Errors introduced into the weather message 

 for various reasons are often difficult to 

 note unless they are of serious nature. Data 

 screening at the BCF Biological Laboratory 

 in San Diego is designed primarily to detect 

 the following errors: 



1. Departures from major 5-digit group- 

 ings: e.g ., teletype garbles, run-on groups 

 of 5 or more digits not separated by a space, 

 failure to shift from lower to upper case, and 

 vice versa. 



2. Major coding errors inbarometric pres- 

 sure readings: all coded messages are inter- 

 preted for values between 950.1 and 1049.9 

 mb.; coded values greater than 500 are assigned 

 to the range 950.1-999.9 mb., and values less 

 than 500 are assigned to the range 1000.0- 

 1049.9 mb. Observed barometric pressures 

 will be refined further in the future when 

 monthly barometric averages and associated 

 statistics become available for screening 

 purposes. 



3. Major errors in sea temperatures: sea 

 temperature observations are screened by 

 comparison with a table of values interpolated 

 from monthly means for the north Pacific 

 Ocean found in the "Atlas of Climatic Charts 

 of the Oceans," U.S. Weather Bureau. Tem- 

 peratures outside a selected range are rejected 

 and placed in a summary listing for inspection 

 by the analyst. 



DISTRIBUTION OF OBSERVATIONS 



Because the data used are compiled pri- 

 marily from merchant marine weather reports, 

 the distribution of observations over the ocean 

 is irregular. Observations are most numerous 

 in the major shipping lanes from San Fran- 

 cisco to Hawaii, San Francisco to Japan, 

 Panama to San Diego, Los Angeles and San 

 Francisco, and from Seattle to Japan. Shipping 

 lanes of secondary importance are from San 

 Diego to Hawaii, Panama to Hawaii, and 

 Panama south to the ports off the west coast 

 of South America. Vast regions of the equa- 

 torial zone and south Pacific Ocean have 

 much less merchant vessel traffic than the 

 north Pacific and, consequently, fewer reports 

 are received from these areas. Sample dis- 

 tributions of the number of usable sea tem- 

 perature observations are given in figures 

 1 and 2. 



In recent years, progress has been made 

 in obtaining coverage from some regions that 

 heretofore have yielded few reports. One quite 

 important region that has been little covered 

 is the area off the west coast of South America. 

 The importance of this area canbe appreciated 

 >y examining the world fishing statistics. 



In 1961, FAO reported that Peru was surpassed 

 only by Japan in world production of fish 

 and fish byproducts. Recently, WMO has asked 

 that cooperating ships steaming in the area 

 south of the equator and from the west coast 

 of South America to long. 140° W. send reports 

 to "Observer San Francisco." Furthermore, 

 the U.S. Weather Bureau has recently estab- 

 lished a Port Visitation Office in Panama and 

 is encouraging vessels passing through the 

 canal into the Pacific to send reports to 

 designated receiving agencies. 



AUTOMATIC DATA PROCESSING OF 

 WEATHER REPORTS 



The Program 



Incoming data are received from the U.S. 

 Weather Bureau, San Francisco International 

 Airport, San Bruno, Calif., via the landline 

 teletype circuit, Service "O", Circuit 8274, 

 mentioned previously. Each transmission is 

 converted to a 5-channel paper tape by a 

 teletype reperforator receiver. Concurrent 

 with the reperforator operation, the trans- 

 mission is duplicated on a page printer 

 machine for message verification purposes 

 and as a backup when the reperforator mal- 

 functions. 



The perforated tape is fed into an IBM 

 047 tape-to-card key punch machine to convert 

 each ship's weather observation to standard 

 data card format. These data cards are then 

 accumulated each month for processing. The 

 data are presorted by octant and longitude 

 into 10-degree wide strips to facilitate proc- 

 essing by the machine and to provide a 

 logical output format. The information on 

 each of the sorted cards is then loaded onto 

 magnetic tape for high-speed input to the 

 computer. The basic steps are schematized in 

 figure 3. 



The program in current use is written in 

 FORTRAN-62 source language for the Control 

 Data Corporation 1604 computer located at 

 the University of California, San Diego. A 

 machine language deck has been prepared from 

 the source language statements for routine 

 processing of synoptic data. The computer 

 program processes the cards and prepares 

 output summaries for each 10-degree strip of 

 longitude separately until all accumulated 

 cards are processed. Data cards are sum- 

 marized at the rate of 5 per second and the 

 output tape is produced at the rate of 40 

 lines per second. The output tape is subse- 

 quently listed on multicopy paper for tabulation 

 assembly. The flow charts in figures 4-6 

 depict the major data processing functions 

 performed; numerical designators are keyed 

 to processing statements within the program 

 listing in Appendix A. 



