To compute the vertical decomposition all cases are vertically interpolated
 to the pressure levels of a reference case. Since the main reanalysis from
 ECMWF ERA_5 and NCEP/NCAR CFSR1 and CFSV2 has the same 37 levels, it is 
 defined as reference case ERA_5. Also ERA_5, CFSR1 and CFSV2 have suitable
 distibution of modes in the vertical classes: 5 modes in class 0, 6 modes in
 class 1, 8 in class 2, 10 in class 3 and 8 in class 4.

 Then, it is defined:
 csvm: reference case ERA_5 for vertical and horizontal modes
 Kmax: fixed number of pressure levels for all cases (same as csvm)
 Fixed vertical modes classes distribution for all case (same as csvm):
 Class 0 : z0a= 1 to z0b= 5
 Class 1 : z1a= 6 to z1b=11
 Class 2 : z2a=12 to z2b=19
 Class 3 : z3a=20 to z3b=29
 Class 4 : z4a=30 to z4b=37

 Were: zca is the first mode of class c and zcb the last.

 The reference case (csvm) and the number of pressure levels (Kmax)
 are defined at the file Reference_Case that is used by all the
 korn shell scripts and used by the GrADS script Case_Dates.gs

 The format inside Reference_Case must be the maintained to the
 scripts (korn shell and GrADS) work. At the korn shell is not
 allowed spaces between the variable, the sign equal and the 
 atributed value. At the GrADS scripts the values were read
 using the function substr after the equal sign.

 The fixed vertical modes distribution is defined at the GrADS
 script Case_Dates.gs