GRIDOPS()

Fortran version:

    SUBROUTINE GRIDOPS( NCOL, NROW, NSPC, NLEV, A, B, C )
    ENTRY      PICKOPS( OPNAME )
    ENTRY      NAMEDOP( OPNAME, NCOL, NROW, NSPC, NLEV, A, B, C )
        INTEGER, INTENT(IN   ) :: NCOL, NROW, NSPC, NLEV     !  dimensions
        REAL   , INTENT(IN   ) :: A( NCOL*NROW*NSPC*NLEV )   !  first  input grid
        REAL   , INTENT(IN   ) :: B( NCOL*NROW*NSPC*NLEV )   !  second input grid
        REAL   , INTENT(  OUT) :: C( NCOL*NROW*NSPC*NLEV )   !  output grid
        CHARACTER*16  OPNAME                     !  operation name (see below)

NO C version:

Summary:

Main routine GRIDOPS generates C = A op B, where op is an operation defined in terms of state variable DIFMODE (as specified below). This is most useful when you're always doing the same operation to the data.

Entry PICKOPS is used to set state variable DIFMODE, and returns the name of the corresponding operation. Uses GETMENU() to select the operation.

Entry NAMEDOP takes an operation name (from PICKOPS), and a full set of GRIDOPS arguments, selects the appropriate DIFMODE, and goes the head of GRIDOPS. This is useful when you alternating among a set of operations you're applying to the data. In this case, it is useful to initialize a table of operations to be performed via PICKOPS.

The supported operations are the following

 1  (pointwise) difference                             A - B
 2  (pointwise) difference                             B - A
 3  (pointwise) ratio                                  A / B
 4  (pointwise) ratio                                  B / A
 5  (pointwise) absolute value of difference          |A - B|
 6  difference normalized by first grid               (A - B)/A
 7  difference normalized by second grid              (B - A)/B
 8  difference normalized by second grid              (A - B)/B
 9  absolute value of difference normalized by A      |A - B|/A
10  absolute value of difference normalized by B      |A - B|/B
11  difference normalized by pointwise mean          2(A - B)/(A + B)
12  difference normalized by pointwise mean          2(B - A)/(A + B)
13  difference normalized by joint root mean square   (A - B)/RMS(A&B)
14  difference normalized by joint root mean square   (B - A)/RMS(A&B)
15  (pointwise) sum                                    A + B
16  (pointwise) maximum                                MAX( A,B )
17  (pointwise) minimum                                min( A,B )
18  value from grid A                                  A
19  value from grid B                                  B
20  (pointwise) product                                A * B

NOTE Nov. 10, 2020: Operation 20 (pointwise product) was introduced for I/O API-3.2 on , by Calvin Arter, UNC Institute for the Environment, especially for use with mask-files.

Preconditions

Valid OPNAME for NAMEDOP (e.g., a name obtained from PICKOPS).

Fortran Usage:

    ...
    REAL           A( 50, 40, 10 )
    REAL           B( 50, 40, 10 )
    REAL           C( 50, 40, 10 )
    CHARACTER*16   OPNAME
    ...
    CALL NAMEDOP( 'A - B', 50, 40, 1, 10, A, B, C )   !  now C = A - B
    ...
    CALL PICKOPS( OPNAME )  !  user selects operation
    ...
    CALL GRIDOPS( 'A - B', 50, 40, 1, 10, A, B, C )   !  now C = A op B
    ...

C Usage:

Next: INDEX1

Up: Utility Routines

To: Models-3/EDSS I/O API: The Help Pages