Hello,
Usual use of PREVIOUS COMPUTATION FILE is to run a computation from a previous one using the same geometry file.
As in a PREVIOUS COMPUTATION FILE, you have elevation of every node, the bottom elevation is classically taken from the elevation of the bottom plane.
That is what you can currently see with your setup.
I can see 1 work around to try to do what you want.
Put telemac3d_init.F in your FORTRAN FILE.
Find the loop after the line
CALL READ_DATASET(T3D_FILES(T3DPRE)%FMT,T3D_FILES(T3DPRE)%LU,
Replace the loop
DO K=1,NPOIN2
H%R(K)=Z(K+NPOIN2*(NPLAN-1))-Z(K)
ZF%R(K)=Z(K)
ENDDO
by
DO K=1,NPOIN2
H%R(K) = MAX(Z(K+NPOIN2*(NPLAN-1))-ZF%R(K),0.D0)
ENDDO
With that, bottom elevation ZF should be kept from what is done before and not replaced by elevation of bottom plane read in your PREVIOUS COMPUTATION FILE.
It should also prevent from getting negative water depth and issues when recomputing the 3D mesh.
I have not tested it but do not hesitate to give your feedback.
I hope that the bottom elevation from your geometry file and the bottom elevation from your previous computation file are not so inconsistent so that values may not lead to other issues.
Hope this helps,
Chi-Tuan