I fixed a bug in noise simulations type 1. The noise source time function was being injected on all the domain slices instead of being injected only in the slice where the master receiver is located. This resulted in the propagation of NPROC noise sources instead of only 1.
Dear SPECFEM maintainers,
I fixed a bug in noise simulations type 1. The noise source time function was being injected on all the domain slices instead of being injected only in the slice where the master receiver is located. This resulted in the propagation of NPROC noise sources instead of only 1.
Best regards, Eduardo Valero Cano.