At first, we set up a standard test case to compare errors among the ordinary
and compact difference (fourth, and sixth) schemes:
Figure 3 displays results from the sixth-order compact case: the error in the
streamfunction and velocity fields after performing 5 days of integration. The
mass transport streamfunction has a large-scale eight-lobe pattern centered on
the seamount. This symmetric structure can be found in all the fields. After 5
days of integration, the model generates spurious currents of O(0.03 cm/s),
much smaller than that of O(6 cm/s) generated with the second order scheme and
O(0.2 cm/s) generated with the fourth order ordinary scheme (Figures 1 and 2
in McCalpin, 1994).