Large values of sig1 and sig2.

[dabail10]

In the subroutine principal_stress we have the following:

        sigP(i,j) = -p5*stressp(i,j)

        ! normalized principal stresses
        sig1(i,j) = (p5*(stressp(i,j) &
                  + sqrt(stressm(i,j)**2+c4*stress12(i,j)**2))) &
                  / strength(i,j)
        sig2(i,j) = (p5*(stressp(i,j) &
                  - sqrt(stressm(i,j)**2+c4*stress12(i,j)**2))) &
                  / strength(i,j)

Where sig1 and sig2 are nondimensional. However, there are occasionally instantaneous values of sig1/sig2 which are very large. That is on the order of 1000. I assume this is a bug and we have an off by a timestep thing here where the strength and the stressp/stressm/stress12 terms are out of sync. This generally seems to happen where the ice concentration is very small (1.0e-5). Or perhaps this is fine.