Why delay in UKESM1?

The feedback we had was:

More trees + shrubs & less grass in high latitudes (compared with GC3.1)
In winter, trees are above the snow
In winter, increased albedo
In winter, anamalous warning in this region
AMOC drops from about 15 Sv to 12 Sv

The fix

There is a parameter which sets the amount of snow on a leaf. In GC3.1 this is very low, but we've considerably increased this for UKESM0.91

We now expect to be running PI control in mid-December.