Welsh 



Table 1 



To appreciate the results of these experiments, I have plotted 

 the initial and successive 5-day predicted positions of the 16°C 

 isotherm for each experiment in Figs. 4 (Experiment 1) and 5 (Experi- 

 ment 2). A comparison of these two figures shows Experiment 2, with 

 its greater transports, to produce greater development. Although 

 our primary purpose here is only to show the general envelope of 

 predictions, continuity between 5-day positions can be discerned in 

 the vicinity of 65°W, where Experiment 2 clearly predicts faster 

 movements and greater amplitude development. The indicated wave 

 speeds in this area are about 7 cm see" for Experiment 1 (Fig. 4) 

 and 18 cm sec" for Experiment 2 (Fig. 5). Warren has suggested 

 a wave speed of 5 cm sec" , with which Experiment 1 more closely 

 agrees; as already noted. Experiment 1 gives volume transport in 

 close agreement with accepted values. 



For the third experiment, six successive, non-overlapping, 

 5-day SST analysis were computed for the period March 1-30, 1960. 

 The analysis should be valid for the middle of each time period, 

 i.e., 2.5, 7.5, 12.5, 17.5, 22.5, and 27.5 days into the month. 

 From each analysis, a prediction was made out to day 32.5 (April 1, 

 1960), with prediction outputs every 5 days after the analysis time. 

 For this experiment, the conversion factor 1500 m sec"-*^ °C~-^ , 



A^ = 3.5 



and 



\ = 7.5 



were used. 



Unfortunately, the six analysis, are themselves disappointing, 

 for they show poor map-to-map continuity. Moreover, it is the ex- 

 perience of the author that successive, non-overlapping periods of 

 ship reports cannot generally be analyzed with good continuity b e- 

 tween maps. It is not surprising, then, that the predictions in 

 this experiment did not bear much resemblence to the verifying 

 analysis. The analyses were simply inadequate. 



For all three experiments, a computation grid with about 60 km 

 between gridpoints and time steps of 6 hours were used. The value 



for 

 sec 



f()/hf^ on the right-hand side of Eq. (11) is 2.5 



1 X 10-^ sec"-"- for h 



-0'"0 



which corresponds to 



fo = 



10-5 ^-1 

 = 4000 m 



This represents a reduction of the effect of the bottom consider- 

 ably greater than the reduction used in our previous report. 



191 



