. T1956 = Jan. a ~ AN Ogee DOCUMENT | COLLECTION THE UNIVERSITY OF CHICAGO DEPARTMENT OF METEOROLOGY « aoe 8 Ans SEA SURFACE TEMPERATURES OF THE NORTH ATLANTIC 1887 =- 1956 Report on Research Prepared under Contract Neo N6ori-02036 | Project NR 082-120 Office of Neval Research January 1956 SEA SURFACE TEMPERATURES OF THE NORTH ATLANTIC 1887 = 1936 by Herbert Riehl The University of Chicago While the meteorological literature contains many allusions to rela- Sions between weather abnormalities and sea surface temperatures, titsie data actually substantiating such relations have been published. Moreover, the physical connection between changes oF ovean temperature and atmospheric flow patterns has not been demonstrated generally. In that respsst, seasonal hurri- cane frequencies offer a promising opportunity for study sincs these storms depend greatly for their maintenance on local heat transfer from ocean to atmospheree One may argue that if a positive anomaly eenter of sea surface temperatures is situated in the principal areas of hurricane formation during the season, hurricane frequency will tend to excesd averages To explore this hypothesis = copy of a collection of Atlantie sea sur- face temperatures for 1887 to 1956 was obtained. This collection was prepared by a project of the United States Weather Bureau under Wo Fo MacDonald during the 1930's. Though it represented an enormous effort, it is nevertheless fractional in that only certain areas of the Atlantis are covered and some of these not continuously. World War I, in particular, accounts for many un= welcome interruptions. According to the writer's understanding a large addi- tional file of sea surface temperatures is available at the National Weather Records Center at Renevetis North Carolina. Moreover, the analysis could be extended to the present. Such extension had been planned if the initial evaluation proved encouraging; this however proved net to be the caseec aQo The Weather Bureau data are arranged on a monthly basis for “squares” bounded by five degrees of latitude and longitude. In some months there were several hundred observations in some individual squares, in others only a few, sometimes less than ten, These uneven frequencies plus the unreliability of many of the measurements made careful processing necessaryo! At first, means for each month were computed for each square and these means averaged over the whole period. The resulting mean monthly values were then plotted on graphso These yielded smcoth seasonal variations in almost all squarese Occasional irregular values could be traced to isolated extremely low or high temperature reports, or months with very few reports. They were considered incorrect and replaced by values on the curve. Figo 1 shows the mean monthly temperatures o* Secular Variation Before proceeding to monthly temperature anomalies it was necessary to determine the secular trend because of the large climatic variation known to have occurred in the Atlantic in recent decades. This was done by computing annual temperatures in each square and expressing these as deviations from the mean annual temperature for the whole pericd. Mean annual temperatures were calculated from the data of fige lo The result is shown in figs 2 which also indicates the periods for which observations were available in each squareo One observes relatively strong year to year variations in many squares, some of these quasi-periodic. Very large changes occurred in the region of 1 Tt could not be ascertained to what extent, if any, changes in methods of measurement have affected the temperatures as recordede No allowance has been made for this factor é The Weather Bureau data extend beyond 50°N, but the polar part of the ocean was not studied in this projecto H =Ze strong temperature gradient near the American coast and during World War Is; the latter of course must be considered as unreliable. Superimposed on these fluctuations are unmistakable secular trends which, however, vary with latitude and longitudes In the tropics cooling predominates up to 1915-20, followed by warming especially after 1925. This pattern also occurs at many squares far-= ther north but other types of curves are alse found such as continuous warming > « we the secular levelling off toward the end of the resords Fige 5 summeri variation by latitude belts. The somewhat irregular character of the secular trends may be resolved partly by plotting charts of annual anomaly. As example, figse 4 and 5 show the anomalies and their analysis for 1936e A distinct and suggestive pattern is in evidences Along the western edge of the ccean positive anomalies are pronounced, with a little increase with lativudee Bus these anomalies de= crease regularly eastward. Slight negative values cecur in the easternumess squares and one suspests that a strong negative anomaly center is situated in Sade Lear ce IS GAGE Ue the southeastern part of the ccean. This anomaly gradient pilus the nm € southwest tilt of the whoie field suggests a connection with sirculation ano=- malies rather than a general radiative influences. Actually the circulation of the whole subtropical anticyclone (measured geostrophically) was above _ average in 19363 thus the departure was in the right sense to explain the oceanic temperature anomaly field. This matter, though very interesting, could not be Deiat aneer ener within the scope of the projecte The analysis of the secular variations can be refined by considering the contribution of individual months or seasons to the annual variations This refinement was not carried cut beyond noting qualitatively that trends were similar in all seasons. Further the secular trends, though relatively hls large when viewed as a whole, are quite small compared to changes in monthly anomaly patterns, say from one July to the next. Thus, as long as one com= pares only a few adjacent years, one may proceed without reference to the long term trend. For this reason straight monthly anomalies were computed from the temperatures of fige le It has been planned, however, to remove secular changes from any correlations with the atmosphere covering the whole period. It should be noted that analysis of the seeular trend by months or seagons could be utilized to normalize the data of fig 1. These are composed from records of varying length and periods, hence not fully comparabis. Nor= malization would produce some changes in the configuration of annual and menthly anomaly patternso Monuthiy Anomalies The eventual purpose of the program was to draw monthly anomaly charts and trace anomaly centers from month to month noting paths and changes in ine tensity. Two difficulties are to be overcome before this can be dones 1) Due to poor or sparse observations large and fictitious oseiliac tions of anomaly can occurs These must be located and eliminatedo 2) The strength of the anomalies may undergo seasonal variations; they may for instance decrease from winter to summer as in the atmosphere, If this is the case, the seasonal variation must be taken into account in following anomaly centers o The second problem was investigated by calculating the variability, or mean deviation, of the (uncorrected) monthly temperatures for each month in each squareo Fige 6 shows the result on graphs and fig. 7 on chartse Out= standing is the lack of important seasonal differences. Certainly ne organized decrease takes place from winter to summers some squares actually have largest 5a variability in summer. A few irregular values result from poor datao This outcome is favorable since it permits analysis without reference to seasono As the next. step, the monthly anomalies were plotted on time eco~ tions for all squares and curves drawne Poor values usually could be located easilye Most often a large positive or negative anomaly appeared in a single month in one square without history either in that or surrounding squares. Such values were disregarded. Except for such irregular values (and the years of World War I) little smoothing proved necessarye Hspecially in the later years when date was most plentiful in many squeres, anomaly variations becane very regular and consistent. The time sections could be used to fill in oeca- sional gaps of one or two months in a records From the curves adjusted monthly anomaly values were plotted on charts. These charts are reproduced as the main part of this reporte The anomalies are in tenths degrees Fahrenheit 9% The charts can be analyzed with some efforts Enough irregularity, how- ‘ ever, is left to suggest another smoothing. This was done by preparing a set ef charts with four-square overlapping meanse This new set could be analyzed with great easee It shows pronounced anomaly centers which sometimes persist for many months and usually, though not always, travel on a clockwise path through the ocean. Unfortunately, these charts could not be prepared for ail years because of the small number of squares available in many of the yearse Hence it was decided to reproduce the original set of charts which contains no analysis but all datae : Due to the large number of charts the data could not be drafted for reproduction. Apology is made for the occasional values which are difficult to reade Underlined numbers: less than ten observations. =Ge= Atmospheric Correlations According to the initial hypothesis the charts should furnish a partial predictor of the hurricane season if the analysis uncovers definite anomaly eenters which travel in relatively steady state over several months. Excepting the Gulf of Mexico the frequent existence of such centers has indeed been demon= strated but the hurricane correlations have failed to materialize. Of course, there were further handicaps. In particular, the anomaly analysis does not cover the whole region of hurricane formation and conditions far upstream are quite unknowne This difficulty was realized before the work was begun; but we thought that the available data should give a fair indications For the western part of the hurricane region temperatures upstream exist in many yearse Moreover one can compare sea temperatures and hurricane frequencies on a contemporary basise Many correlations were carried out, for the area as a whole and for smaller portions. Since these correlations proved unsuccessful they will not be discussed furthere It should be noted, however, that only a partial corre= lation was expectede Atmospheric circulation anomalies should furnish another indicator, even though circplation anomalies and sea temperatures may also be partly correlatede For the purpose of studying the relation between hurricanes end circulation in the subtropics and temperature westerlies, monthly sea-level pressure anomalies (1899-1939) were obtained through courtesy of the Extended Forecast Section of the United States Weather Bureaue These were analyzed and compared with ocean temperatures and hurricane frequencies. One notes quickly that there is some tendency for hurricane frequency to be low if in late spring the subtropical anticyelone has above average strength with east-west elonga= tione The lag correlation, however, is not good enough for forecasting pur- poses; a combined surface circulation = sea temperature index produces no im= provement. Conceivably extension of the work to the period following 1945 might do better since upper-air data can be used from then onward. When the time scale of correlation is increased to five years and more much better relations appears Only the secular trends are important in con= sidering these longer periods, The marked year-to-year changes drop out 2 a "turbulenceo” Indeed, when fig. 2 is viewed as a whole, it gives the appear= ance of a typical turbulence record with underlying trend. Only a few indications of leng period correlaticns based on nen overlapping five-year means can be given here. As already noted, the sea ten perature at first declined, later rose over a largs part of the western Atlan= tic including the subtropics (fig. 2). For comparison, rige & shews the pres=- sure anomaly at 30°N, 65°W. Clearly, the pressure ross as the sea temperature decreased, then the reverse took placee At the same tims, the geostrephic vorticity in the Bermuda area at first became more eyslonisce These date furnish strong ciues nding of secular trends o Even the hurricane frequencies become fairly manageable on the five- year tims scale. Figs 9 correlates hurricanes frequencies and sea temperatures in the region of formation for non-overlapping five-year periodse Though not wholly satisfactory, a parallel trend is nevertheless evidents One may raise the question whether figs. 8 and 9 demonstrate a general principle, namely that ocean temperature and atmosphere are correlated only when time steps of about five years or more are considered, and that other factors dominate shorter period fluctuations. =o= Turning to hurricane tracks, one observes that since the beginning of the century the location of the tracks has undergone a marked cycle. In the early years recurvatures in September took place mostly east of Florida (fig. 10). They then shifted westward to the Gulf between 1910 and 1920; later they returned at first to Florida and adjoining waters, finally to the West Atlantic. The shift in average recurvature longitude is no less than 20° near latitude 25°N and must be accepted as reale During the years when the tracks migrated westward ocean temperatures decreaged, and during the years when the tracks returned eastward, they ros@o Figo ll shows five-year average anomalies corresponding to parts of the periods with easternmost and westernmost track positions. Patterns are out-of=phases3 the period 1932-36 agress with fig. 5. Correlation bstween sea temporatures and hurricane tracks is excellent, but a direct connection appears difficult to establish. A change in resurvature longitude, as indicated by fige 10, should from all knowledge of hurricane motion be related to changes in the position of mean troughs and ridges in the westerlies. Direct evidence does not exist from the years studisd, but a glance at figo 8 will show that the track changes are correlated well with sea-level pressure near Bermuda. Tracks shifted westward with rising pressure and eastward with falling pres- suree If the pressure changes reflect east-west displacements of the Azores- Bermuda High, the track displacements can be interpreted as due to general circulation changeso Conclusion It- is the writer's belief that in spite of many difficulties of sea temperature observations and sea temperature records anomaly charts have been constructed which by and large are fairly accurate. Second approximations could be made to the whole series of charts beginning with fig. 1, but it is doubtful whether this would prove worth the effort The immediate aim for which the charts were prepared has not been attained. But it is evident thet they can be used for many studies of impor- tance in cceanography and long range forecasting. A few suggestions of inter= esting relations have been made, especially in figse 4 and 5 and figse 8 and 93 but further exploration of these topics is beyond the scope of this projecto For this reason the initial summary diagrams and all anomaly charts have been reproduced in the hepe that they will prove useful to other investigators. Acknowledgment The writer wishes to asknowledge the assistance of Mrso Ee Se Jordan who supervised preparation of the anomaly charts and did most of the analysose thes shi ha er 16,8 sit nani nih j ee nine ee TN ise food hay ayy wae eee ae or A PANE a ee Urs Rete, Balke Figure 1 Mean monthly sea surface temperatures (°F). (ES CNN aly TW We ke ee = hea ior Bane (| 54.9 61.7\639\' \ UGUST Figure 2 Departure of annual sea surface temperature from mean (°F) for all five-degree. squares (coordinates given for southeast corner }o { Ee 4 r 1 a I) I = is D | eae ry = Tr + ° q C| a = 4 —pb RE se: BE ee] PECK Saneyn . a Sees SABA +1 oe es SS PP OL PS EEL SS SLL EES v al te ' (asl me AN aah Plt CCC MIAN YT en iS EE SN A Hay WN | ALY Li Coane pe meaeNicee a a ik Ue a aneLevese A God ST ale Aa : eee ACT a A pauta eae SSOP 00, EG SW TT ew TTT TNA an. Cael ee eS a YL [VY SOM ols a “MRT ek Tee han SUS ; Figure $3 Course of annual sea surface temperature anomaly summed over latitude belts. In each belt all available five=degree squares have been usede 85 90 95 00 05 10 15 20 25 30 35 Figure 4 Sea surface temperature anomalies for 1936 (°F). Figure 9 Analysig of the data of fig. 4 (°F), 8 60 40 20 Reet. ate 5\0 wy \9 40 a, i — 30 a Pe - 20 Figure 6 Annual course of mean deviation of monthly sea surface temperatures (°F) for all five-degree squares (coordinates given for southeast corner } « Numbers on left indicate average annual value of monthly deviations. Scale given at bottomo | 0.9 0.8 0.6 0.7 0.7 Jen di Oe 1d 25-|90 25-185 25-|80 25-|75 25-|70 25-165 25-60 ORY Wa a A 8 a) ls ark owe! | 20-|70 be wore | 20-|60 20-/55 20-|50 WANES 20-|20 ] Meanie J ale fe fe /\ sii, RAR AS de el Dear A pele hes bi fuieke shirk qj LS qj O ( Vos oO i “J Gl w Lt de poe eae KE he \@ heh ° © 9° o mS aaa eS i | ; i : Nicer iv. a Te eh at k 4 iN) is estan i en ; { iene ae sia. Wiis HHT Hs Figure 7 Monthly charts of mean deviation of monthly sea surface temperatures (°F). hn aigND Re aaa A hs Mae ens [ats Geir th meas eet . Bi hee eee cn eae .| Po aglvales 3 eG Ae al ! =< Wi t Sy) ae 5) A A358 1p sjaqgara VS 4 Figure 38 Course of five-year means of July surface pressures anomaly at 30°N, 65°W (upper) and of surface geostrophie vorticity anomaiy for July in the 20°=square centered on 30°N, 65°W (lower). Figure 9 Course of five-year means of annual sea surface temperature anomaly for the belt 15-25°N, 65-85°W (upper) and of hurricans frequency in the whole Atlantic hurricane region (lower). Ne FS SOS: (QW) Ajowoun -d 5 20 25 30 35 time (years) 10 00 05 w ) s a one i ie 84094 G sad Ajowoub-1 15 20 25 30 35 10 90 95 00 05 oo Co3d 8 on ¢ mM AW sapak g sad Aouenbas auodjsany - time (years) Figure 10 September hurricane tracks with duration of three days and more, 1901-1944. (eo) a o m (e) 2 Figure 11 Sea surface temperature anomalies (OF) for September. Top: mean for 1911-1915; bottom: mean for 1932-19366 Pevensey eae WANT I Aa! nd ne cee Sree raya: livviacs Aigo Meee ne DEG thes a @ TALL T one Ht Ly Beets aes SS ALLE LHe] Ae 7 TE eer \: \ i EN Ghigo neoee ce "ann Adah La Wy ety ee naetiare Er Bees Mra aN aa x \ ; \_ gla “16 -u\ 38 , Ne 2K ‘ NOOO : a \ 4% \ [ «| “S20 | 6 bales hile LTTE LTE TTA LH 6 ‘0 2} 4| bo PEEL : Ee : S\folholely> < 5 | | erreretteeee TTT ») Halk seeks Noa \. ale \ \ ATA \e 40 a \\ \ \ A aN \ ft \ ea \ \ \ 20 \ 6) \ 4 \ rect? HK ; i ett \ l MDE WT 6 d gia \ hea tel apyer 0 \ g\ao\t \ -4\" \ \ o| va} -a| a} ¥\-Sl S\re, Te aan ee Hi SLE YI ne i If }- om tale eel -4 folds eee ara Woe hee, ' me de : Ath — Stel) 1 oN i ee SUS, el Y Ne eA 1S ed) eae aon 08 ee Lif iH pf He Ie i el Ow é | ALT LEAT Ineo iE Li] ct seesen™ is Bas , WKS Seba se mie! a eae PEER AAS iereanteaae nasal LELLTTEEETET o, An [-] \\ Me : eepgacpeeeee tcc a i BESrE aseton ww + tio Se ? Li Ty LTT] peeeeuen ¢ TIF ELL si ns } pees: HATH aaa ) AL) aooe oe acces as 1 oon C EC) eo © Le) st LH oe ALT THY HiagananeR rT chgt inh ZEEE HHH ED i BABECCEREEEECE ae TEX LOELTET EL Ses EEE eae ssh GRBBOOe eet Bad ans KS LETT: feng eeoenn Lhe agg ey ane, aoul pooenesenn a Boe com Pore pene cRaee aes : ee iit | hy tA ie [16] i saad eaanaee : reeeeeogor HEE eae [ tatctabael see Sera EERE SEESceeSeeC. Ls ae F Pee 0) 169 yeoman Sayer oe as PRED ESR eC Ferre ereecatontes Sopra en auee age AaB) ‘¢ (m9. PRE CETTE ise Ly LT munencee aca geeaananaDRel nee a [fafa iA ab seats Se, hone cH eile fis) peaeee SNS ie ts \ Sa Os : “" 2 -* srt at ie a # ee 1¢ 20 A ye Sah Sag) et? Si 3 2 Ass \ eee _-S ge ayes 40 i pe pen FADE 5 Vee a \ 6 Boe 35 3s as ao ZEEE AA ai PLE anal a CPE TTT [ LA 3 DERnRETET eT att FETE FEEELET TL bebe 5 EREREECeCe CEPT ERE 5 rape Ne auboorcce Ast “AON 70 e560 58S bes LALA ragareaaa ssn ie | ane BEE ae “ Hee se | PEELE | i SSB eae te : THEN ator HSS : ees Bases aS EREEEE ° a ; Bia : As oer enatitanagnauatitec z ‘el E-) a0 SN T0 seh 60 ee eas , =f Bete 7 RERREE : [=] ee noe ee a PERE C PEELE ee : EE Vg] Soon Bee eos SEE Be 80 3 70 35 0 5 30 45 20 AE SEEEBEEECE SaaeeCe ae ACEP EEEEP ECE! gc |e | | | PL a S705. 607 3580) LTH LETT Ee a ogee cal Bee) acon OF ee cee ee TELAT eS ng, Ribose TEER nT EES CEE tapas SET S GU 44 ee C) eassraeateesce “COPE EER Boge ABBERRDEDORY rss RRRBBEcobSo cone : Beem oy ATH Ee eae EE e| Y} ik t[ i] H ® ee) a\3 va\5 f HEE ap Bele acon. ss oe eer {] LT ALT Caeenan DPA ms g, LPP EGREENE ARNG LTT napaneee ne Ty por et | ‘ee BeBe ace ‘ HEHE f EERE suet ee HE ai Bune eu. PHT HE HANS Lip HH 28 \\; ate aust Cee ath atta PEL er eceeeee ! “Hae Gin fea Ty) | - | i Bacon. nooo tones ease 70.6560 85~«80 TR BEREEE ELAS Ly z a ry ee as HS late iis ee any s 20 a5 E) sS Ce) 7065 75 80 Mig g PEEREROBBBCO ecm PAA Pera aes z 30 Hi He | Diaeie [fear e| ¢ ae \ T tha ¥ fe | etl ie fs is i j Ly Mays ENGR EEOC ee eee SGnsggaendoees Ly LET Hebe | ALPE peboapeaneeeeegctees ra ul AGE reer erey s boa 10 feted me ° Epneeee a a re ? Her STALE] LPL] Ly rae FAT abe “ : Sete zane a eeegesae COT Lh PEPE aaa : meee iH Fess Hey eek: LL) es Le/ a a ALPES cea COE CEs ia oc OTS erg ek en 85 a Sct LL? ros } a PAE HE onc ; as 6 s Le chee f, eee. ‘e Ce aaa: reas aseee agaee ow + cig Ee a3 SEE 30 ee ely Pra oh aor Ty coe Gage if as LTTE PC ea § HLH Hee ee 3 i“ Gina eee: HH aoe eee LI] Coe noma feaaaiecrecceea oe | | Ge ay c) 49 70 eso 5 -80=C:t«—« SD ay SS abel ihe 5) GRO rtm esis comme: ae a0 wea Ae TPT: ETE ay assay SSO. . FCEEEEECS EAE EEG ELEGY, So nagesn £ fe ~ - / 7 4 ny ed ais =e : t-. RMN ie TS ote re ae Pi; Se ig / / ~ 1 lb lb wider 10 ( iy ny | = ae | ] ef lafel| N = ~ z 4 = f Ney allele -t ae cl I [ae a elie |2r| se a H ae is N LK if SS: \>\ al «| 2 oe a i [+] DERE as ? gErceeceaenaey Cig apame’ 22 6 IS : Ane KLE PRL BUBP Cte TP ALLTEL as ee s LTT HH A | - nee Se mies auaoee ; ZEAEL Tes Cog, CRUE DEES asCon CCS el a0 ey i