1. Using initial and updated velocity files

   The initial velocity file must be in the same coordinate system as the PRONTO model. The best method to get the proper coordinates is to run PRONTO once with the linear velocity option (iop_vin=1). Then alter the output velocity file (vel.fin) to have the desired velocities and rename this file to vel.init. Now re-run PRONTO with iop_vin=2 to use an a priori velocity distribution.

   If you start with a new vel.init file, PRONTO will not do an inversion. I think the problem may be with the coordinates in the vel.init file. When I first created a vel.init file, my coordinates were at simple grid nodes, like:
   

   -1.000000       -0.695100       300.000000
   -0.500000       -0.695100       300.000000
   

   After one run, the coordinates in vel.fin correspond to the nodes in the PRONTO model, for example:
   

      -1.207300     -0.6952000       1149.422    
     -0.9572999     -0.6952000       1148.259    
   

   The solution is to change the velocities in the vel.fin file to the desired initial velocities.
   WPC-10/98

2. Velocity gradients

   I originally used a near surface velocity of 300 m/s overlying a velocity of 2500 m/s. PRONTO produced the message:

     Error! - No exit option selected in RAYCROSS.
               Receiver no. =   28
               Raypath entering cell: i=     7 j=    32  Raypath abandoned!
     Error! - No exit option selected in RAYCROSS.
               Receiver no. =   29
               Raypath entering cell: i=     7 j=    32  Raypath abandoned!
   

   This large a gradient may be too much for the finite difference forward modeller to handle. I decreased the upper velocity to 1000 m/s and got the program to run to completion.
   WPC-10/98

3. Small grid spacing

   I get the program to run with an 0.1 m grid spacing if I set the zero'th order constraint to 100! If I don't do this, I can't get a successful inversion

   I can't get the program to run with grid spacing less than 0.1. Here's the error message:

           LSQR iteration =  19   relative error = nan
           LSQR iteration =  20   relative error = nan
           LSQR iteration =  21   relative error = nan
           LSQR iteration =  22   relative error = nan
           LSQR iteration =  23   relative error = nan
           LSQR iteration =  24   relative error = nan
           LSQR iteration =  25   relative error = nan
         Slowness model updated.
         Iteration number =   2
         Begin forward computations:
           Source number   1: Traveltimes done.
     Error! - No exit option selected in RAYCROSS.
               Receiver no. =    1
               Raypath entering cell: i=    11 j=   205  Raypath abandoned!
     Error! - No exit option selected in RAYCROSS.
               Receiver no. =    2
               Raypath entering cell: i=    11 j=   205  Raypath abandoned!
     Error! - No exit option selected in RAYCROSS.
               Receiver no. =    3
               Raypath entering cell: i=    11 j=   205  Raypath abandoned!
   

   Could the problem be in the LSQR routine? Or perhaps in raystart or raycross?
   WPC-10/98

4. Constraint weights

   I am confused about the constraint weights, especially the zeroth order constraint. What does it mean? Do I need to assign a value to the zero'th order constraint. Help!!

   I am not clear on the significance of the weighting values for the constraint equations. How large can they be > 100. What is the difference between 10 and 100 in terms of the constraints?

5. Velocity models

   What is the significance of the reference velocity. How does it factor into the inversion procedure. The inverse model tries to match the reference velocity model? Do I make the reference model as close to known values, like VSP or borehole, or surface information. I don't know a good value for the subsurface velocity, otherwise, I would not be inverting the data!

Return to PRONTO documentation.