icebergs failure in 1/12 degree configuration

[MJHarrison-GFDL]

I am running a 1/12 global MOM6-SIS2-icebergs configuration which is failing with the following error:

====

WARNING from PE 380: diamonds, adjust: Berg did not move or bounce during iterations AND was not in cell. Adjusting!

diamonds, adjust: The adjusting iceberg is: 4310241141 380
diamonds, adjust: The adjusting lon,lat,u,v: -105.906374305403
89.9858827691345 -4.581405195299609E-002 0.170125968426817
diamonds, adjust: The adjusting xi,ji: 0.978580521913551
0.881252174498347
pos_within_cell: lon=[ -72.189972, -42.445068, -105.879997, -105.879997]
pos_within_cell: lat=[ 89.968498, 90.000000, 89.984253, 89.964783]
pos_within_cell: x,y= -105.906374, 89.985883
pos_within_cell: lon=[ 0.009635, 0.000000, -0.004309, -0.009636]
pos_within_cell: lat=[ -0.029992, 0.000000, -0.015146, -0.033873]
pos_within_cell: x,y= -0.003869, -0.013577
calc_xiyj: coeffs alpha,beta,gamma -9.6352E-03 -1.9271E-02 1.4963E-02 2.9992E-02 -3.8812E-03 -1.1265E-02
calc_xiyj: coeffs delta,epsilon,kappa -9.6352E-03 -1.9271E-02 1.4963E-02 2.9992E-02 -3.8812E-03 -1.1265E-02
calc_xiyj: coeffs dx,dy= -1.3504E-02 1.6415E-02
calc_xiyj: coeffs A,B,C= 2.7516E-04 -5.2189E-04 2.4686E-04
calc_xiyj: coeffs D= 1.6733E-10
Roots for b/2a, sqrt(d) = 9.4833E-01 4.7011E-02
Roots for y = 9.0132E-01 9.9534E-01 9.0132E-01
calc_xiyj: xi,yj= 0.1004E+01 0.9013E+00
pos_within_cell: xi,yj= 1.003769 0.901318
sum_sign_dot_prod5: x=380: -72.1900 -72.1900 -105.8800 -105.8800 -105.8800 -105.9064
sum_sign_dot_prod5: X=380: -72.1900 -72.1900 -105.8800 -105.8800 -105.8800 -105.9064
sum_sign_dot_prod5: y=380: 89.9685 90.0000 90.0000 89.9843 89.9648 89.9859
sum_sign_dot_prod5: l=380: -1.0621E+00 -4.7561E-01 4.1536E-04 5.1357E-04 -7.1096E-01
sum_sign_dot_prod5: p=380: -1.0000 -1.0000 1.0000 1.0000 -1.0000
calc_xiyj: 380lon=[ 0.01, 0.00, -0.00, -0.01]
calc_xiyj: x2..x4 - x1380 -0.01 -0.01 -0.02
calc_xiyj: 380lat=[ -0.03, 0.00, -0.02, -0.03]
calc_xiyj: y2..y4 - x1380 0.03 0.01 -0.00
calc_xiyj: coeffs alpha..kappa380 -9.6352E-03 -1.9271E-02 1.4963E-02 2.9992E-02 -3.8812E-03 -1.1265E-02
calc_xiyj: b<0 in quadratic root solver!!!!380
calc_xiyj: coeffs a,b,c,d,dx,dy380 2.7516E-04 -5.1379E-04 2.4218E-04 -6.4594E-10 -1.3528E-02 1.6975E-02

FATAL from PE 380: diamonds, calc_xiyj: We have complex roots. The grid must be very distorted!

====

icebergs_nml:use_roundoff_fix=.true.