Wintertime Expansion and Contraction of the Terra Nova Bay Polynya
157
Fig. 6. Same as Fig. 5 except that the fractional cloud cover is allowed to take on either 0 or
100% in accordance with the U-shaped in-situ fractional cloud distribution
ing modeled polynya extents, associated with either 0 or 100% cloud fraction, that
were closest to the original régression line. Finally, the régression was recomputed using the revised time sériés of modeled polynya extent.
The scatter plot and the revised régression line are shown in Fig. 6a. The corrélation coefficient is 0.63, explaining 40% of the variance (Table 3). Figures 6b-d
show the open water fraction derived from the satellite data and the open water
fraction estimated by the polynya model and régression analysis. The mean modeled polynya extent fbr this case is 17±3 km (Table 3), which is similar to the value
derived from the constant cloud cover case. The improvement in the corrélation
coefficient suggests that the unknown cloud fraction can explain some of the
observed fluctuation in polynya extent. However, the model still fails to completely capture the large polynya expansions during May and June 1988 and July
1989. These events are discussed more fully below.