tauscaling)
use tracer_mod, only: rho_dust
USE comcstfi_h
implicit none

real, intent(in) :: qccn,nccn ! needed if doubleq
real, intent(in), optional :: tauscaling ! useful for realistic thresholds

real, intent(out) :: rccn

real coeff


if (present(tauscaling)) then
  coeff = tauscaling ! threshold on realistic values
else
  coeff = 1. ! threshold on virtual values
endif

if ((nccn .le. nccn_threshold/coeff) .or. (qccn .le. qccn_threshold/coeff)) then

   rccn = rccnmin

else

   rccn  =  qccn * 0.75 / pi / rho_dust / nccn

   if (rccn .le. r3ccnmin) then
       rccn = rccnmin
   else if (rccn .ge. r3ccnmax) then
       rccn = rccnmax
   else
       rccn = rccn**(1./3.)
   endif

endif
     
end subroutine updaterccn
!============================================================================
!============================================================================
!============================================================================



end module updaterad