Southwest corner of model grid is 4351 N. Lat, 4920 W. Long 



Prediction starts; M3NTH 5 DAY 20 HR 1800 YR 

 Time step = 3600. SEC Grid size is 5 KM. 



Layer No. 



1 

 2 

 3 

 4 

 5 

 6 



Layer Depths 



- 

 30 ■ 



70 M. 

 70 M. 



70 - 150 M. 



150 

 300 

 500 



300 M. 

 500 M. 

 1000 M. 



Lateral Friction 

 0.3E 07 

 0.3E 07 

 0.1E 07 

 0.1E 07 

 0.1E 07 

 0.1E 07 



1980 



Diffusion Coef . 

 0.5E 05 

 0.5E 05 

 0.3E 05 

 0.1E 04 

 0.1E 03 

 0.1E 03 



Vertical Friction 

 0.005 

 0.005 

 0.005 

 0.005 

 0.005 

 0.005 



Vertical Friction Along Bottom/Water Interface is 0.010 



Figure B-2. — Computer Output Heading of the Labrador Current Model. 



direction of the current systems with time. 

 Hourly current predictions are available and the 

 model can be instructed as to the frequency of 

 output desired. The output consists of the V and 

 V vector velocities, temperature, and salinity dis- 

 tributions for each layer of the model. In addi- 

 tion, the. density distribution of the layers may 

 also be called out if desired. The initial heading 

 of the model output lists such facts as the loca- 

 tion of the area modeled, month/day /hour and 

 year of the start of the predictive period, the 

 time step, grid size and the various coefficients 

 used by the model as shown in Figure B-2. 



The model is presently being run on an IBM 

 360/370 computer at the University of Connecti- 

 cut, Storrs, CT. A 10 day predictive run takes 

 608,000 words of core memory and will use ap- 

 proximately 95 minutes of Central Processor 

 Unit (CPU) time. 



An increase in predictive accuracy of the pro- 

 gram can be realized through the boundary 

 monitoring of the modeled area at certain critical 

 locations. Additional data sampling across the 

 Labrador Current and the Gulf Stream where 

 they enter the modeled area will allow for in- 

 creases in both predictive accuracy and length of 



prediction. The monitoring of boundaries, if 

 clone by air deployable probes every 8 to 10 days, 

 could allow a probable predictive period of up 

 to a month before re-initiation is necessary. The 

 original one degree square model was tested suc- 

 cessfully for up to eight days. This expanded 

 model has not been tested against real data, so 

 the length of the predictive period is only spec- 

 ulative at this time. 



An extensive data handling system has been 

 developed to speed Hie preparation of the data 

 into a form which is directly usable by the model. 

 For any given data collection survey, the region 

 to be modeled is selected and the temperature/ 

 salinity/depth data are gathered throughout the 

 region using some form of a continuous vertical 

 sampling instrument. Data will generally be in 

 the form of magnetic or punched tapes which 

 contain individual station information. This in- 

 cludes the station number, latitude and longitude, 

 water depth and the serial depth/temperature/ 

 salinity distributions. After some degree of 

 quality control, generally in the form of either 

 eyeball analysis or computer processing, a data 

 set is produced. This data set may list depth/ 

 temperature/salinity distributions spaced as close 



B-3 



