Model Change Bulletin

                                   MCB#3                February 12, 2013

                            SCREEN3 (dated 13043)


   This Model Change Bulletin documents changes made to the SCREEN3 model.
   
The original 16-bit SCREEN3 executable from 1996 and previous 32-bit versions 
did not work on Windows 64-bit Operating Systems.  SCREEN3 was modified  
slightly and recompiled using Intel(R) Visual Fortran Compiler XE 13.1.0.149 
[IA-32].  SCREEN3 now runs on Windows 7, x64 computer systems.

Newer compilers are less tolerant of deviations from older Fortran Programming
standards.  The above compiler identified several such errors and they had to 
be corrected before the executable could be built.  The following is a list of 
those corrections along with some minor changes.

1)  Several edits were made to the SCREEN3A.FOR source code in the header 
area to change the creation date from 96043 to 13043 and from MCB #2 to MCB #3

2)  The length of CINP had to be increased from 50 to 80.  CINP is passed to a
subroutine LERUPR where the receiving variable is of a length of 80 characters.
The original line:

      CHARACTER*50 CINP

was changed to:

      CHARACTER*80 CINP
      
3)  The capability to open a fixed filename, SCREEN.INP, was added but not 
implemented.  This would normally curtail the need to use redirection.  The 
following line was added to initialize the input filename variable, INPFIL:

      INPFIL = 'SCREEN.INP'

4)  If a fixed filename is required for use under another operating system, 
the following statement needs to be uncommented before recompiling.

C      OPEN(IRD,FILE=INPFIL,STATUS='OLD')

5)  There were too many commas in Format statement 103 near line number 1943.  
The commas were removed.  The resulting line became:

103   FORMAT(' ',1X,'***  SCREEN3 MODEL RUN  ***'
     &         /2X,'*** VERSION DATED ',A5,' ***'//1X,A79//
     &       1X,'SIMPLE TERRAIN INPUTS:',/,
     &       1X,'   SOURCE TYPE                  =        POINT ',
     &                                  'w/ DRY DEPOSITION'/
     &       1X,'   EMISSION RATE (G/S)          = ',G16.6,/
     &       1X,'   STACK HEIGHT (M)             = ',F12.4,/
     &       1X,'   STK INSIDE DIAM (M)          = ',F12.4,/
     &       1X,'   STK EXIT VELOCITY (M/S)      = ',F12.4,/
     &       1X,'   STK GAS EXIT TEMP (K)        = ',F12.4,/
     &       1X,'   AMBIENT AIR TEMP (K)         = ',F12.4,/
     &       1X,'   RECEPTOR HEIGHT (M)          = ',F12.4,/
     &       1X,'   URBAN/RURAL OPTION           = ',7X,A5,/
     &       1X,'   SURFACE ROUGHNESS LENGTH (M) = ',F12.4,/
     &       1X,'   BUILDING HEIGHT (M)          = ',F12.4,/
     &       1X,'   MIN HORIZ BLDG DIM (M)       = ',F12.4,/
     &       1X,'   MAX HORIZ BLDG DIM (M)       = ',F12.4)

6) The version number, VERSN, was changed in Block Init from 96043 to 13043:

      DATA VERSN/'13043'/

7) In the CALL PRISE statement, the scalar variable, xnw, was being passsed to
the dimensioned variable, xdown. xdown was always dimensioned to "1".  In
Subroutine PRISE, xdown was changed to the scalar, xnw.  The same was true for
z and zeff.  These two variables were changed to znw and zefflgl.  The 
subroutine is now:

      subroutine PRISE(istab,ws,ta,ts,hs,exitw,diam,fmterm,xnw,ndown,
     1 fbterm,znw,zefflgl,xmax,zmax)

8)  xnw, znw, and zefflgl were equated to their dimensioned counterparts:

c     scalar arguments (xnw,znw,zefflgl) are passed by 'call prise' to
c     dimensional constants (xdown, z, and zeff) with a value of 1 (ndown).
c     The following three lines were added as well as a change to the 
c     sub args to address a message citing this previous coding 
c     as a compiler error: 
      xdown(ndown) = xnw
      z(ndown) = znw
      zeff(ndown) = zefflgl

and after processing, znw and zefflgl were converted back:

c      
c     added to pass results back through call statement as scalar
c     arguments:
      znw = z(ndown)
      zefflgl = zeff(ndown) 

xnw was not need after the Return was made.

9)  The compiler objected to mixed integer and floating point variable 
arguments in the function AMAX1.  The value 10 was changed to a float point 
value by adding a decimal point.  There were two other instances of this
happening.  All three statements were edited to:

         UC10M(I) = UC * (ZREF/AMAX1(10.,HS))**0.20
and:

      U0 = U0TEN*(AMAX1(10.,HS)/ZREF)**0.20
      U1 = U1TEN*(AMAX1(10.,HS)/ZREF)**0.20
      
10)  The G10.4 formats were widened to G11.4 formats due to a remark
generated by the compiler.  Several format statements with dashed lines had
to be modified so the new format would align with the dashes above the
output.

11)  Source code SCREEN3C.FOR was not modified.
       
Any questions should be directed to us via the SCRAM web site "Contact Us"
link.