rate budget with imbalance BUT MODEL CALIBRATED!!

hello everybody

I am having an issue with a groundwater model that was given to me !!!, with 9 layers, thickness of 1st layer is 2 m (kx 8.6m/d), 2nd layer 10m (kx=0.0000864m/d), layer 3 15m (kx obtained by pest), and the other layers.... my interest regards the thrid layer.   
*  I have some layers with thickness ranging from 44 meters to 25m, and the size of the elements range from 2m to 25 towards the border of m model
*  with a BC to the north water table that is  +-267m and the lower BC (south) 256.44,
*  the area of entrance  is 1500m to the north and to the south is 3000m for each aquifer.
*  the model is covenging with an error of 15 centimeters OBSERVED VS CALIBRATED, but the problem is that i dont know why the RATE BUDGET IS GIVING MY VALUES COMPELTELY WORONG

I have an imbalance which is twice the entrance and i dont know to to fix this issue, discretization?

the model converges without problem, there is only one  log >>>>          PCG has a matrix with 159023th diagonal element <= 0.

I REALLY APPREACIATE YOUR EXPERIENCE TO HELP ME SOLVE THIS ISSUE!!! ( enclosed you have an image showing the 1st layer RESULTS WITH THE VALUES FOR THE RATE BUDGET)

ALEXANDER

The diagonal in the matrix must not be <= 0. There are several possible reasons for negative entries along the principal diagonal of the matrix. One possible reason is related with strong contrasts in material properties of adjacent elements. Try to refine the mesh horizontally/vertically. Another possible reason may be related with topological errors of the mesh triggered by inconsistent input data (e.g. overlapping elements/element edges, wrong numeration of the nodes). Accordingly, the model cannot be considered as being "calibrated".