=============================================================================== International Comprehensive Ocean-Atmosphere Data Set (ICOADS): Release 2.4 Translation Specifications: US Maury Collection 22 September 2002 =============================================================== Document Revision Information (previous version: 9 September 2002): Updates (heading only) for Release 2.4. Author: S. Woodruff ------------------------------------------------------------------------------- {1. Introduction} The US Maury Collection data were extracted from NCDC (1998), and processed through a series of quality control steps. The CD-ROM data are split into a separate file corresponding to each original microfilm reel, keyed into the format described by [1]. Two parallel files (per reel) were then used as input to this translation: (1) The pre-processed records in the digitization format; and (2) abbreviated records in the quality control ("qc") format, including time elements adjusted to the 24-hour clock, interpolated positions, and other information. References to the qc format are by "qc" with field number appended [2]. The digitization format contains two types of 173-character records: (1) header, and (2) data. These two record types were linked together by voyage number (the concatenation of the 2-digit original microfilm reel number and 4-digit frame number, and a 1-digit voyage sequence number). The voyage number in the header record is named cvoyh and that in the data record cvoyd. These and other references to fields in the digitization format are by field name, as defined in . Each field name uniquely defines a character position or range of positions in either the header or data record. Within the digitization file, record type was determined by checking for blank in column eight, indicating a header record (column eight of the data record is the first position of year, which we assumed was always extant). The following consistency checks were made among header and data records, and the qc records, based on voyage number: (a) Each digitization file was required to begin with a header record, and each header record to contain a form type of 1 or 2, or the program terminated with an error. (b) The data records following a header record were required to agree in voyage number, or they (and the parallel qc records) were rejected. (c) The qc records were required to agree with the sequentially corresponding digitization (either header or data) records in voyage number, or both the qc and digitization records were rejected. The US Maury digitization data comprise two general form types: (1) daily, and (2) reports more than once a day. The data record format contains fields for up to three air and sea surface temperatures (each with a field available to indicate observation hour), one barometric measurement (also with a field available to indicate observation hour), and wind entries for the first, middle, and latter parts of the day. However, the second and third temperature groups were almost always missing in the keyed data, as was the barometer observation hour; thus these fields were ignored for purposes of constructing fields HR, AT, and SST. Moreover, as expected, an extant hour associated with the first temperature group was almost always associated with form type 1, and, when extant, was almost always set to 9am. Elms (1999) proposed assigning HR = 9am (unless another time was reported) to the form type 1 reports, and then using wind data from the latter part of the day, centered about 9am, to complete each daily report: Part of day Hours ------------ ----- First 12-20 Middle 20-04 Latter 04-12 As discussed in Elms (1999), the proposed method assumed that the parts of the day were uniformly defined in this way in the US Maury Collection, and that the maritime day generally ended at noon. This statement from Oliver and Kingston (1970) may also be applicable: "Daily entries were made in the journal or log- book for the previous 24 hours." However, it was later suggested (Mike Chenoweth, personal communication) that many temperatures may have actually been taken at local noon for observational convenience, rather than at 9am. But in this case it is not certain whether the temperature observation would normally have been taken at the beginning or end of each maritime day. In view of the many uncertainties, we adopted the following simplified procedure for form type 1 reports: --HR was set to missing. --Data from the first temperature group were always used for AT and SST (since the other groups held negligible amounts of data). --Wind data from the first (or only) part of the day were always used (since the other groups held smaller amounts of usable data). For form type 2 reports, the procedure was the same except hour was normally extant. We decided to use wind data from the first (or only) part of the day, rather than the latter, to simplify procedures and maximize the amount of wind data. As a result, however, it seems likely that the wind data used to construct the form type 1 reports are separated in time and space from the temperature data (perhaps by as much as a day in time and in ship's movement). An exception was made for a small number of voyages [3] keyed as form type 1, but that actually consisted (partly or entirely) of form type 2 data (as was verified through microfilm examination). Extant hours were assigned in the qc format for these voyages, or portions of the voyages, believed to be mis-labeled as form type 1. Otherwise qc hour was always missing for voyages labeled form type 1. The qc hour was utilized for setting HR in these form type 1 voyages, but form type was not changed. {2. Translation to LMR6 location and regular fields} In the event of a missing or erroneous year, month, latitude, or longitude, the report was written out to the reject file. For other fields that were assembled from external input, the input data (or a subset as indicated by "Attm5") were included in the error attachment in the event of data problems or unusable data (concatenation is indicated by "//"). Fields in the digitization format were missing if all blank. Representation of missing data in the qc format is specified in [2]. 1) B10 10-degree box calculated by {wrlmr6} 2) YR year input = year (qc3) 3) MO month input = month (qc4) 4) DY day input = day (qc5) 5) HR hour input = year, month, day, and hour (qc3-6); longitude (qc8); and form type (cft) (Attm5=uhr; 4-character field from step d) (a) Hour (qc6) was used for HR, regardless of form type (unless missing, it was multiplied by 100 after input, to yield hundredths). For form type 1, qc6 was extant only for a few voyages (or portions of voyages) in which form type was incorrectly assigned [3] (thus most form type 1 data were not subject to UTC adjustment). (b) Year, month, day were converted to julian day (via {ixdtnd}), unless: (i) julian calculation not attempted (missing year/month). (ii) julian calculation failed (missing/illegal day, e.g., 31 June). [NOTE: This calculation would also fail in the event of an illegal year/month, but the qc format contained no illegal values for year, month, hour, or longitude.] (iii) hour or longitude was missing (which would otherwise lead to a subsequent failure of the UTC calculation). In these 3 cases any extant hour was set to missing (since it could not be adjusted), and the remaining steps were skipped. (c) Hour 2400 was changed to 0000, and one was added to julian day (thus incrementing day, and year and/or month, if applicable). (d) Hour, longitude, and julian day were input to {rxltut}, which converted from local to UTC hour (uhr), and also returned UTC julian day. (e) {rxnddt} was called to convert UTC julian day to year, month, and day. (thus also incrementing year, month, and/or day, if applicable). Externals: {lmrlib.01E:ixdtnd,rxltut,rxnddt} 6) TI time indicator input = no corresponding field (a) TI=0 (nearest whole hour). 7) LON longitude input = longitude (qc7) (a) Divided by 100 (otherwise already in LMR units). 8) LAT latitude input = latitude (qc8) (a) Divided by 100 (otherwise already in LMR units). 9) LI lat/lon indicator input = latitude/longitude indicator (qc9; already in LMR units) 10) DCK deck input = no corresponding field (a) Set DCK=701 (US Maury Collection). 11) SID source ID input = no corresponding field (a) Set SID=69 (US Maury Collection). 12) PT platform type input = no corresponding field (a) Set PT=5 (ship). 13) QI quality indicator input = no corresponding field 14) DS dup status input = no corresponding field 15) DC dup check input = no corresponding field 16) TC track check input = no corresponding field (a) Set TC = 1 (track checked). 17) PB pressure bias input = no corresponding field (a) If SLP is extant, this field was set as follows depending on whether a temperature correction was applied (a gravity correction was always applied): PB=1: temperature and gravity corrections applied PB=2: only gravity correction applied 18) DI wind direction indicator input = no corresponding field (a) DI=1 (32-point compass). [NOTE: As discussed in Elms (1999), this was an assumption; individual ships may have used, e.g., 16- or 8-point compasses.] 19) D wind direction input = wind direction (cwd1) (a) These abbreviations (only if absent any trailing characters) were converted to special LMR codes as follows: Abbrev. LMR code ------------ -------- C (calm) 361 V (variable) 362 B (baffling) 362 (b) If the abbreviation formed a recognized 32-point compass abbreviation, it was converted to numeric code (1-32) and then to degrees as follows (codes and degrees from Release 1, Table F2-1, DI=1 column): Abbrev. code degrees Abbrev. code degrees ------- ---- ------- ------- ---- ------- NXE 1 11 SXW 17 191 NNE 2 23 SSW 18 203 NEXN 3 34 SWXS 19 214 NE 4 45 SW 20 225 NEXE 5 56 SWXW 21 236 ENE 6 68 WSW 22 248 EXN 7 79 WXS 23 259 E 8 90 W 24 270 EXS 9 101 WXN 25 281 ESE 10 113 WNW 26 293 SEXE 11 124 NWXW 27 304 SE 12 135 NW 28 315 SEXS 13 146 NWXN 29 326 SSE 14 158 NNW 30 338 SXE 15 169 NXW 31 349 S 16 180 N 32 360 (c) Otherwise, including any abbreviations containing non-blank characters after position four, the abbreviation was considered unrecognizable and written out to the error attachment. 20) WI wind speed indicator input = no corresponding field (a) WI=5 (Beaufort force) 21) W wind speed input = wind force (cwf1) (a) The original data were either force 0-12, 2-digit codes 20-33, or 2- or 3-digit codes corresponding to additional text terms (as defined in ). Some of the 3-digit codes were equivalent in textual meaning to codes 20-32. All the recognized codes were mapped to force 0-12 as follows (for additional terms, and "--", W was set to missing): Beaufort (1805) 2-digit 3-digit Force --------------- ------- ------- ----- Calm 20 -- 0 Light air 21 -- 1 Light breeze 22 501 2 Gentle breeze 23 601 3 Moderate breeze 24 511 4 Fresh breeze 25 441 5 Strong breeze 26 571 6 Moderate gale 27 512 7 Fresh gale 28 442 8 Strong gale 29 572 9 Whole gale 30 -- 10 Storm 31 -- 11 Hurricane 32 -- 12 Baffling 33 -- -- (b) Both original and mapped force were converted to meters per second using {fxbfms}; i.e., the "old" midpoints from Release 1, Table K5-5. (c) All unrecognized data were written out to the error attachment. External: {lmrlib.01E:fxbfms} 22) VI visibility indicator input = no corresponding field 23) VV visibility input = no corresponding field 24) WW present weather data not converted = present weather (cx) and indicator (cxi) [NOTE: Fields cx and cxi are available in the supplementary attachment (the bulk of the data were reported as Beaufort or "4-choice" weather codes, as opposed to WMO Code 4677). But the data were not used because of conversion questions, and data quality concerns including in the coding of non-significant versus missing data (Elms, 1999).] 25) W1 past weather input = no corresponding field 26) W2 second past weather input = no corresponding field 27) SLP sea level pressure input = barometric pressure (cb), 1st attached thermometer (cbt1), and temperature indicator (ct1) or temperature corrections [4] (Attm5=cb//cbt1 after any corrections) (a) Trailing/leading blanks in the 4-digit cb field were converted to zero. Then an internal read converted the field to an integer (letting "b" represent blank, thus ensuring, e.g., that 28bb, 280b, and 2800 all yield 2800.) Barometer units were inferred, and the integer value divided accordingly, based on the first digit of cb: (i) If 2 or 3, English inches were assumed (division by 100) (ii) If 7 , millimeters were assumed (division by 10) (b) The temperature correction was calculated depending on the (original) temperature units (ct1): (i) Fahrenheit: using {fwbptf} (ii) Celsius: using {fwbptc} and added to the reported pressure reading (PB is set to 1). However, the calculation was contingent on these conditions (otherwise pressure was left uncorrected, and PB set to 2): (i) The 1st attached thermometer reading was required to fall within the range -50:50C (it was tested in a form already converted to Celsius, and corrected if applicable; see AT for details about the handling of all temperatures). (ii) If the original temperature units were Fahrenheit (Celsius), barometer units were required to be inches (millimeters). The remaining steps were executed regardless of whether the temperature correction was made. (c) The gravity correction was calculated by {fwbpgv} using gmode=2, and added to the pressure. (d) Barometer units were transformed into hPa from: (i) inches: using {fxeimb} (ii) millimeters: using {fxmmmb} Externals: {lmrlib.01E:fwbptf,fwbptc,fwbpgv,fxeimb,fxmmmb} [NOTE: Inventories showed that the 2nd and 3rd attached thermometer fields (cbt2 and cbt3) and the hour of barometer observation (chb), were almost always missing, so these fields were not used. There was no indication in the keyed data to indicate barometer type (e.g., aneroid or mercurial); we assumed all were mercurial in this processing.] 28) T1 temperature indicator input = temperature indicator (ct1) or temperature corrections [4] (a) If the voyage was subject to temperature corrections, ct1 was ignored and: T1=9 (other; refer to metadata). (b) Otherwise if ct1=1, T1=7 (degrees Fahren.; mixed/undetermined precision) or if ct1=2, T1=3 (degrees Celsius; mixed/undetermined precision) [NOTE: In the keyed data, only ct1=1 or 2 values were present (temperature values were not tested for the presence of a blank/numeric tenths digit).] 29) AT air temperature input = air temp. (1st entry; ca1) and indicator (ct1) or temperature corrections [4] (Attm5=ca1 after any corrections) (a) If the voyage was subject to temperature corrections, units were obtained from [4] rather than ct1, and the temperature value was shifted if applicable. (b) The field was divided into a tenths character (fourth position) and higher-order characters (first three positions). The two segments of the field were converted into separate integers, and then the real value was created (higher-order + tenths-character/10). Default input on the higher-order positions had the effect of right-justifying the data in the event of trailing blanks within those three positions (e.g., letting "b" represent blank, "89bb" became 89.0 rather than 890.0 as it would under a rigorous format interpretation). (c) If ct1=2, or if the units were specified as Fahrenheit by [4], {fxtftc} was used to convert to Celsius (otherwise data were already in Celsius). [NOTE: 1st attached thermometer (cbt1; for SLP) and water temperature at surface (cs1; for SST) were handled similarly. External {fxtrtc} was not used because no data were flagged as Reaumur.] Externals: {lmrlib.01E:fxtftc,fxtrtc} 30) WBT wet bulb temperature input = no corresponding field 31) DPT dew point temperature input = no corresponding field 32) SST sea surface temperature input = water temp. at surf. (1st entry; cs1), and indicator (ct1) or temperature corrections [4] (Attm5=cs1 after any corrections) (a) (See AT for details about the handling of all temperatures.) Externals: {lmrlib.01E:fxtftc,fxtrtc} 33) SI sea surface temperature method indicator input = no corresponding field 34) N total cloud amount input = proportion of sky clear (csc) (a) Proportion of sky clear was converted to oktas of sky covered by {ixt0ok}. External: {lmrlib.01E:ixt0ok} 35) NH lower cloud amount input = no corresponding field 36) CL low cloud type input = no corresponding field 37) HI cloud height indicator input = no corresponding field 38) H cloud height input = no corresponding field 39) CM middle cloud type input = no corresponding field 40) CH high cloud type input = no corresponding field 41) WD wave direction input = no corresponding field 42) WP wave period input = no corresponding field 43) WH wave height input = no corresponding field 44) SD swell direction input = no corresponding field 45) SP swell period input = no corresponding field 46) SH swell height input = no corresponding field 47) C1 country code input = no corresponding field 48) C2 second country code input = no corresponding field 49) SC ship course input = no corresponding field 50) SS ship speed input = no corresponding field 51) A barometric tendency input = no corresponding field 52) PPP amount of SLP change input = no corresponding field 53) IS ice accretion input = no corresponding field 54) ES ice thickness input = no corresponding field 55) RS ice accretion rate input = no corresponding field 56) II ID indicator input = no corresponding field Tests were made against ID after editing (next field): (a) If ID = blank, II=missing (b) Otherwise, II=10 (composite information from early ship data) 57-64) ID(8) ID/call sign input = name of ship (cnship) Extensive editing was used to reduce the 24-character cnship field to a set of unique identifiers eight or fewer characters in length. First a set of global rules was applied, then abbreviation or truncation of selected names as specified in Table 1, and finally the resulting string was truncated at eight characters and stored in the ID array. Also the editing changed some cnship fields to better match others (Table 2). Following are the global editing rules: (a) Unnecessary (embedded) blanks were removed (trailing blanks were ignored). A blank was removed if: (i) Two (or three) blanks occurred together. (ii) Blank followed period (". ") (tests showed no occurrences of blank preceding period). (iii) Blank preceded or followed ampersand (" &" or "& "). (b) Remaining embedded blanks were replaced by underline ("_"). Table 1. Modification of individual ship names. Names containing the search substring (starting at position 1 within cnship) were transformed into the concatenation of: (i) the shorter replacement substring, plus (ii) a residual substring formed starting at the indicated position within cnship (in some cases this overlaps with part of the search string). An asterisk (*) suffixes the replacement substring, indicating modification (there were no asterisks originally in cnship), followed if applicable by an ampersand (&) to replace the word "and." A special rule converting cnship strings such as NO_LIST into blank is discussed in the footnote. -------------------------------------------------------------------------- Search substring (from pos. 1) Replacement Residual substring from ========================================================================== ABSTRACT_JOURNALS AJ* 19 ADELAIDE ADELA* 9 ALEXANDER ALEXA* 10 AMAZON AMA* 11 AMERICA_OF AME* 11 AMERICAN AMERI* 9 ASHBURTO ASHBUR* 9 BENJAMIN BENJA* 9 CAROLINE CAROL* 9 CATHERINE CATHE* 10 CELESTIAL CELES* 10 CHALLENGE CHALLE* 10 CHARLES_ CHAR* 8 CHARLOTTE CHARL* 10 CINCINNAT CINCIN* 10 CITY_OF CITY* 8 CORNELIA CORNE* 9 CRESCENT CRESC* 9 DEFIANCE DEFIA* 12 EARL_OF_ EARL* 9 EASTERN_ EAS* 9 ELECTRIC ELECT* 9 ELIZA-AND ELIZ*& 10 ELIZABETH ELIZAB* 11 EMPRESS_OF EMPRE* 11 FLYING FLYI* 7 FRANCES_ FRAN* 8 FRANCIS_ FRAN* 8 FRANKLIN FRANK* 12 FREDERIC FRED* 9 GENERAL GEN* 8 GEORGE_AND G*& 11 GEORGE_ GEO* 7 GOLDEN GOL* 7 HARRIET HARRI* 8 HENRY HEN* 6 ISAAC_HOWLAND I_HOW* 17 ISABELLA_ ISABE* 12 JEFFERSON JEF* 10 JUDGE_SHAW JUD_S* 14 KATE_AND_ K*& 10 KATHLEEN KAT* 12 MACEDONIAN MACEDO* 10 MAJESTIC MAJ* 12 MANCHESTER MAN* 14 MARGARET MAR* 9 MARY_ANNA MAR_A* 14 MARY_FRA MAR_FR* 9 MAYFLOWER MAYF* 13 MECHANICS MEC* 10 MERCATOR_ MERCA* 16 NO_ (blank)# NA NORTH_AMERICA N_AMER* 14 NORTHERN_ N* 9 PLYMOUTH PL* 9 PRINCE_OF P* 10 PRINCESS PR* 6 QUEEN_OF_ Q* 9 REINDEER_ REIND* 12 RESOLUTE_ RES* 13 RETRIEVER RETRI* 8 RICHARD RICHA* 8 SAMUEL SAM* 7 SARAH_H. SA* 6 SIR_ROBERT SIR_R* 12 SPARKLING SPARK* 10 SPLENDID SPLE* 12 STAR_OF_THE STAR_T* 13 STEPHEN STEP* 9 SULPHUR, SUL* 9 THOMAS_ THOM* 8 UNITED_STATES_ U.S.* 14 VIRGINIAN VIRGIN* 9 WESTERN WESTE* 8 WILLIAM_ WILLI* 8 YOUNG_EAGLE Y_EAG* 15 -------------------------------------------------------------------------- # This special rule converts NO_ followed by LIST, LIST_SHIP, SHIP_LIST, or SPIP_LIST to blank, rendering ID missing (see Table 2). ---------- Table 2. List of modified ship names (cnship) identical in positions 1-8 (as a result, or in spite, of editing). The "No." column gives the position of the name in a sorted inventory of the 2021 original names, and frequency gives the number of voyages containing the original name. -------------------------------------------------------------------------- Modified Original Frequency No. 12345678|9012345678901234 123456789012345678901234 ==== ========|================ ======================== ========== 1337 | NO SPIP LIST 1 1336 | NO SHIP LIST 1 1334 | NO LIST 4 1335 | NO LIST SHIP 1 52 AFRICAN_|MET.ABSTRACT AFRICAN MET. ABSTRACT 2 51 AFRICAN_|MET.ABSTRACT AFRICAN MET. ABSTRACT 1 406 CORAL_OF|_NEW_BEDFORD CORAL OF NEW BEDFORD 8 407 CORAL_OF|_NEW_BEDFORD_S CORAL OF NEW BEDFORD S 5 450 D.FERNAN|DO D.FERNANDO 1 449 D.FERNAN|DO D. FERNANDO 1 693 GEN*_JON|ES GENERAL JONES 4 690 GEN*_JON|ES GENERAL JONES 1 1248 MINERVA_|SMYTH MINERVA SMYTH 3 1247 MINERVA_|SMYTH MINERVA SMYTH 5 1647 SAM*_ROB|ERTS SAMUEL ROBERTS 4 1648 SAM*_ROB|ERTSON SAMUEL ROBERTSON 3 1844 THOM*B.W|ALES THOMAS B.WALES 3 1843 THOM*B.W|ALES THOMAS B. WALES 2 -------------------------------------------------------------------------- 65) OS observation source input = no corresponding field 66) OP observation platform input = no corresponding field 67) T2 second temperature indicator input = no corresponding field 68) IX station/weather indicator input = no corresponding field 69) WX wave period indicator input = no corresponding field 70) SX swell period indicator input = no corresponding field 71) IRD IMM receipt date input = no corresponding field 72) A6 allowance 6 flag input = no corresponding field {3. Contents of supplemental attachment (Attm4)} Data stored in the supplemental attachment consisted of the entire data record (173 characters); followed by a selection of fields from, or derived from, the associated header record (through character 241); and selected fields from the qc file (total 250 characters): # Pos. Total # Field Record range of pos. name type Description of field (of derived field) --- ------- ------- ----- ------ ---------------------------------------- 1 1-7 7 cvoyd data voyage number ... ... ... ... ... 47 172-173 2 cmvq data magnetic variation QC indicator NA 174-175 2 cts2 header (fr ship type, ctship, according to [5]) 4 176-177 2 cft header form type 5 178-193 16 comm header commander (first 16 positions only) [6] 6 194-217 24 cfr header from city 7 218-241 24 cto header to city 2 242-246 5 qc2 qc reel sequence number 5 247-248 2 qc5 qc day (local time) (99 indicates missing) 6 249-250 2 qc6* qc hour (local time) (99 indicates missing) --- ------- ------- ----- ------ ---------------------------------------- * Whenever qc6 was 24, zero was inadvertently written out to the supplemental attachment. This resulted from an error in the conversion program, but can be fixed by interpretation of hour zero as hour 24 of qc5 + 1 (as noted in [2], qc6 originally ranged 1-24, with 24 signifying hour 0 of the next day. As intended, qc5 was included in the supplementary attachment in original form. ---------- {4. Notes} [1] Reels keyed in the Collection Reels 1-2 and 45 were missing from the digital data obtained from the CD-ROM. This was explained by CD-ROM documentation "about.txt" (excerpted as follows) except that it erroneously refers to reel44 (references should be to reel45): "There were eighty-eight reels of microfilmed records in the collection. Reels One, Two, and Forty-four do not appear on the CD-ROM. Reel One is simply an inventory, and we were unable to produce clear paper copies from Reel Two. Reel Forty-four, keyed in a different format, was used in the pilot digitization project to help finalize the formats that were adopted for the remainder of the collection. Reel Forty-four will later be converted to a common format." At this writing, NCDC has been unable to locate the details of the "pilot" format used for reel45. Moreover, notes indicated that only about half the data from the reel were keyed, owing to some paper copies being lost enroute to China. These data will need to be re-digitized under a future project. [2] Quality control (qc) format: missing: description: qc1 voyage number (a7 n/a reel+frame+voyage sequence qc2 reel sequence number ,1x,a5 n/a orig. sequence number on reel qc3 year ,1x,i4 9999 local time qc4 month ,1x,i2 99 local time qc5 day ,1x,i2 99 local time qc6 hour ,1x,i2 99 local time 1-24 (=0 next day) qc7 latitude ,1x,i5 9999 per LAT qc8 longitude ,1x,i5 99999 per LON qc9 lat/lon indicator ,1x,i1) 9 per LI Four examples of the 41-character record layout (preceded by header lines): 1 2 3 4 12345678901234567890123456789012345678901 ----------------------------------------- 0300031 12345 0300031 12346 1850 12 31 24 -9000 35999 4 0300031 12347 9999 99 99 99 9999 99999 9 0300031 12348 1851 1 1 2 9999 99999 9 QC records that were blank following reel sequence number referred to header records. The reel sequence number referred to the original CD-ROM files. The files corresponded one-for-one with the pre-processed digitization files, in which voyage numbers 1402671, 1402681, 7602921, and 7602941 (with headers and/or data out of order) were reordered. Input of these records was handled as indicated by the format statement, so that fields most conveniently handled as integers (e.g., for input to {lmrlib} routines for conversion of time to julian day and UTC) were made available in that form. However, to detect a header record, year was read first as a character string to determine if if was blank, and in that case all the numeric fields (or individual fields that are missing in the non-header case) were converted to a uniform integer missing value. [3] Voyages mis-labeled as form type 1 (partially or totally form type 2) The following voyages were labeled form type=1 (daily), but also included extant hours in the quality control file for all or part of the voyage: Line range Approx.* Voyage Reel start end reports number Comments ---- ------------- ------- ------- --------------------------------------- 44 9668 17442 7764 4407351 entire voyage with extant qc6; atypical (assumed correct) hour pattern in data 50 10991 11892 902 5003701 92 reports with missing qc6 at start 58 1524 1755 232 5800651 42 reports with missing qc6 at start 71 4768 4876 84 7109391 91 reports with missing qc6 at start 72 6487 8199 1409 7204311 entire voyage with extant qc6 ------- ~10391 * The count of reports is approximate because the line range may encompass reports without these characteristics, or with them but rejected for other reasons. ---------- [4] Temperature units and other corrections The following 107 voyages (cvoyd) were subject to units or other corrections of the temperature fields (cbt1, ca1, cs1). The temperature indicator (T1) was set to 9 for reports subject to corrections. If a "C" ("F") appears under a field, Celsius (Fahrenheit) was used as the units regardless of the setting of ct1 (thus allowing mixed units within a report). A prefix of "L-" ("R-") for a field indicates that data (or most of the data within a given voyage) appear to have been digitized with an erroneous shift by one character to the left (right). To correct this, the data were shifted in the opposite direction via an "end-around" shift (i.e., characters ordered 1234 were reordered 4123 for "L-", or 2341 for "R-"). However, the shift was not made (data unchanged) if the end-around character (1 or 4) was not a blank. Units are always given for all three fields, but it should be noted that in many voyages (e.g., those footnoted) only a subset of fields was actually reported. cvoyd cbt1 ca1 cs1 cvoyd cbt1 ca1 cs1 ------- ---- ---- ---- ------- ---- ---- ---- 1502821 C C C 6703351 C F F 1502851 C C C 6703401 C F F 1502891 C C C 6900021 C C C 1502921 C C C 6900251 C C C 1502931 C C C 6900271 C C C 1502941 C C C 6900301 C C C 1502991 C C C 6900341 C C C 1503021 C C C 6900511 C C C 1503041 C C C 6900661 C C C 2701881 C C C 6900701 C C C 4605761 C C C 6900731 C C C 4605861 C C C 6900841 C C C 4605871 C C C 6901501 C C C 4605881 C C C 6901721 C C C 4605901 C C C 6901781 C C C 4905971 F F L-F 6901831 C C C 4905981 F F L-F 6901981 C C C 4906221 F F L-F 6902201 C C C 4906241 F F L-F 6902251 C C C 4906321 F F L-F 6902271 C C C 4906341 F F L-F 6902391 C C C 4906411 F F L-F 6902441 C C C 4906461 F F L-F 6902501 C C C 4906471 F F L-F 6902681 C C C 4906531 F F L-F 6902831 C C C 4906651 F F L-F 6902851 C C C 4906661 F F L-F 6902901 C C C 4906741 F F L-F 6903031 C C C 4906771 F F L-F 6903051 C C C 4906781 F F L-F 6903261 C C C 4906811 F F L-F 6903331 C C C 4906851 F F L-F 6903501 C C C 4907001 F F L-F 6903711 C C C 4907021 F F L-F 6903801 C C C 4907031 F F L-F 6903931 C C C 4907161 F F L-F 6904021 C C C 4907221 F F L-F 6904181 C C C 5601261 C C C 6904221 C C C 5601511 C C C 6904401 C C C 5601761 C C C 6904491 C C C 5601901 F C C 6904561 C C C 5700711 C F F 6904711 C C C 5700781 C F F 6904991* F R-F F 5700871 C F F 6905751* F R-F F 5701061 C C C 7400041 C C C 5701231 C C C 7400061 C C C 5701881 C C C 7400071 C C C 6703151 C F F 7400101 C C C 6703201 C F F 7400121 C C C 6703221 C F F 7400141 C C C 6703251 C F F 7400161 C C C 6703291 C F F 7506861* F R-F F 6703331 C F F 7702531 C C C 8405121 F C C ------- ---- ---- ---- ------- ---- ---- ---- * AT only; for voyage 7502531 there were ~10 apparent overflows of half a degree from ca1 into the first position of cs1 (no action taken for this). ---------- [5] Mapping of Type of ship field (ctship) into cts2 Inventories of the Collection indicated a total of 55 different patterns within the digitized 15-character ship type field, ctship. These were mapped into a 2-character designation, cts2 (1-51, plus blank for missing) as follows. In three cases (blank, 29, and 34) two apparently equivalent patterns were mapped to the same cts2 value: cts2 ctship (first) ctship (second) cts2 ctship (first) ctship (second) 12 123456789012345 123456789012345 12 123456789012345 123456789012345 -- --------------- --------------- -- --------------- --------------- (blank) NOT LISTED 1 3 MAST CLIPPER 2 BARK 3 BARQUE 4 BRIG 5 BRIGANTINE 6 BRIT. SCHOONER 7 BRITISH 8 BRITISH BARK 9 BRITISH BARQUE 10 BRITISH RMS 11 CHILIAN 12 CHILIAN BARK 13 CLIPPER 14 CUTTER 15 DUTCH BARK 16 DUTCH FRIGATE 17 FINNISH 18 FRENCH 19 FRENCH SLOOP 20 FRIGATE 21 GERMAN 22 GERMAN BRIG 23 PACKET 24 PILOT 25 PILOT BOAT 26 PORTUG.CORVETTE 27 PORTUGUESE BRIG 28 PRUSSIAN 29 R.M.S.BRITISH R.M.S.S.BRITISH 30 REVENUE 31 RUSSIAN CORVET 32 RUSSIAN STEAMER 33 SARDINI. FRIGAT 34 SCHOONER SCOONER 35 SCHOONER,HMS 36 SCREW SLOOP 37 SHIP 38 SLOOP 39 SLOOP OF WAR 40 SPANISH 41 STEAM FRIGATE 42 STEAMER 43 STEAMSHIP 44 SURVEY SHIP 45 SWEDISH BRIG 46 TRANSPORT 47 U.S. SHIP 48 U.S. SQUADRON 49 U.S. STORE SHIP 50 U.S.FRIGATE 51 WHALER -- --------------- --------------- -- --------------- --------------- [6] Truncation of Commander field (comm) To save space, the 24-character field was truncated after the sixteenth character. In general this eliminated only blanks in positions 17-24, except it resulted in non-blank truncation of the 46 following names: # 1234567890123456|78901234 # 1234567890123456|78901234 -- ----------------|-------- -- ----------------|-------- 1 ALFRED A. ENGVIS|T 24 NATHANIEL BROWN |SR. 2 ALVIN O.BLACKIST|ON 25 NATHANIEL J.KINS|MAN 3 ANTONIO RICARDO |GRACA 26 NATHANIEL PUTNAM| JR 4 BENJ.LOWNDES JOH|NSTON 27 NATHANIEL PUTNAM| JR. 5 CAPT-LIEUT TERIA|EFF 28 OLIVER H.SAUNDER|S 6 CHARLES H.KENNED|Y 29 P.A.MARTINS DE S|ILVA 7 CHRISTOPHER GILP|ATRICK 30 P.V.COSTANEIRO E| PINTAO 8 E.G.JILION COMMO|S 31 RICHARD F.CORNIN|G 9 FERDINAND CROCKE|R 32 RICHARD J.HOFFNE|R 10 FRANCIS A.BEERSL|EY 33 RICHARD T. CORNI|NG 11 FRANCIS F.NICKER|SON 34 ROB'T B. CUNNING|HAM 12 FRANKLIN BUCHANA|N 35 SAM HUTCHINSON J|R 13 HUMPHREY W.SEABU|RY 36 SAM'L EASTERBROO|K 14 ISAAC HARDING JR|. 37 SAM'L HUTCHINSON| JR. 15 JACOB L.CLEVELAN|D 38 SAMUEL BARTLETT |JR. 16 JAMES F. ARMSTRO|NG 39 STEPHEN JEFFERSO|N 17 JAMES LEMASSURIE|R 40 STEPHEN JELLERSO|N 18 JOCIO EUSEBIO D'|OLIVEIRA 41 STEPHEN MICKERSO|N 19 JOHN C. HUBBARD |JR. 42 STEPHEN NICKERSO|N 20 JOHN PETER LUNDB|ERG 43 THORNTON B.RENNE|LL 21 JOSIAH RICHARDSO|N 44 WASHINGTON PLUME|R 22 LEFTENANT MOTELE|V 45 WASHINGTON T.WAL|KER 23 LT.CMDR. B.J. LO|TTEN 46 WILLIAM RICHARDS|ON -- ----------------|-------- -- ----------------|-------- {References} NCDC (National Climatic Data Center), 1998: The Maury Collection: Global Ship Observations, 1792-1910 (CD-ROM, Version 1.0, February 1998). NCDC, Asheville, NC. Elms, J.D., 1999: Digitizing U.S. historical marine collections (1792-1976). Proceedings of the International Workshop on Digitization and Preparation of Historical Surface Marine Data and Metadata (Toledo, Spain, 15-17 September 1997). H.F. Diaz and S.D. Woodruff, Eds., WMO/TD-No.957, MMROA Report No. 43, World Meteorological Organization, Geneva, 7-13. Oliver, J., and J.A. Kington, 1970: The usefulness of ships' log-books in the synoptic analysis of past climates. Weather, 25, 520-528.