SUNMO - NMO for an arbitrary velocity function of time and CDP	     

  sunmo <stdin >stdout [optional parameters]				     

 Optional Parameters:							     
 tnmo=0		NMO times corresponding to velocities in vnmo	     
 vnmo=2000		NMO velocities corresponding to times in tnmo	     
 anis1=0		two anisotropy coefficients making up quartic term   
 anis2=0		in traveltime curve, corresponding to times in tnmo  
 cdp=			CDPs for which vnmo & tnmo are specified (see Notes) 
 smute=1.5		samples with NMO stretch exceeding smute are zeroed  
 lmute=25		length (in samples) of linear ramp for stretch mute  
 sscale=1		=1 to divide output samples by NMO stretch factor    
 invert=0		=1 to perform (approximate) inverse NMO	             
 ixoffset=0		do not consider cross-line offset                    
                       =1 read cross-line offset from trace header          
 upward=0		=1 to scan upward to find first sample to kill       

 Notes:								     

 For constant-velocity NMO, specify only one vnmo=constant and omit tnmo.   

 The anisotropy coefficients anis1, anis2 permit non-hyperbolicity due	     
 to layering, mode conversion, or anisotropy. Default is isotropic NMO.     

 For NMO with a velocity function of time only, specify the arrays	     
	   vnmo=v1,v2,... tnmo=t1,t2,...				     
 where v1 is the velocity at time t1, v2 is the velocity at time t2, ...    
 The times specified in the tnmo array must be monotonically increasing.    
 Linear interpolation and constant extrapolation of the specified velocities
 is used to compute the velocities at times not specified.		     
 The same holds for the anisotropy coefficients as a function of time only. 

 For NMO with a velocity function of time and CDP, specify the array	     
	   cdp=cdp1,cdp2,...						     
 and, for each CDP specified, specify the vnmo and tnmo arrays as described 
 above. The first (vnmo,tnmo) pair corresponds to the first cdp, and so on. 
 Linear interpolation and constant extrapolation of 1/velocity^2 is used    
 to compute velocities at CDPs not specified.				     
 The same holds for the anisotropy coefficients as a function of time and   
 CDP.									     

 Moveout is defined by							     

   1		 anis1							     
  --- x^2 + ------------- x^4.						     
  v^2	     1 + anis2 x^2						     

 Note: In general, the user should set the cdp parameter.  The default is   
	to use tr.cdp from the first trace and assume only one cdp.          
 Caveat:								     
 Nmo cannot handle negative moveout as in triplication caused by	     
 anisotropy. But negative moveout happens necessarily for negative anis1 at 
 sufficiently large offsets. Then the error-negative moveout- is printed.   
 Check anis1. An error (anis2 too small) is also printed if the	     
 denominator of the quartic term becomes negative. Check anis2. These errors
 are prompted even if they occur in traces which would not survive the	     
 NMO-stretch threshold. Chop off enough far-offset traces (e.g. with suwind)
 if anis1, anis2 are fine for near-offset traces.			     

 NMO interpolation error is less than 1% for frequencies less than 60% of   
 the Nyquist frequency.						     

 Exact inverse NMO is impossible, particularly for early times at large     
 offsets and for frequencies near Nyquist with large interpolation errors.  

 Credits:
	SEP: Shuki, Chuck Sword
	CWP: Shuki, Jack, Dave Hale, Bjoern Rommel
      Modified: 08/08/98 - Carlos E. Theodoro - option for lateral offset
      Modified: 07/11/02 - Sang-yong Suh -
          added "upward" option to handle decreasing velocity function.

 Technical Reference:
	The Common Depth Point Stack
	William A. Schneider
	Proc. IEEE, v. 72, n. 10, p. 1238-1254
	1984

 Trace header fields accessed: ns, dt, delrt, offset, cdp, sy