Constraining the Temperature History of the Past Millennium Using Early Instrumental Observations

Clim. Past, 8, 1551–1563, 2012 www.clim-past.net/8/1551/2012/ Climate doi:10.5194/cp-8-1551-2012 of the Past © Author(s) 2012. CC Attribution 3.0 License.

Constraining the temperature history of the past millennium using early instrumental observations

P. Brohan1, R. Allan1, E. Freeman2, D. Wheeler3, C. Wilkinson4,5, and F. Williamson3,4,5 1Met Ofﬁce Hadley Centre, Exeter, UK 2NOAA/STG Inc., USA 3Sunderland University, Sunderland, UK 4University of East Anglia, Norwich, UK 5Catholic University of Valparaiso, Chile

Correspondence to: P. Brohan (philip.brohan@metofﬁce.gov.uk)

Received: 4 April 2012 – Published in Clim. Past Discuss.: 4 May 2012 Revised: 17 August 2012 – Accepted: 6 September 2012 – Published: 11 October 2012

Abstract. The current assessment that twentieth-century tions – supporting their use for longer-term climate recon- global temperature change is unusual in the context of the last structions. However, some of the climate model simulations thousand years relies on estimates of temperature changes in the CMIP5 ensemble show much larger volcanic effects from natural proxies (tree-rings, ice-cores, etc.) and climate than this – such simulations are unlikely to be accurate in model simulations. Conﬁdence in such estimates is limited this respect. by difﬁculties in calibrating the proxies and systematic differences between proxy reconstructions and model simulations. As the difference between the estimates extends into 1 Introduction the relatively recent period of the early nineteenth century it is possible to compare them with a reliable instrumental es- The temperature history of the past millennium provides timate of the temperature change over that period, provided vital context for predictions of future change, and attri- that enough early thermometer observations, covering a wide butions of recent change to anthropogenic causes (Jones enough expanse of the world, can be collected. and Mann, 2004). Back to about 1850 large-scale tem- One organisation which systematically made observations perature changes are fairly well-known from thermometer and collected the results was the English East India Com- measurements (Brohan et al., 2006), but longer time scale pany (EEIC), and their archives have been preserved in the reconstructions are based on a variety of natural proxies British Library. Inspection of those archives revealed 900 (tree-rings, ice cores, speleothems, etc.) and have a large log-books of EEIC ships containing daily instrumental mea- uncertainty (Fig. 1). surements of temperature and pressure, and subjective esti- The proxy reconstructions not only disagree amongst mates of wind speed and direction, from voyages across the themselves, but they share a reliance on calibration to re- Atlantic and Indian Oceans between 1789 and 1834. Those cent instrumental (thermometer) records. Calibration is a sta- records have been extracted and digitised, providing 273 000 tistical estimate of the scaling factor relating a change in new weather records offering an unprecedentedly detailed the proxy value (ring width etc.) to a change in temper- view of the weather and climate of the late eighteenth and ature, and it is necessary to assume that this scaling fac- early nineteenth centuries. tor does not change with time or temperature. Recent years The new thermometer observations demonstrate that the have seen a lot of research into improved calibration tech- large-scale temperature response to the Tambora eruption niques, but despite these technical improvements, the uncer- ◦ and the 1809 eruption was modest (perhaps 0.5 C). This pro- tainty in the reconstructions remains large (Frank et al., 2010; vides an out-of-sample validation for the proxy reconstruc- Jones et al., 2009).

Published by Copernicus Publications on behalf of the European Geosciences Union. 2 BROHAN ET AL: EEIC EARLY OBSERVATIONS

under-represent the effects of large volcanic eruptions (Mann et al., 2012). Because the main difference between the reconstructions is in their calibration, the spread in centennial and longer timescale temperatures is highly correlated with the spread in shorter timescale variability: reconstructions that have a larger variability over (say) the first few decades of the nine- 1552teenth century also P. show Brohan larger et variability al.: EEIC over early the observations whole reconstruction. So if the reconstructions could be more tightly constrainedTable 1. Proxy over a seriesshort period, (top) and this modellingwould reduce groups the uncer- (bottom) 0.5 B2000 HCA..2006 MSH..2005 MBH1999 taintyproviding over data the whole used in millennium. Figs. 1 and 9. The early nineteenth cen- BOS..2001 ECS2002 O2005 JBB..1998 DWJ2006 MJ2003 RMO..2005 tury is a strong candidate for such a validation period, as it includesB2000 short-termBriffa temperature(2000); Briffa variability et al. ( of2004 uncertain) size 0 associatedBOS..2001 with largeBriffa volcanic et al. ( eruptions2001) (1809 and Tambora in 1815),DWJ2006 and thereD’Arrigo are some et thermometeral. (2006) observations for the period,HCA..2006 which couldHegerl potentially et al. (2006) be used to quantify the ECS2002 Esper et al. (2002); Cook et al. (2004) −0.5 temperature variability much more precisely. Precise infor- mationMJ2003 on temperatureMann changes and Jones over(2003 this) period would also MSH2005 Moberg et al. (2005) Temperature anomaly (C) Temperature resolveO2005 a major disagreementOerlemans between(2005) the GCM simulations, which vary greatly in the simulated magnitude of the 1809 −1 RMO..2005 Rutherford et al. (2005) andMBH1999 Tambora eruptions.Mann et al. (1999) AccurateJBB.1998 instrumentalJones etweather al. (1998 observations, 2001) have been re- coveredbcc-csm1-1 for limited regions Beijing Climate going back Center, well China before Meteorological 1800 (Ca- muffo and Bertolin,Administration 2011; Alcoforado (Wu et et al. al.,, 2012 2012),) so to do 0.5 bcc−csm1−1 MPI−ESM−P thisGISS-E2-R validation is merely NASA a matterGoddard of Institute recovering forSpace enough Studies such FGOALS−gl GISS−E2−R MIROC−ESM CCSM4 observations, covering(Schmidt a large et al. enough, 2006) area of the Earth, to constrainFGOALS-gl the large-scale LASG, temperature. Institute of Atmospheric Physics, Chi- 0 nese Academy of Sciences (Zhou et al., 2008) MIROC-ESM Japan Agency for Marine-Earth Science and Technology, Atmosphere and Ocean Research −0.5 2 New instrumentalInstitute weather (The recordsUniversity for of the Tokyo), early and nine- Na- teenth century tional Institute for Environmental Studies Temperature anomaly (C) Temperature (Watanabe et al., 2010) −1 AmongstMPI-ESM-P the archives Max in Planck the British Institute Library for Meteorology (BL) in London are some 4000 logbooks(Raddatz from et al.ships, 2007 in; theMarsland service et ofal., the2003 En-) CCSM4 National Center for Atmospheric Research 800 1000 1200 1400 1600 1800 2000 glish East India Company(Gent et (EEIC); al., 2011 each) recording the details Year and events of a voyage from England to the Indies (usually India, China or both) and back, typically taking the best part Fig. 1. Northern Hemisphere surface temperature estimates for the of two years. The EEIC received its charter from Elizabeth last millennium. Upper panel: recent proxy reconstructions (after Fig. 1. Northern Hemisphere surface temperature estimates for the I intime 1600, scale and temperatures many of its is earliest highly voyages correlated became with the famous spread lastfigure millennium. 6.10 of Jansen Upper panel: et al., recent2007). proxy Lower reconstructions panel: GCM simula-(after in shorter time scale variability: reconstructions that have a tions from CMIP5 (where possible (MIROC-ESM-P & GISS-E2-R) because of their excellent records of new lands; for example figure 6.10 of Jansen et al. (2007)). Lower panel: GCM simula- thatlarger by Henry variability Middleton over (say) to the the Moluccas first few decades in 1604-6 of (Foster, the nine- tionsthese from have CMIP5 been (wherecorrected possible for drift (MIROC-ESM-P by subtracting a& linear GISS-E2-R) trend fit- 2010).teenth These century early also voyages show larger were variability recorded over in diaries; the whole log- re- theseted to have their been unforced corrected parent for experiment). drift by subtracting In each a case linear the trend black fit- line construction. So if the reconstructions could be more tightly shows instrumental observations (Brohan et al., 2006). Data used is books — formally prepared documents of a standard format ted to their unforced parent experiment). In each case the black line constrained over a short period, this would reduce the uncer- listed in Table 1. — did not begin to appear until the 1650s. Their preparation shows instrumental observations (Brohan et al., 2006). Data used is tainty over the whole millennium. The early nineteenth cen- listed in table 1. was part of an officer’s duties until the gradual expansion of thetury Company is a strong in the candidate 1830s into for such a quasi-military a validation and period, politi- as it includes short-term temperature variability of uncertain size An alternative estimate of temperature changes is given cal body responsible for overseeing British interests in In- diaassociated and beyond. with Those large archivedvolcanic eruptionsin the BL, (1809 therefore, and Tambora extend by general circulation model (GCM) simulations, and the in 1815), and there are some thermometer observations for WCRP Coupled Model Intercomparison Project – Phase 5 from the 1600s through to the 1830s, and are well known to historiansthe period, (Farrington, which could 1999). potentially They document be used tosocial quantify condi- the (CMIP5 – Taylor et al., 2012) includes an ensemble of temperature variability much more precisely. Precise infor- state-of-the-art GCM simulations covering the period 850– tions, discipline, medicine and health, the trade and transport ofmation goods, on people temperature and passengers. changes over They this touch period on firstwould con- also 1850 (Fig. 1). Comparison of the GCM simulations with the resolve a major disagreement between the GCM simulations, proxy reconstructions shows systematic differences: the sim- tact with new lands and peoples, convey colonial attitudes andwhich cultures, vary greatlyand describe in the long simulated lost coastal magnitude towns of and the vil- 1809 ulations usually have little inter-decadal variability (which and Tambora eruptions. would support those proxy reconstructions showing least Accurate instrumental weather observations have been re- variance) but often show large responses to volcanic erup- covered for limited regions going back well before 1800 (Ca- tions (which are generally much less pronounced in the proxy muffo and Bertolin, 2011; Alcoforado et al., 2012), so to do reconstructions). This large difference has led to the sug- this validation, it is merely a matter of recovering enough gestion that tree-ring based proxy reconstructions systemat- such observations, covering a large enough area of the Earth, ically underrepresent the effects of large volcanic eruptions to constrain the large-scale temperature. (Mann et al., 2012). Because the main difference between the reconstructions is in their calibration, the spread in centennial and longer

Clim. Past, 8, 1551–1563, 2012 www.clim-past.net/8/1551/2012/ P. Brohan et al.: EEIC early observations 1553

2 New instrumental weather records for the early and variability over a large area of tropical and sub-tropical nineteenth century ocean for the late eighteenth and early nineteenth century – a time and region where other observations are almost com- Amongst the archives in the British Library (BL) in London pletely missing. About 10 % of these logbook observations are some 4000 logbooks from ships in the service of the En- have been examined in previous studies (Chenoweth, 1996, glish East India Company (EEIC); each recording the details 2000; Farrington et al., 1998), but most of them have never and events of a voyage from England to the Indies (usually been digitised or examined, and none have made it into the India, China or both) and back, typically taking the best part standard climate datasets for widespread use. of two years. The EEIC received its charter from Elizabeth I in 1600, and many of its earliest voyages became famous because of their excellent records of new lands; for example 3 Digitisation of the weather records that by Henry Middleton to the Moluccas in 1604-6 (Foster, 2010). These early voyages were recorded in diaries; log- The BL’s EEIC logbook collection has been catalogued (Far- books – formally prepared documents of a standard format rington, 1999), but that catalogue, though extensive, does not – did not begin to appear until the 1650s. Their preparation distinguish those logbooks that contain instrumental data. So was part of an officer’s duties until the gradual expansion of research was undertaken in the BL archives to produce a cat- the Company in the 1830s into a quasi-military and political alogue detailing exactly which logs contain instrumental ob- body responsible for overseeing British interests in India and servations; whether the observations are of pressure, air or beyond. Those archived in the BL, therefore, extend from the sea surface temperature; the name of the ship; the year of 1600s through to the 1830s, and are well-known to historians its voyage; and the ship’s route with dates. It also includes (Farrington, 1999). They document social conditions, disci- additional information such as how frequently the readings pline, medicine and health, the trade and transport of goods, were taken, and whether any unusual weather events took people and passengers. They touch on first contact with new place. Many of the logbooks dating from 1790 or later con- lands and peoples, convey colonial attitudes and cultures, and tained some instrumental observations, but not every log con- describe long lost coastal towns and villages. Many even contained observations, and on occasion observational records tain detailed drawings of coastlines, ships, mammals, birds were sporadic. and sea creatures. Using this catalogue, the 891 logbooks including con- Ship’s logbooks are also valuable sources of historical sistent instrumental records were selected for digitisation: climate data (Chenoweth, 1996; Wheeler et al., 2006; Bro- the earliest that of the Melville Castle, starting in Febru- han et al., 2009, 2010), and the EEIC logbooks include ary 1789; and the last that of the Sherborne, finishing in daily records of the weather along the routes taken by the August 1834. Figure 2 shows a typical example, logbook ships: they cover large parts of the Atlantic and Indian records for one day from EIC ship Carmarthen. Oceans, and include the occasional foray into the Pacific. All That logbook records a voyage from London to Bombay the logbooks contain wind speed and direction records, as and back to the UK, through the Atlantic and Indian Oceans, this was vital information for early navigators, but the later and round the Cape of Good Hope. The voyage took 20 logs, starting in about 1790, are even more valuable, as some months (May 1810 to January 1812); records were only made of them contain daily thermometer and barometer observa- on days when the ship was at sea, but even so the logbook in- tions as well as the wind reports. The principal instigator of cludes 372 such daily records. the addition of instrumental observations was Alexander Dal- Digitising each day’s observations from all 891 logbooks rymple – the Company’s, and later the Royal Navy’s, first proved to be a major undertaking. To make it possible to hydrographer. Dalrymple was both an explorer and an enthu- work on the logbooks outside the BL, the books were pho- siastic scientist, and, as hydrographer, he was responsible for tographed. This produced about 140 000 digital images, each ensuring that the EEIC ships could transport goods to and showing one page. Most pages contained two day’s records from England as quickly as possible and at minimum risk on a standard pre-printed form (Fig. 2 shows the top half of of loss. With this in mind, he equipped the East Indiaman one page) though, in rare cases, variant form types were used Grenville with a set of meteorological instruments for her that only contained one day’s records; also hand-drawn forms voyage in 1775 under Captain Burnet Abercrombie (Dalrym- were occasionally used – presumably to make up a short- ple, 1778), and set a pattern to be later adopted by officers on fall in printed forms. These images were indexed and stored all EEIC ships. in the electronic media archive of the US Climate Database It is the routine inclusion of regular instrumental weather Modernisation Program (CDMP), managed from the Na- observations that distinguishes the EEIC logbooks from their tional Oceanic and Atmospheric administration’s (NOAA) contemporaries, such as the Royal Navy and the Hudson’s National Climatic Data Center (NCDC). Bay Company (which did not routinely make such records CDMP also managed the transcription of the weather ob- until many years later). The EEIC logbook records offer a servations from the images. The data to be transcribed were potential source of detailed information on climate change selected – these are the highlighted sections in Fig. 2 – and www.clim-past.net/8/1551/2012/ Clim. Past, 8, 1551–1563, 2012 41554 BROHAN ET AL: EEIC EARLY OBSERVATIONS P. Brohan et al.: EEIC early observations

format is not entirely regular, so judgement was often required in identifying the elements to be transcribed. Values were sometimes recorded in unconventional methods (e.g. complex fractions or the use of dashes (–) to represent the number zero), and sometimes positioned on the wrong part of the form. Every logbook was pre-screened to notify the keying operators of any strange and unusual recording methods or deviations from the most common recording practises, and the keying operators were vigilant in identifying irregu- larities. Unusual entries often required a full review of the logbook to determine how the observer was recording the questionable element and if they were consistent throughout with their recordings. Once the logbook was thoroughly in- spected, an educated decision was made on how to transcribe the values to the common format outlined in the keying instructions. Once a logbook had been keyed in its entirety, it was then quality controlled by CDMP and distributed for further format conversions and analysis. In all 272 852 daily records were transcribed. To be useful to the community of climate and other researchers who use historical marine observations, each record must be converted into the International Maritime Me- teorological Archive format (Woodruff, 2007). In most re- spects such conversion is straightforward – conversion of latitude from degrees-minutes-seconds to decimal degrees,

Fig. 2. Logbook of EIC ship Carmarthen for 24th September 1810. Ship’s days run from noon to noon 12 hours ahead of the civil day, soand of temperatures from Fahrenheit into Celsius. Conver- this coversFig. the afternoon 2. Logbook of the 23rd and theof morning EIC of ship the 24th.Carmarthen For each hour there isfor space 24 to enter September the ship’s course, 1810. its speed (in Knots and Fathoms), and the wind direction. To the right of this table is a section for general remarks (which almost always includes referencesion of pressure measurements is slightly more complicated, to the windShip’s speed); and days at the bottom run is from a table of noon summary to data noon for the day. 12 The h elements ahead digitised of the are highlighted civil day, in yellow. so from top to ◦as not only must the measurements be converted from inches bottom theythis are: the covers date (September the 24th); afternoon the wind force of (a thelight breeze 23rd — Beaufort and the force 2), morning the wind direction of the (West by24th. South - 258.75 magnetic), the noon position (9 degrees 27 minutes North, 64 degrees 28 minutes East), the barometric pressure (30.05 inches of mercury)of mercury to hectopascals, but corrections may be ap- and the airFor temperature each (82hour degrees 45 there minutes is — 82 space.75◦ Fahrenheit). to enter the ship’s course, its speed (in plied for systematic biases in the method of measurement Knots and Fathoms), and the wind direction. To the right of this table is a section for general remarks (which almost always includes (Sect. 3.2.2). Conversion of wind direction from 32 or 64- reference to the wind speed); and at the bottom is a table of sum- point compass directions to degrees east is also straightfor- −1 mary data for the day. The elements digitised are highlighted in yel- ward, but the wind speed must be converted to ms from low. From top to bottom they are as follows: the date (24 Septem- verbal descriptions such as light gale or moderate monsoon ber); the wind force (a light breeze – Beaufort force 2); the wind (Sect. 3.2.3). Temperature, pressure, and wind records are direction (West by South – 258.75◦ magnetic); the noon position (9 further discussed below. degrees 27 min North, 64 degrees 28 min East); the barometric pres- As well as the units conversion and adjustment, the oppor- sure (30.05 inches of mercury); and the air temperature (82 degrees ◦ tunity was taken to apply some basic quality control to the 45 min – 82.75 Fahrenheit). ship positions. In many cases the hemisphere ﬂags (E/W or N/S) attached to position observations were missing or ob- viously wrong, occasionally obvious errors would appear in staff were trained to read and key the speciﬁed elements. A latitudes and longitudes as well. All these problems can be detailed set of instructions was prepared for the keying oper- seen plainly in a plot of the course of the ship and such er- ators to ensure that the data was transcribed into a uniform roneous values, when found, were corrected if the correction format. Budget constraints limited transcription to the noon was obvious, and set to missing otherwise. observations of air temperature, barometric pressure, and lo- The IMMA format allows attachments, and the original cation; the wind direction and force closest to noon, and all version of each record is attached to each IMMA record, details of the state of the weather and sea. Each element was so that the un-converted and un-corrected data can be re- double keyed to a give a minimum transcription accuracy covered if necessary. All the IMMA records are provided of 99 %. as Supplement. For the most part the logbooks recorded elements to a basic standard that can easily be understood today. However, 3.1 Ship routes and observational coverage the age of the documents made the transcription unusually challenging: the handwriting is not easy to read and con- A sailing ship travelling between England and the Indies, be- tains frequent and variable abbreviations, and the document fore the opening of the Suez canal in 1869, had to follow

Clim. Past, 8, 1551–1563, 2012 www.clim-past.net/8/1551/2012/ 6 BROHAN ET AL: EEIC EARLY OBSERVATIONS

Coutts voyage of 1817-1819), and multiple thermometers are representing 82.75◦F , as seen in figure 2), similar to the typ- also occasionaly seen (e.g. Gilbert and Blunt manufactures ical format for latitude or longitude. As the observations pre- on board the Neptune voyage of 1814-15). It is likely that a date the development of the modern Stevenson-type screen, diverse range of instruments was used across the EEIC fleet. the major difficulty in comparing them to modern observa- The vast majority of the temperature and pressure obser- tions will be their exposure — the details of how the ther- vations were made at noon (a handfull of logbooks record mometer was screened from solar radiation. It is likely that morning and afternoon observations on the same day); the the thermometers were less well screened than the modern location of the instruments is not known for certain, but it is standard, and also contaminated by ship heating (Chenoweth, likely that the thermometer and barometer were kept together 2000; Berry and Kent, 2005). They will therefore be biased in the captain’s quarters and adjacent gallery at the stern of warm, and the bias will be larger in regions where the surface the vessel — Dalrymple’s report includes the following “this solar radiation is large. Figure 5 shows the distribution of the thermometer belonged to Mr. Russell, and hung in the open temperature anomalies (difference from recent values). The air in the balcony” (Dalrymple, 1778).

BROHAN ET AL: EEIC EARLY OBSERVATIONS 5 3.2.1 Temperature

4 questionable element and if they were consistent throughout ular throughout the period: the Mozambique Channel could The temperatures are recorded in degrees Fahrenheit: usually to a precision of 1 degree but sometimes to a quarter or a 60 with their recordings. Once the logbook was thoroughly in- only be used if arriving in boreal summer (when the south- tenth of a degree. In some instances thermometer values were 2 30 spected, an educated decision was made on how to transcribe west monsoon blows in the northern Indian Ocean) the alter- recorded in the format of degrees and seconds (e.g. 82◦450, the values to the common format outlined in the keying in- native route was used throughout the year. Figure 3 shows 0 0 structions. Once a logbook had been keyed in its entirety, examples of both routes. The route back was simpler - a P. Brohan et al.: EEIC early observations 1555 Latitude it was then quality controlled by CDMP and distributed for −30 −2 further format conversions and analysis. In all 272852 daily −60 records were transcribed. 1600 ● ● −4 ●● ●●●● ● ● ● ● ● ● To be useful to the community of climate and other ● ● ● ●●● ● ● 60 ● 1400 ● ● ● ● ● ●● researchers who use historical marine observations, each ● ● −45 0 45 90 135 ● ● ● ● 1200 ● ● ● ● 30 ● ●●● record must be converted into the International Maritime Me- ● ● Longitude ● ●● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● 1000 ● ● ● ● ●● ● ● ● ● ● teorological Archive format (Woodruff, 2007). In most re- ●● ●● ●●● ● ● ● ● ● ● ● ● ● ●● 1 ● ● ● ●● ● ● ● ● 0 ● ●● ● ●●● ● ●● ● ● ●● ● 800 spects such conversion is straightforward - conversion of lat- ● ● ● ● ● ● ● ● Latitude ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ●● ● 600 itude from degrees-minutes-seconds to decimal degrees, and ● ● ● −30 ● ● ●● ● 0.5 ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ●● ●● ● ● of temperatures from Fahrenheit into Celsius. Conversion ●● ● ● 400 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●●● ●●● ● ● ● ● ●● ● ● ● ● ●● ●●●● ● ● ● ● −60 ● ● ● ● ● ● ● of pressure measurements is slightly more complicated, as 0 ● ● 200 ● ● ● not only must the measurements be converted from inches ● ● ●

AT anomaly (C) AT ● of mercury to hectopascals, but corrections may be applied −45 0 45 90 135 180 −0.5 for systematic biases in the method of measurement (section Fig. 3. Daily positions on the outward (red) and return (blue) voy- Longitude ● 3.2.2). Conversion of wind direction from 32 or 64-point ages of the Astel in 1812–1813 (squares) and Thomas Grenville in −1 Fig. 3. Daily positions on the outward (red) and return (blue) voy- ● 1827–1828 (circles). ● compass directions to degrees east is also straightforward, 60 ● 1790 1800 1810 1820 1830 ● ages of the Astel in 1812-13 (squares) and Thomas Grenville in ● ● ● ● ● −1 ● ● ● but the wind speed must be converted to ms from ver- 50 ● ● ● Year ● ● ● 1827-28 (circles). ● ● ● ● ● ● ● ● bal descriptions such as ‘light gale’ or ‘moderate monsoon’ ● ● 40 ● ● ● ● ● ● a route constrained by the global wind ﬁelds: the prevail- ● ● ● ● ◦ (section 3.2.3). Temperature, pressure, and wind records are ● ● Fig. 5. Air temperature (AT) anomalies ( C): observations minus ing winds close to the Equator are the easterly trades, so 30 an air-temperature climatology for 1961–90 (Rayner et al., 2003). further discussed below. ● directsailing route to round the East the from Cape England using meant the easterly travelling trades, southwest north- 20

Number of ships Upper panel: truncated mean AT anomaly in each 2x2 degree As well as the units conversion and adjustment, the oppor- westthrough into the the mid-Atlantic Atlantic down and to then theback latitudes to England of the Southern with the ● 10 ● square. Lower panel: truncated mean AT anomaly in each year. ● tunity was taken to apply some basic quality control to the NorthernHemisphere Hemisphere westerlies, westerlies. and using those winds to make the ● ● Note that the observed temperatures are not bias-adjusted. ship positions. In many cases the hemisphere flags (E/W or Figurenecessary 4 shows easting. the Once coverage in the Indianof the Oceanobservations the ships from could all ●0● ● ● ● N/S) attached to position observations were missing or ob- 891either logs. sail The up observations through the are Mozambique strongly concentrated Channel (between along 1790 1800 1810 1820 1830 Madagascar and the continent of Africa) and use the south- viously wrong, occasionally obvious errors would appear in the standard routes, but with enough variation to explore a Year temperatures are consistently warmer than their recent equiv- latitudes and longitudes as well. All these problems can be west monsoon to carry them over to India, or make all their alents, but the difference is much more likely to be a result large area of the Atlantic and Indian Oceans - occasional◦ Fig. 4. Geographical coverage of the observations. Upper panel: to- Easting in the strong westerly winds around 35–40 south of exposure bias than an indication that surface temperatures seen plainly in a plot of the course of the ship and such er- ships do take radically different routes - visiting the Red sea Fig.tal 4. numberGeographical of observations coverage of in the each observations. 2 × 2 degree Upper square. panel: Lower to- and then sail directly north to their destination. Both choices were warmer in 1789-1834 than in 1961-90. The mean tem- roneous values, when found, were corrected if the correction and Persian Gulf, or looping through the South Pacific on talpanel: number number of observations of ships providing in each 2x2 observations degree square. in each Lower year. panel: remained popular throughout the period: the Mozambique number of ships providing observations in each year. perature changes little over the period of the observations, was obvious, and set to missing otherwise. the way to China. The records are fairly evenly distributed Channel could only be used if arriving in boreal summer with modest falls in 1809 and 1816 — almost certainly a The IMMA format allows attachments, and the original through(when time, the southwest with at least monsoon 30 ships blows contributing in the northern in every Indian year and make of instruments used on board the EEIC vessels is version of each record is attached to each IMMA record, so betweenOcean) 1794 the and alternative 1833. route was used throughout the year. rarely recorded within the logbooks, and in most cases no that the un-converted and un-corrected data can be recovered Figure 3 shows examples of both routes. record has been found indicating the manufacturer or type if necessary. All the IMMA records are provided as supple- 3.2 Temperature,The route back pressure was simpler and – wind a direct route round the Cape of instrument used. Dalrymple’s 1775 voyages used barom- using the easterly trades, north-west into the mid-Atlantic mentary information. eters and thermometers of Nairne and Blunt manufacture and then back to England with the Northern Hemisphere The logbooks contain instrumental observations of air pres- (Dalrymple, 1778) but it is known from some of the more westerlies. 3.1 Ship routes and observational coverage sure and temperature, and qualitative descriptions of wind assiduously maintained logbooks that barometers of differ- Figure 4 shows the coverage of the observations from all speed. The details of how the measurements were made are ent manufacture were also used, such as Dolland, Barraud, 891 logs. The observations are strongly concentrated along A sailing ship travelling between England and the Indies, be- not known, and we should expect some biases even in the in- Troughton, and Gilbert. On occasion more than one barome- the standard routes, but with enough variation to explore a fore the opening of the Suez canal in 1869, had to follow a strumental observations, caused by limitations in the instru- ter was in use (e.g. Dolland, Barraud and Troughton on board large area of the Atlantic and Indian Oceans – occasionally route constrained by the global wind fields: the prevailing ments used and the observational protocols. The model and the Thomas Coutts voyage of 1817–1819), and multiple ther- ships do take radically different routes – visiting the Red sea mometers are also occasionaly seen (e.g. Gilbert and Blunt winds close to the Equator are the easterly trades, so sailing makeand of Persian instruments Gulf, used or looping on board through the EEIC the South vessels Pacific is rarely on manufactures on board the Neptune voyage of 1814–1815). to the East from England meant travelling southwest through recordedthe way within to China. the logbooks, The records and are in fairly most evenly cases distributed no record It is likely that a diverse range of instruments was used across the Atlantic down to the latitudes of the southern hemisphere hasthrough been found time, indicating with at least the 30 manufacturer ships contributing or type in every of instru- year the EEIC fleet. westerlies, and using those winds to make the necessary east- mentbetween used. 1794Dalrymple’s and 1833. 1775 voyages used barometers and The vast majority of the temperature and pressure obser- ing. Once in the Indian Ocean the ships could either sail thermometers of Nairne and Blunt manufacture (Dalrymple, 3.2 Temperature, pressure and wind vations were made at noon (a handful of logbooks record up through the Mozambique Channel (between Madagascar 1778) but it is known from some of the more assiduously morning and afternoon observations on the same day); the and the continent of Africa) and use the Southwest monsoon maintained logbooks that barometers of different manufac- The logbooks contain instrumental observations of air pres- location of the instruments is not known for certain, but it is to carry them over to India, or make all their Easting in the ture were also used, such as Dolland, Barraud, Troughton, likely that the thermometer and barometer were kept together ◦ sure and temperature, and qualitative descriptions of wind strong westerly winds around 35–40 South and then sail di- andspeed. Gilbert. The On details occasion of how more the than measurements one barometer were was made in are use in the captain’s quarters and adjacent gallery at the stern of rectly north to their destination. Both choices remained pop- (e.g.not Dolland, known, Barraud and we should and Troughton expect some on biasesboard eventhe Thomas in the the vessel – Dalrymple’s report includes the following “this instrumental observations, caused by limitations in the in- thermometer belonged to Mr. Russell, and hung in the open struments used and the observational protocols. The model air in the balcony” (Dalrymple, 1778).

www.clim-past.net/8/1551/2012/ Clim. Past, 8, 1551–1563, 2012 6 BROHAN ET AL: EEIC EARLY OBSERVATIONS BROHAN ET AL: EEIC EARLY OBSERVATIONS 7

Coutts voyage of 1817-1819), and multiple thermometers are representing 82.75◦F , as seen in figure 2), similar to the typ- consequences of the two large tropical volcanic eruptions in to be an artifact of the movement of the ship as it appears also occasionaly seen (e.g. Gilbert and Blunt manufactures ical format for latitude or longitude. As the observations pre- the period. equally in stormy and calm regions. on board the Neptune voyage of 1814-15). It is likely that a date the development of the modern Stevenson-type screen, In 1818, Alexander Adie patented the sympiesometer, a 3.2.2 Pressure diverse range of instruments was used across the EEIC fleet. the major difficulty in comparing them to modern observa- mercury-less marine barometer containing coloured almond oil and hydrogen gas (Middleton, 1964). Several of the The vast majority of the temperature and pressure obser- tions will be their exposure — the details of how the ther- The pressures were recorded in inches of mercury, usually to later EEIC voyages carried a sympiesometer, either in tan- vations were made at noon (a handfull of logbooks record mometer was screened from solar radiation. It is likely that a decimal precision of 1/100 of an inch, but occasionally as dem with the mercury barometer or as a standalone pressure morning and afternoon observations on the same day); the the thermometers were less well screened than the modern a fraction (e.g. 29 7 ”). The pressures have been corrected 8 gauge: 31 of the 893 digitized logbooks have records from location of the instruments is not known for certain, but it is standard, and also contaminated by ship heating (Chenoweth, for gravity (using the observed ship latitude) and for temper- sympiesometers. Of those 31 logs, 30 also contained records likely that the thermometer and barometer were kept together 2000; Berry and Kent, 2005). They will therefore be biased ature (using the associated air temperature where available - from a mercury barometer, and in some cases simultanious in the captain’s quarters and adjacent gallery at the stern of warm, and the bias will be larger in regions where the surface no barometer attached temperatures were recorded). Figure observations from both instruments were recorded. the vessel — Dalrymple’s report includes the following “this solar radiation is large. Figure 5 shows the distribution of the 6 shows the distribution of the pressure anomalies (differ- There are large and systematic differences between the thermometer belonged to Mr. Russell, and hung in the open temperature anomalies (difference from recent values). The ence from recent values). The pressures are systematically sympiesometer and barometer measurements. The symp- air in the balcony” (Dalrymple, 1778). 1556 P. Brohan et al.: EEIC early observations iesometer was designed to be portable and to respond rapidly to pressure changes, rather than for accuracy and stability; 3.2.1 Temperature and time-series of sympiesometer measurements (not shown) 20 often show large drifts in pressure readings over a voyage. 4 The temperatures are recorded in degrees Fahrenheit: usu- 15 So the indications are that sympiesometer records will need ally to a precision of 1 degree but sometimes to a quarter or a 60 60 10 close attention to calibration and correction in order to be tenth of a degree. In some instances thermometer values were 2 30 30 useful for historical reconstructions. 82◦450 5 recorded in the format of degrees and seconds (e.g. , The pressure observations included in the attached IMMA 0 0 0 0 records are all believed to be from mercury barometers, but Latitude Latitude −5 it is possible that in some of the later voyages the pressure −30 −30 −2 observations are actually from a sympiesometer. That is, a −10 sympiesometer has been used in place of the usual mercury −60 −60 1600 −15 barometer but the substitution is not mentioned in the log- −4 book. 60 1400 −20 −45 0 45 90 135 −45 0 45 90 135 180 1200 30 Longitude Longitude 3.2.3 Wind speed 1000 1 5 0 The wind speed observations in the logbooks are, as is usual 800 Latitude ● at sea, subjective assessments based on the sails carried and 600 0 −30 0.5 ● ● ● the state of the sea. The vocabulary of such assessments was ● ● ● ● ● ● ● ● ● ● 400 ● ● ● ● ● ● ● ● ● ● not formally standardised until the 1830s, when Sir Francis ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −60 ● ● ● ● ● ● ● ● ● ● −5 ● ● ● ● 0 ● ● ● ● ● ● ● Beaufort succeeded in introducing an official scale, but, even 200 ● ● ● ● ● ● ● ● ● ● ● ● ● ● in this pre Beaufort-scale age, sailors were very consistent ● ● ● AT anomaly (C) AT ● −10 ● −45 0 45 90 135 180 −0.5 in their description of the winds, and it is possible to make ● ● ● ● Pressure anomaly (hPa) ● Longitude ● quantitative estimates of the wind speed from the language in −1 −15 the logs (CLIWOC, 2003). Figure 7 shows the most frequent ● wind-force terms used and their Beaufort equivalents where ● 60 ● 1790 1800 1810 1820 1830 1790 1800 1810 1820 1830 ● ● ● available. ● ● ● ● ● ● Year 50 ● ● ● Year ● ● ● ● ● ● About 80% of the terms encountered can be converted to ● ● ● ● ● ● ● 40 ● ● ● ● ◦ ● ● Fig. 6. Air pressure (AP) anomalies from mercury barometers Beaufort forces, and so to 10-metre wind speeds in m/s. ● ● ● Fig. 5. Air temperature (AT) anomalies ( C): observations minus an ● ◦ ● ● Fig. 5. Air temperature (AT) anomalies ( C): observations minus Fig. 6. Air Pressure (AP) anomalies from mercury barometers 30 air-temperature climatology for 1961–1990 (Rayner et al., 2003). (hPa): observations minus a sea-level pressure climatology for Figure 8 shows the distribution of the inferred wind-speed an air-temperature climatology for 1961–90 (Rayner et al., 2003). (hPa): observations minus a sea-level pressure climatology for ● Upper panel: truncated mean AT anomaly in each 2 × 2 degree 1961–1990 (Allan and Ansell, 2006). Upper panel: mean AP anomalies (difference from recent values). The notable fea- 20 1961–90 (Allan and Ansell, 2006). Upper panel: mean AP anomaly

Number of ships Upper panel: truncated mean AT anomaly in each 2x2 degree × ● square. Lower panel: truncated mean AT anomaly in each year. Note anomaly in each 2 2 degree square. Lower panel: mean AP in each 2x2 degree square. Lower panel: mean AP anomaly in each ture of figure 8 is the large anomalies in the trade-wind re- 10 ● square. Lower panel: truncated mean AT anomaly in each year. anomaly in each year. ● that the observed temperatures are not bias-adjusted. gions. It’s possible that the trade winds were stronger around ● year. ● Note that the observed temperatures are not bias-adjusted. ●0● ● ● ● 1800, but as no equivalent anomaly appears in the pressure fields it’s more likely that the CLIWOC dictionary is 1790 1800 1810 1820 1830 3.2.1 Temperature andThe consistently mean temperature about 5hPa changes (0.15 little inches) over below the periodtheir recent of the slightly mis-calibrated in those regions (stronger trades im- Year temperatures are consistently warmer than their recent equiv- equivalents.observations, This with bias modest has been falls observed in 1809 before and 1816 in pre-1855 – almost ply a stronger sub-tropical high or a deeper equatorial low). alents, but the difference is much more likely to be a result The temperatures are recorded in degrees Fahrenheit: usu- marinecertainly observations a consequences (Ansell of et theal., 2006;two large Brohan tropical et al., volcanic 2010) Although all the pressure and temperature observations of exposure bias than an indication that surface temperatures Fig. 4. Geographical coverage of the observations. Upper panel: to- ally to a precision of 1 degree but sometimes to a quarter or a anderuptions the cause in the is still period. unknown. It is unlikely to be an effect in the selected logbooks were digitised, budget constraints were warmer in 1789-1834 than in 1961-90. The mean tem- tal number of observations in each 2x2 degree square. Lower panel: tenth of a degree. In some instances thermometer values were of gravity or temperature correction because it doesn’t vary meant that not all of the much more numerous and vari- number of ships providing observations in each year. perature changes little over the period of the observations,◦ 0 recorded in the format of degrees and seconds (e.g. 82 45 , with3.2.2 temperature Pressure or latitude, and it seems equally unlikely ous wind observations were. There are also wind observa- withrepresenting modest falls 82.75 in◦F, 1809 as seen and in 1816 Fig. —2), similaralmost to certainly the typical a format for latitude or longitude. As the observations predate The pressures were recorded in inches of mercury, usually to the development of the modern Stevenson-type screen, the a decimal precision of 1/100 of an inch, but occasionally as major difficulty in comparing them to modern observations a fraction (e.g. 29 7/800). The pressures have been corrected will be their exposure – the details of how the thermometer for gravity (using the observed ship latitude) and for temper- was screened from solar radiation. It is likely that the ther- ature (using the associated air temperature where available – mometers were less well screened than the modern standard, no barometer attached temperatures were recorded). Figure 6 and also contaminated by ship heating (Chenoweth, 2000; shows the distribution of the pressure anomalies (difference Berry and Kent, 2005). They will therefore be biased warm, from recent values). and the bias will be larger in regions where the surface so- The pressures are systematically and consistently about lar radiation is large. Figure 5 shows the distribution of the 5 hPa (0.15 inches) below their recent equivalents. This bias temperature anomalies (difference from recent values). has been observed before in pre-1855 marine observations The temperatures are consistently warmer than their re- (Ansell et al., 2006; Brohan et al., 2010) and the cause is still cent equivalents, but the difference is much more likely to unknown. It is unlikely to be an effect of gravity or temper- be a result of exposure bias than an indication that surface ature correction because it does not vary with temperature temperatures were warmer in 1789–1834 than in 1961–1990. or latitude, and it seems equally unlikely to be an artifact of

Clim. Past, 8, 1551–1563, 2012 www.clim-past.net/8/1551/2012/ 8 BROHAN ET AL: EEIC EARLY OBSERVATIONS

8tions in the more than 3000 logbooks in the BL archive that and ship routes BROHAN were constantET AL: EEICover the EARLY period OBSERVATIONS covered by were not examined in this study (because they had no in- the EEIC observations so they can be used directly to look tionsstrumental in the more pressure than or 3000 temperature logbooks observations). in the BL archive So much that atand temperature ship routes variations were over constant the period over 1795–1833 the period (when covered by weremore not information examined onin windthis study fields is (because still potentially they had available no in- therethe were EEIC enough observations ships contributing so they can to give be reliable used directly results). to look strumentalin the BL pressure EEIC logbook or temperature archive. observations). If extracted in Soa future much Extractingat temperature a set of pseudo-observations variations over the from period each 1795–1833 GCM run, (when project, these records would provide information on sub- by sampling from the model output fields at the date and lo- more information on wind fields is still potentiallyth available there were enough ships contributing to give reliable results). in thedaily BL weather EEIC variability logbook back archive. into the If17 extractedcentury. in a future cationExtracting of each a observationset of pseudo-observations allows a direct comparison from each GCMbe- run, tween observations and simulations (figure 9: upper panel). project, these records would provide information on sub- by sampling from the model output fields at the date and lo- It’s clear that the observational coverage is sufficient to show daily weather variability back into the 17th century. cation of each observation allows a direct comparison be- 4 Constraining proxy reconstructions and GCM simu- the effect of the volcanoes in the simulations, and it’s also tween observations and simulations (figure 9: upper panel). lations clear that the observations support those simulations with a smallIt’s temperatureclear that the response observational to the Tambora coverage and is sufficient 1809 erup- to show 4 ConstrainingThe biases in the proxy observed reconstructions air temperatures and mean GCM that simu- it’s tions.the Mucheffect of of the the variation volcanoes between in the simulations simulations, is a reflec- and it’s also lationsdifficult to compare temperatures in the early nineteenth cen- tionclear of the that uncertain the observations forcing produced support by the those eruptions simulations (Wag- with a tury with present day values, but the observational method nersmall and Zorita, temperature 2005; Schmidt response et al., to the 2011), Tambora so this says and little 1809 erup- The biases in the observed air temperatures mean that it’s abouttions. the Much accuracy of theof any variation GCM, but between it does simulations demonstratethat is a reflec- difficult to compare temperatures in the early nineteenth cen- thetion large of volcanic the uncertain response forcing present produced in several by simulations the eruptions did (Wag- not occur. tury with present day values, but the observational method ner and Zorita, 2005; Schmidt et al., 2011), so this says little Light Air (1) ● about the accuracy of any GCM, but it does demonstrate that Fresh Breeze (5) ● Light Breeze (2) ● the large volcanic response present in several simulations did P. BrohanModerate etBreeze al.: (4) EEIC early observations● 1557 Pleasant Breeze (4) ● Calm (0) ● not occur. Light Wind ● ● LightFresh Air Trade (1) (6) ● 10 Fresh BreezeModerate (5) (4) ● ● LightSteady Breeze Breeze (2) (5) ● ● Wind ● ● Moderate Breeze (4) ● 60 Moderate Trade (5) ● PleasantPleasant Breeze Trade (4) (5) ● ● 5 FreshCalm Gale (0) (8) ● ● StrongLight Breeze Wind (6) ● ● 30 Fresh TradeIncreasing (6) ● 10 ModerateLight Trade (4) (4) ● ● SteadyIncreasing Breeze Breeze(5) ● ● DecreasingWind ● ● 0 0 Steady Trade (6) ● ● 60 Moderate Trade (5) Latitude PleasantStrong Trade Gale (5) (9) ● ● Light ● ● −30 5 Fresh Gale (8) ● Strong Trade (7) ● Strong Breeze (6) ● 30 −5 Moderate Wind ● IncreasingBreeze ● ● −60 Light Trade Steady(4) ● Increasing Breeze ● ● Fresh Monsoon 0 0 DecreasingFine Breeze (5) ● ● SteadyDecreasing Trade Breeze(6) ● ● −10 Latitude StrongFresh Gale Wind (9) (6) ●● −45 0 45 90 135 180 Blowing FreshLight (6) ●● −30 StrongHard Trade Gale (7) (10) ●● Longitude Freshening ●● −5 Moderate Wind ● More Moderate ● Breeze ● Moderating ● 2 −60 Steady ● Moderate Monsoon (5) ● FreshStrong Monsoon Wind (8) ● ● FineBlowing Breeze Strong (5) (9) ● DecreasingSteady Breeze Monsoon ● ● 1 −10 FreshPleasant Wind Monsoon (6) ● ● −45 0 45 90 135 180 ● ● BlowingBlowing Fresh Hard (6) (10) ● ● ● Brisk Trade (5) ● ● ● ● ● ● Hard Gale (10) ● ● ● ● ● ● Longitude● ● ● ● ● 0 ● ● ● ● ● ● Trade ● ● ● ● ● ● ● ● ● ● Freshening ● ● ● ● ● ● ● ● Fresh ● ● ● ● ● More Moderate ● ● Pleasant Gale (5) ● ● Moderating ● 2 Brisk Gale (6) ● Moderate Monsoon (5) ● −1 Air ● StrongModerate Wind Gale (8) (7) ● Speed anomaly (m/s) ● BlowingUnsettled Strong (9) Wind ● SteadyStrong Monsoon Monsoon (7) ● ● 1 Pleasant Monsoon ● −2 ● Blowing Hard (10) ● ● ● Brisk Trade (5) ● ● ● ● ● ● ● ● ● 1790 1800 ● 1810 ● ● ●1820 ● 1830 ● ● 0 ● ● ● ● ● ● Trade 0 2 4 6 8 10 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Fresh ● ● ● Year ● Percent of observations ● ● Pleasant Gale (5) ● Brisk Gale (6) ● −1 Air ● Fig. 8.−1Wind-speed (WS) anomalies (m s ): observations minus Fig.Moderate 7. The mostGale (7) common● wind-force descriptors in the logs. Red Fig.a wind-speed 8.Speed anomaly (m/s) Wind Speed climatology (WS) anomalies for 1961–1990 (m/s): observations from the ERA-40 minus a re- Unsettled Wind ● points mark descriptors● that can be converted to a Beaufort force wind-speed climatology for 1961–90 from the ERA-40 reanalysis Strong Monsoon (7) analysis−2 (Uppala et al., 2005). Upper panel: truncated mean WS using the CLIWOC dictionary (the Beaufort category is given in (Uppalaanomaly et in al., each 2005). 2 × Upper2 degree panel: square. truncated Lower mean panel: WS truncated anomaly mean in each 2x2 degree square. Lower panel: truncated mean WS anomaly brackets after the descriptor in these cases). Grey points are terms WS anomaly1790 in each year.1800 1810 1820 1830 which can’t be converted.0 2 4 6 8 10 in each year. Percent of observations Year

Fig. 7. The most common wind-force descriptors in the logs. Red and time series of sympiesometer measurements (not shown) Fig. 8. Wind Speed (WS) anomalies (m/s): observations minus a Fig.points 7. The mark most descriptors commonwind-force that can be converteddescriptors to in a Beaufort the logs. force Red often show large drifts in pressure readings over a voyage. wind-speed climatology for 1961–90 from the ERA-40 reanalysis pointsusing mark the descriptors CLIWOC dictionary that can be (the converted Beaufort tocategory a Beaufort is given force in So the indications are that sympiesometer records will need (Uppala et al., 2005). Upper panel: truncated mean WS anomaly in usingbrackets the CLIWOC after the dictionary descriptor in (the these Beaufort cases). Grey category points is are given terms in close attention to calibration and correction in order to be each 2x2 degree square. Lower panel: truncated mean WS anomaly bracketswhich after can’t the be descriptor converted. in these cases). Grey points are terms useful for historical reconstructions. in each year. which can’t be converted. The pressure observations included in the attached IMMA records are all believed to be from mercury barometers, but the movement of the ship as it appears equally in stormy and it is possible that in some of the later voyages the pres- calm regions. sure observations are actually from a sympiesometer. That In 1818, Alexander Adie patented the sympiesometer, a is, a sympiesometer has been used in place of the usual mercury-less marine barometer containing coloured almond mercury barometer but the substitution is not mentioned oil and hydrogen gas (Middleton, 1964). Several of the later in the logbook. EEIC voyages carried a sympiesometer, either in tandem with the mercury barometer or as a standalone pressure 3.2.3 Wind speed gauge: 31 of the 893 digitized logbooks have records from sympiesometers. Of those 31 logs, 30 also contained records The wind-speed observations in the logbooks are, as is usual from a mercury barometer, and in some cases simultaneous at sea, subjective assessments based on the sails carried and observations from both instruments were recorded. the state of the sea. The vocabulary of such assessments was There are large and systematic differences between the not formally standardised until the 1830s, when Sir Francis sympiesometer and barometer measurements. The symp- Beaufort succeeded in introducing an ofﬁcial scale, but, even iesometer was designed to be portable and to respond rapidly in this pre Beaufort-scale age, sailors were very consistent to pressure changes, rather than for accuracy and stability; in their description of the winds, and it is possible to make

www.clim-past.net/8/1551/2012/ Clim. Past, 8, 1551–1563, 2012 BROHAN ET AL: EEIC EARLY OBSERVATIONS 9 1558 P. Brohan et al.: EEIC early observations

0.5 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

−0.5 ● bcc−csm1−1 FGOALS−gl

● ● ● ● ● ● ● ● ● ●● MIROC−ESM ● ●●● ● ● ●●●●●●●●●●●●● ● ●●●●●●●●● ● ● ● ●●●●●●●●●● ●●●●● ●●●●●●●●●●● ● ●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ●● −1 ●●●●●●●●●●●●●●●●●● MPI−ESM−P ●●●● 0 ● ● ●●●●●●●●●●●● ● ● ●●●● ● ● ●● ●●●●● ● ● ● ●●●●●●●●● ● ● ●●●●●●●● ● ●● ●● ● ● ●●● GISS−E2−R ●●● ● ●● ●●● ●●● AT anomaly (C) AT ● ● ● ●● ● ● ●●●● ● ● ● ● ●●● ●●●●● ●● CCSM4 ● ●● ● ●● ● ● ●● ● ●● ●● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● −1 ● ● ● ● ● ● −1.5 ● ● ● ● ●

● ● ● −2 ● ●

●

−2 ● −2● −1 0

0.6 B2000 RMO..2005 ● BOS..2001 MBH1999 DWJ2006 JBB..1998 HCA..2006 0.4 ECS2002 ● ● MJ2003 ● MSH..2005 ● ● ● O2005 0.2 ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 ● ● ● ● ● ● ● ● ● ● ● −0.2 ●

AT anomaly (C) AT ● ● ●

● −0.4 ●

● −0.6

1790 1800 1810 1820 1830 Year

Fig. 9. Comparison of observed, simulated, and proxy-derived large-scale near-surface temperature variability over the early nineteenth century. Upper panel:Fig. tropical 9. Comparison (observations of observed, coverage) simulated, and marine proxy-derived temperatures large-scale near-surface from observations temperature variability (black) over and the the early CMIP5 nineteenth simulations. Lower panel: Northern Hemispherecentury. Upper temperatures panel: tropical (observations from observations coverage) marine (black) temperatures and proxy from observations reconstructions. (black) and the Inset: CMIP5 relationship simulations. Lower between observations panel: northern hemisphere temperatures from observations (black) and proxy reconstructions. Inset: relationship between observations coverage (x) and Northerncoverage (x)Hemisphere and northern hemisphere (y) temperature (y) temperature in the in the simulations, simulations, with with best best fit line fit (slope line 1.2). (slope All series 1.2). normalised All series to have normalised mean to have mean zero over 1795–1805.zero The over grey 1795-1805. vertical The grey lines vertical mark lines the mark dates the dates of two of two large large volcanic eruptions eruptions (1809 and (1809 1815). and Data 1815). used is listed Data in tableused 1. is listed in Table 1. quantitative estimates of the wind speed from the language in (stronger trades imply a stronger sub-tropical high or a the logs (CLIWOC, 2003). Figure 7 shows the most frequent deeper equatorial low). wind-force terms used and their Beaufort equivalents where Although all the pressure and temperature observations available. in the selected logbooks were digitised, budget constraints About 80% of the terms encountered can be converted to meant that not all of the much more numerous and vari- Beaufort forces, and so to 10-m wind speeds in m s−1. Figure ous wind observations were. There are also wind observa- 8 shows the distribution of the inferred wind-speed anoma- tions in the more than 3000 logbooks in the BL archive that lies (difference from recent values). were not examined in this study (because they had no in- The notable feature of Fig. 8 is the large anomalies in strumental pressure or temperature observations). So much the trade-wind regions. It is possible that the trade winds more information on wind fields is still potentially avail- were stronger around 1800, but as no equivalent anomaly ap- able in the BL EEIC logbook archive. If extracted in a fu- pears in the pressure fields it is more likely that the CLI- ture project, these records would provide information on WOC dictionary is slightly miscalibrated in those regions sub-daily weather variability back into the 17th century.

Clim. Past, 8, 1551–1563, 2012 www.clim-past.net/8/1551/2012/ P. Brohan et al.: EEIC early observations 1559

Table 2. Ships from which observations were taken (1 of 2 – starting Table 2. Continued. dates 1789 to 1803). Ship Name Years of operation Ship Name Years of operation Earl Howe 1798–1810 Melville Castle 1789–1790,1792–1793,1796–1802 Lord Duncan 1798–1806 Rose (2) 1789–1790,1799–1800 Ocean (3) 1798–1800 Barwell (1) 1790–1791,1795–1796 Tellicherry 1798–1799 Belvedere 1790–1791 Earl Cornwallis 1798–1800 Earl Of Abergavenny (2) 1790,1797–1800 Orpheus 1798–1800 Marquis Of Lansdown 1790–1791,1793–1800 Charlton 1799–1806 Ocean (1) 1791–1797 Asia (4) 1799–1803 Bridgewater (3) 1791–1793,1796–1797 Hindostan (2) 1799–1800 Lascelles 1792–1796 Duke Of Buccleugh (1) 1799–1800 Middlesex (2) 1792–1795 Preston 1799–1800 Royal Admiral (1) 1792–1796 Herculean 1800–1801 Swallow (3) 1792–1794 Dorsetshire 1800–1801,1803–1804,1806,1811– Ganges (1) 1792–1795 1812,1814–1815,1817–1818,1820–1823 General Goddard 1792–1793 Earl Spencer (2) 1800–1801,1803–1810 Pigot (2) 1793–1794 Neptune (5) 1800–1801,1804–1807,1809–1810,1812–1815 Ceres (2) 1793–1794 Hugh Inglis 1800–1801,1810–1812 Warley (1) 1793–1794 Lady Burges 1800–1805 Berrington 1793–1794 Walthamstow 1800–1801,1804–1805,1808–1813 General Coote 1793–1794 Lord Nelson 1800–1801,1806–1807 Rodney (2) 1793–1796 Ceres (4) 1800–1805,1808–1809 Princess Amelia (3) 1793–1796 City Of London 1800–1801,1803–1808,1812–1813 Francis (2) 1793–1796 Bengal 1800–1802,1806–1807 Exeter (2) 1793–1794,1800–1801,1803–1804,1810–1811 Canton 1800–1805,1808–1811 Lord Thurlow 1793–1794,1797–1802 Georgiana (1) 1800–1803,1805–1807 Lord Walsingham 1793–1794,1797–1799 Hawke (5) 1800–1801 Minerva (1) 1793–1796,1799–1800 Earl St 1800–1801,1808–1813 Earl Of Chesterﬁeld 1793–1794 Henry Dundas 1801–1802 Earl Of Wycombe 1794–1795,1797–1799 Calcutta (4) 1801–1804 Sir Edward Hughes 1794–1795,1797–1803 Alfred (2) 1801–1802,1807–1808,1810–1811 Woodford (1) 1794–1805,1807–1808,1810–1811 Caledonian (2) 1801–1803 Thetis (1) 1794–1797 Henry Addington (2) 1801–1802,1805–1806,1811–1812,1814–1815 Rockingham (1) 1794–1795,1798–1802 Walpole (5) 1801–1802 Walpole (4) 1794–1795 Ocean (4) 1801–1803,1805–1806,1808–1809 Phoenix (3) 1794–1795 Northampton (2) 1801–1805,1807–1810,1818–1819 Lord Hawkesbury 1794–1802,1804–1806 Princess Mary (2) 1801–1805 Taunton Castle 1794–1795,1799–1800,1804–1805,1809–1810 Fort William (2) 1801–1802 Europa (2) 1794–1795 Monarch 1801–1802 Queen (4) 1794–1798 Experiment (2) 1801 Carnatic (2) 1794–1795,1801–1802 Princess Of Wales (3) 1795–1797 Manship (2) 1801–1803 Earl Of Oxford 1795–1796 Sarah Christiana 1801–1802 Cirencester 1795–1796,1812–1813 Comet (2) 1801–1803 London (13) 1795–1796 Marquis Of Ely 1802–1805,1811–1814,1819–1820 Bellona 1795–1798 Marquis Wellesley 1802–1803,1806–1808 Hillsborough (2) 1795–1798 Castle Eden 1802–1804 Kent (5) 1795–1797 Lady Jane Dundas 1802–1807 Woodcot 1795–1796 Thames (2) 1802–1805,1809–1810,1812–1813 Brunswick (1) 1795–1797 Sir William Bensley 1802–1811 Cuffnells 1796–1800,1802–1805,1809–1810 Tottenham 1802–1803,1806–1810 Princess Charlotte (1) 1796–1797 Travers 1802–1806 Albion (2) 1796–1798 Alnwick Castle 1802–1813,1815–1816 Royal Charlotte (5) 1796–1807,1809–1815 Marchioness Of Exeter 1802–1803,1811–1814,1816–1817 Essex (4) 1796–1798 Devaynes 1802–1808,1811–1812 True Briton (4) 1796–1798,1801–1802 Ann (1) 1803–1809,1814–1815 Airly Castle 1796–1797,1804–1806 Experiment (4) 1803–1805 Walmer Castle 1796–1798,1802–1805,1815–1816 Cumberland 1803–1804,1809–1810 Boddam 1796–1800 Warren Hastings (2) 1803–1804 Manship (1) 1796–1800 Harriet (3) 1803–1811 Good Hope (3) 1796–1799 Elphinstone 1803–1811 Henry Addington (1) 1796–1798 Tigris (2) 1803–1805,1810–1815 Ganges (3) 1797–1802 Marquis Cornwallis (2) 1803 Prince William Henry 1797–1799 Windham (2) 1803–1806,1816–1817 Britannia (4) 1797–1805 Europe (2) 1803–1807 Warley (2) 1797–1800,1805–1809,1811–1814 Euphrates 1803–1805 Hope (2) 1797–1808,1811–1816 General Stuart 1803–1804,1807,1811–1814 Arniston 1797–1798,1804–1807,1810–1811 Essex (5) 1803–1805,1808–1809,1819–1820 Eurydice 1797–1799 Carmarthen 1803–1818 Sulivan 1797–1798 Union (4) 1803–1804,1808–1812,1815–1818 Osterley (3) 1798–1800

www.clim-past.net/8/1551/2012/ Clim. Past, 8, 1551–1563, 2012 1560 P. Brohan et al.: EEIC early observations

Table 3. Ships from which observations were taken (2 of 2 – starting dates 1803 to 1833). Table 3. Contined.

Ship Name Years of operation Ship Name Years of operation Ocean (5) 1803–1805 Lowther Castle 1813–1828,1831–1834 Lord Melville (1) 1803–1808,1811–1816 Bombay 1814–1815,1817–1822,1825– Dover Castle 1804–1805,1809–1810 1828,1831–1834 Indus 1804–1805,1810–1815 Prince Regent 1814–1815,1818–1829,1833–1834 Alexander (3) 1804–1809,1814 Lady Melville 1814–1821,1824–1827,1829–1834 Lord Eldon 1804–1814 Minerva (7) 1815–1822,1825–1832 Waller 1804 Surrey (2) 1815 Winchelsea (3) 1804–1807,1810–1815,1817– General Kyd 1815–1816,1823–1832 1818,1820–1823,1831–1832 James Sibbald 1815–1816,1826–1829 Ocean (6) 1804–1814 Sovereign (2) 1816–1817 Huddart 1804–1805,1808–1809,1815–1816 Northampton (3) 1816 United Kingdom 1804–1805,1807–1808 Fort William (3) 1816–1817 Bombay Castle 1805–1806 Mangles 1816–1819 Surrey (1) 1805–1810 Buckinghamshire 1816–1824 Northumberland (5) 1805–1818 Providence (1) 1816–1817 Royal George (4) 1805–1811,1814–1818 Sir William Pultney 1805–1807,1815–1816 Larkins (1) 1816–1817 Streatham (4) 1805–1814,1817–1818 Earl Of Balcarras 1816–1833 Glory 1805–1807 Vansittart (4) 1817–1824,1827–1834 William Pitt (2) 1805–1807,1810–1820 Lord Castlereagh (1) 1817–1820 Phoenix (5) 1805–1809,1816–1819 Waterloo (1) 1817–1832 Wexford 1805–1817 Bridgewater (5) 1817–1830 David Scott (2) 1806–1807,1812–1813,1815–1816 Herefordshire 1817–1818,1821–1826,1829–1834 Glatton (4) 1806–1807,1809–1810,1812–1815 Barkworth 1817–1818 Sir Stephen Lushington 1806–1811 Castle Huntley 1818–1821,1824–1825,1828– Regent 1807,1816–1819,1822 1831,1833–1834 Lady Castlereagh 1807–1817 General Hewett 1818–1825 Admiral Gardner 1807–1808 London (14) 1818–1823,1826–1833 Union (5) 1807–1808,1813–1814 Canning 1818–1832 Lord Keith 1808–1811,1814–1819 Duke Of York (2) 1818–1826,1829–1830 Princess Amelia (4) 1809–1825 Dunira 1818–1819,1822–1823,1830–1833 Thomas Grenville 1809–1832 Thomas Coutts 1818–1825,1828–1833 Warren Hastings (3) 1809–1810,1814–1821,1825– Henry Porcher 1818–1819 1828,1831–1834 Matilda 1819–1820 Coutts 1809–1810,1812–1815 Kellie Castle 1819–1830,1833–1834 Lady Lushington 1809–1814,1818–1819 Inglis 1819–1832 Farlie 1809–1814,1818–1819 Thames (5) 1819–1821,1824–1827,1829–1834 Charles Grant 1810–1816,1819–1830,1832–1833 Cornwall 1819–1820,1826 Surat Castle (2) 1810–1815 Windsor (2) 1819–1820,1825–1828,1832–1833 Lady Carrington 1810,1812–1817 Marchioness Of Ely 1820–1821,1826–1829 Midas 1810–1811 Orwell 1820–1831 1811–1812,1815–1816,1819– Warren Hastings (5) Kent (7) 1821–1824 1820,1823–1826 Royal George (5) 1821–1824 Carnatic (3) 1811–1820 John Palmer 1811 Farquharson 1821–1828,1831–1834 Cambridge 1811–1812,1825–1827 William Fairlie 1822–1833 Scaleby Castle 1811–1812,1814–1828,1831–1834 Berwickshire 1822–1833 William Pitt (3) 1811–1812 Sir David Scott 1822–1825,1830–1833 Harleston 1811–1812 Duchess Of Athol 1822–1826,1828–1833 Moffat 1811–1812,1818–1819 Repulse 1823–1830 Rose (4) 1811–1822,1824–1827,1833–1834 Claudine 1824–1825 General Harris 1812–1831 Macqueen 1824–1825,1830–1833 Broxbornebury 1812–1813,1825–1828 Java 1825 Asia (6) 1812–1826,1832–1833 Clyde (2) 1825–1826 Marquis Of Huntley 1812–1813,1818–1823,1831–1834 George The Fourth 1826–1833 Perseverance (2) 1812–1814,1818–1819 Edinburgh 1826–1831 Marquis Camden 1812–1814,1821–1829,1832–1833 Reliance 1828–1833 Juliana 1812–1813 Abercrombie Robinson 1828–1833 Astell 1812–1813,1818–1821,1824– Maitland 1828–1829 1825,1830–1831 Asia (10) 1829–1830 Princess Charlotte Of Wales 1812–1822,1825–1828 Susan (2) 1830–1831 Coldstream 1812–1813,1816–1817,1822–1823 Marquis Of Hastings 1830–1831 Cabalva 1812–1817 Lord Lowther 1830–1833 David Scott (1) 1813–1814 Duke Of Sussex 1831–1832 Marquis Of Wellington (1) 1813–1822,1827,1829–1830 Duke Of Buccleugh (2) 1831–1832 Atlas (4) 1813–1830 Bencoolen 1832–1833 Sherborne (2) 1833–1834

Clim. Past, 8, 1551–1563, 2012 www.clim-past.net/8/1551/2012/ P. Brohan et al.: EEIC early observations 1561

4 Constraining proxy reconstructions and GCM new weather observations extracted from those logs provide simulations material for detailed reconstructions of the weather and climate between 1789 and 1834 and offer new insights into The biases in the observed air temperatures mean that it is pre-industrial climate variability. For all three meteorological difficult to compare temperatures in the early nineteenth cen- variables studied (temperature, pressure and wind) it is clear tury with present-day values, but the observational method that the data can be used for investigating variability over and ship routes were constant over the period covered by the period of measurement, though comparison with mea- the EEIC observations so they can be used directly to look surements made decades or centuries later will require close at temperature variations over the period 1795–1833 (when attention to observational biases. there were enough ships contributing to give reliable results). The observations demonstrate that the large-scale temper- Extracting a set of pseudo-observations from each GCM run, ature change, over the Atlantic and Indian Oceans, associated by sampling from the model output fields at the date and lo- with the two big tropical volcanic eruptions in 1809 and 1815 cation of each observation allows a direct comparison be- was modest (perhaps 0.5 ◦C). Some of the GCM simulations tween observations and simulations (Fig. 9: upper panel). It in the CMIP5 ensemble show much larger volcanic effects is clear that the observational coverage is sufficient to show than this – such simulations are unlikely to be accurate in the effect of the volcanoes in the simulations, and it is also this respect. Recent annually-resolved proxy reconstructions clear that the observations support those simulations with a of Northern Hemisphere temperature show a varied but sim- small temperature response to the Tambora and 1809 erup- ilarly modest volcanic response (about 0.2–0.7 ◦C); the new tions. Much of the variation between simulations is a reflec- observations therfore provide an out-of-sample validation for tion of the uncertain forcing produced by the eruptions (Wag- the proxy reconstructions – supporting their use for longer- ner and Zorita, 2005; Schmidt et al., 2011), so this says little term climate reconstructions. about the accuracy of any GCM, but it does demonstrate that the large volcanic response present in several simulations did not occur. Supplementary material related to this article is Comparison with the proxies is more complicated, as they available online at: http://www.clim-past.net/8/1551/ do not usually provide field reconstructions – just time series 2012/cp-8-1551-2012-supplement.zip. for the entire Northern Hemisphere (NH). However, comparing the GCM results subsampled to the coverage of the observations with their NH averages (Fig. 9: inset) indicate that the Acknowledgement. This work was made possible by the UK De- NH temperature anomalies are linearly related to the obser- partment for Environment, Food and Rural Affairs (Defra), which vational anomalies 1TNH ≈ 1Tobs × 1.2, and Fig. 9 (lower paid for the investigation and photography of the original log-books; panel) compares the proxy reconstructions to the observa- and by the US Climate Database Modernisation Program (CDMP) tional series scaled by this factor. which transcribed the weather records from the photographs. PB With the exception of the glacier based reconstruction was also supported by the Joint DECC and Defra Integrated Cli- mate Programme, DECC/Defra (GA01101). (O2005), which has (unsurprisingly) too little variance on these short time scales, the agreement between the proxy re- We acknowledge the World Climate Research Programme’s constructions and the observations is good. It is not easy to Working Group on Coupled Modelling, which is responsible say which of the proxy series is the best, but as a group they for CMIP, and we thank the climate modelling groups (listed in match the observations well. As this comparison is for a pe- Table 1) for producing and making available their model output. For CMIP the US Department of Energy’s Program for Climate riod outside that used to calibrate the proxies (both in time Model Diagnosis and Intercomparison provides coordinating and temperature), the observations form a validation for the support and led development of software infrastructure in partner- proxy reconstructions – demonstrating that the proxies can ship with the Global Organization for Earth System Science Portals. be used to extrapolate back into the past and into different climates, with success. Edited by: B. Vinther

5 Conclusions References

The records of the English East India Company (EEIC), Alcoforado, M. J., Vaquero, J. M., Trigo, R. M., and Taborda, J. P.: Early Portuguese meteorological measurements (18th century), archived in the British Library, offer a remarkable new in- Clim. Past, 8, 353–371, doi:10.5194/cp-8-353-2012, 2012. sight into the weather and climate of the late eighteenth and Allan, R. J. and Ansell, T. J.: A new globally complete monthly early nineteenth centuries. Their archives include 891 ships’ historical gridded mean sea level pressure data set (HadSLP2): logbooks containing daily temperature and pressure mea- 1850–2003, J. Climate, 19, 5816–5842, 2006. surements, and wind-speed estimates, each covering a voy- Ansell, T. J., Jones, P. D., Allan, R. J., Lister, D., Parker, D. E., age from England to India or China and back. The 273 000 Brunet, M., Moberg, A., Jacobeit, J., Brohan, P., Rayner, N. A., www.clim-past.net/8/1551/2012/ Clim. Past, 8, 1551–1563, 2012 1562 P. Brohan et al.: EEIC early observations

Aguilar, E., Alexandersson, H., Barriendos, M., Brandsma, T., Farrington, A.: A Catalogue of East India Company Ships, Journals Cox, N. J., Della-Marta, P. M., Drebs, A., Founda, D., Ger- and Logs 1600–1834, British Library, 1999. stengarbe, F., Hickey, K., Jonsson, T., Luterbacher, J., Nordli, Farrington, A., Lubker, S., Radok, U., and Woodruff, S.: South At- O., Oesterle, H., Petrakis, M., Philipp, A., Rodwell, M. J., lantic winds and weather during and following the Little Ice Age Saladie, O., Sigro, J., Slonosky, V., Srnec, L., Swail, V., – a pilot study of English East India Company Ship Logs, Mete- Garcia-Suarez, A. M., Tuomenvirta, H., Wang, X., Wanner, H., orol. Atmos. Phys., 67, 253–257, 1998. Werner, P., Wheeler, D., and Xoplaki, E.: Daily mean sea level Foster, W.: The Voyage of Sir Henry Middleton to the Moluccas, pressure reconstructions for the European-North Atlantic Re- 1604–1606, Hakluyt Society, 2010. gion for the period 1850–2003, J. Climate, 19, 2717–2742, Frank, D., Esper, J., Zorita, E., and Wilson, R.: A noodle, doi:10.1175/JCLI3775.1, 2006. hockey stick, and spaghetti plate: a perspective on high- Berry, D. I. and Kent, E. C.: The effect of instrument exposure on resolution paleoclimatology, WIREs Climate Change, 1, 507–16, marine air temperatures: an assessment using VOSClim data, Int. doi:10.1002/wcc.53, 2010. J. Climatol., 25, 1007–22, doi:10.1002/joc.1178, 2005. Gent, P. R., Danabasoglu, G., Donner, L. J., Holland, M. M., Hunke, Briffa, K. R.: Annual climate variability in the Holocene: interpret- E. C., Jayne, S. R., Lawrence, D. M., Neale, R. B., Rasch, P. J., ing the message of ancient trees, Quaternary Sci. Rev., 19, 87– Vertenstein, M., Worley, P. H., Yang, Z.-L., and Zhang, M.: The 105, 2000. Community Climate System Model Version 4, J. Climate, 24, Briffa, K. R., Osborn, T. J., Schweingruber, F. H., Harris, I. C., and 4973–4991, doi:10.1175/2011JCLI4083.1, 2011. Jones, P. D.: Low-frequency temperature variations from a north- Hegerl, G. C., Crowley, T. J., Hyde, W. T., and Frame, D. J.: Climate ern tree ring density network, J. Geophys. Res., 106, 2929–2941, sensitivity constrained by temperature reconstructions over the 2001. past seven centuries, Nature, 440, 1029–1032, 2006. Briffa, K. R., Osborn, T. J., and Schweingruber, F. H.: Large-scale Jansen, E., Overpeck, J., Briffa, K. R., Duplessy, J. C., Joos, temperature inferences from tree rings: a review, Global Planet. F., Masson-Delmotte, V., Olago, D., Otto-Bliesner, B., Peltier, Change, 40, 11–26, 2004. W. R., Rahmstorf, S., Ramesh, R., Raynaud, D., Rind, D., Solom- Brohan, P., Kennedy, J., Harris, I., Tett, S. F. B., and Jones, P. D.: ina, O., Villalba, R., and Zhang, D.: PalaeoClimate, chap. 6 of Uncertainty Estimates in Regional and Global Observed Temper- Climate Change 2007: The Physical Science Basis. Contribution ature Changes: a new dataset from 1850, J. Geophys. Res., 111, of Working Group I to the Fourth Assessment Report of the In- D12106, doi:10.1029/2005JD006548, 2006. tergovernmental Panel on Climate Change, edited by: Solomon, Brohan, P., Allan, R., Freeman, J. E., Waple, A., Wheeler, S., Qin, D., Manning, M., Marquis, M., Averyt, K. B., Tignor, D., Wilkinson, C., and Woodruff, S.: Marine observations M., Miller, H. L., and Chen, Z., 433–497, Cambridge University of old weather, B. Am. Meteorol. Soc., 90, 219–230, Press, 2007. doi:10.1175/2008BAMS2522.1, 2009. Jones, P. D. and Mann, M. E.: Climate over past millennia, Rev. Brohan, P., Ward, C., Willets, G., Wilkinson, C., Allan, R., and Geophys., 42, RG2002, doi:10.1029/2003RG000143, 2004. Wheeler, D.: Arctic marine climate of the early nineteenth cen- Jones, P. D., Briffa, K., Barnett, T. P., and Tett, S. F. B.: High- tury, Clim. Past, 6, 315–324, doi:10.5194/cp-6-315-2010, 2010. resolution palaeoclimatic records for the last millennium: inter- Camuffo, D. and Bertolin, C.: The earliest temperature observations pretation, integration and comparison with General Circulation in the world: the Medici Network (1654–1670), Clim. Change, Model control run temperatures, Holocene, 8, 455–471, 1998. 111, 335–363, doi:10.1007/s10584-011-0142-5, 2011. Jones, P. D., Briffa, K. R., and Osborn, T. J.: The evolution of cli- Chenoweth, M.: Ship’s Logbooks and “The Year Without an Sum- mate over the last millennium, Science, 292, 662–667, 2001. mer”, B. Am. Meteorol. Soc., 77, 2077–2094, 1996. Jones, P. D., Briffa, K. R., Osborn, T. J., Lough, J. M., van Ommen, Chenoweth, M.: A new methodology for Homogenization of 19th T. D., Vinther, B. M., Luterbacher, J., Wahl, E. R., Zwiers, F. W., century marine air temperature data, J. Geophys. Res., 105, Mann, M. E., Schmidt, G. A., Ammann, C. M., Buckley, B. M., 29145–29154, 2000. Cobb, K. M., Esper, J., Goose, H., Graham, N., Jansen, E., Kiefer, CLIWOC: CLIWOC multilingual dictionary: an English-Spanish- T., Kull, C., Kuttel, M., Mosley-Thompson, E., Overpeck, J. T., Dutch-French dictionary of wind force terms used by mariners Riedwyl, N., Schulz, M., Tudhope, A. W., Villalba, R., Wanner, from 1750 to 1850, KNMI publication 205, De Bildt, The H., Wolff, E., and Xoplaki, E.: High-resolution palaeoclimatol- Netherlands, 2003. ogy of the last millennium: a review of current status and future Cook, E. R., Esper, J., and D’Arrigo, R. D.: Extra-tropical North- prospects, Holocene, 19, 3–49, doi:10.1177/0959683608098952, ern Hemisphere land temperature variability over the past 1000 2009. years, Quaternary Sci. Rev., 23, 2063–2074, 2004. Mann, M. E. and Jones, P. D.: Global surface temperatures Dalrymple, A.: Journal of a voyage to the East Indies in the ship over the past two millennia, Geophys. Res. Lett., 30, 1820, Grenville, Captain Burnet Abercrombie, in the year 1775, Philos. doi:10.1029/2003GL017814, 2003. T. R. Soc. Lond., 68, 389–418, 1778. Mann, M. E., Bradley, R. S., and Hughes, M. K.: Northern Hemi- D’Arrigo, R. D., Wilson, R., and Jacoby, G.: On the long-term con- sphere temperatures during the past millennium: inferences, un- text for late twentieth century warming, J. Geophys. Res., 11, certainties, and limitations, Geophys. Res. Lett., 26, 759–762, D03103, doi:10.1029/2005JD006352, 2006. 1999. Esper, J., Cook, E. R., and Schweingruber, F. H.: Low-Frequency Mann, M. E., Fuentes, J. D., and Rutherford, S.: Underes- Signals in Long Tree-Ring Chronologies for Reconstructing Past timation of volcanic cooling in tree-ring-based reconstruc- Temperature Variability, Science, 295, 2250–2253, 2002. tions of hemispheric temperatures, Nat. Geosci., 5, 202–205, doi:10.1038/ngeo1394, 2012.

Clim. Past, 8, 1551–1563, 2012 www.clim-past.net/8/1551/2012/ P. Brohan et al.: EEIC early observations 1563

Marsland, S. J., Haak, H., Jungclaus, J. H., Latif, M., and Roeske, Taylor, K. E., Stouffer, R. J., and Meehl, G. A.: An Overview of F.: The Max-Planck-Institute global ocean/sea ice model with CMIP5 and the Experiment Design, B. Am. Meteorol. Soc., 93, orthogonal curvilinear coordinates, Ocean Modell., 5, 91–127, 485–498, doi:10.1175/BAMS-D-11-00094.1, 2012. 2003. Uppala, S. M., Kallberg,˚ P. W., Simmons, A. J., Andrae, U., Middleton, W. E. K.: The history of the barometer, Baros Books, da Costa Bechtold, V., Fiorino, M., Gibson, J. K., Haseler, J., 1964. Hernandez, A., Kelly, G. A., Li, X., Onogi, K., Saarinen, S., Moberg, A., Sonechkin, D. M., Holmgren, J. K., Datsenko, N. M., Sokka, N., Allan, R. P., Andersson, E., Arpe, K., Balmaseda, and Karlen, W.: Highly variable Northern Hemisphere temper- M. A., Beljaars, A. C. M., van de Berg, L., Bidlot, J., Bormann, atures reconstructed from low- and high-resolution proxy data, N., Caires, S., Chevallier, F., Dethof, A., Dragosavac, M., Fisher, Nature, 433, 613–617, 2005. M., Fuentes, M., Hagemann, S., Holm,´ E., Hoskins, B. J., Isak- Oerlemans, J.: Extracting a Climate Signal from 169 Glacier sen, L., Janssen, P. A. E. M., Jenne, R., McNally, A. P., Mahfouf, Records, Science, 308, 675–677, doi:10.1126/science.1107046, J.-F., Morcrette, J.-J., Rayner, N. A., Saunders, R. W., Simon, P., 2005. Sterl, A., Trenberth, K. E., Untch, A., Vasiljevic, D., Viterbo, P., Raddatz, T. J., Reick, C. H., Knorr, W., Kattge, J., Roeckner, E., and Woollen, J.: The ERA-40 re-analysis., Q. J. Roy. Meteorol. Schnur, R., Schnitzler, K. G., Wetzel, P., and Jungclaus, J.: Will Soc., 131, 2961–3012, doi:10.1256/qj.04.176, 2005. the tropical land biosphere dominate the climate-carbon cycle Wagner, S. and Zorita, E.: The inﬂuence of volcanic, solar and feedback during the twenty-ﬁrst century?, Clim. Dynam., 29, CO2 forcing on the temperatures in the Dalton Minimum 565–574, doi:10.1007/s00382-007-0247-8, 2007. (1790–1830): a model study, Clim. Dynam., 25, 205–218, Rayner, N. A., Parker, D. E., Horton, E. B., Folland, C. K., Alexan- doi:10.1007/s00382-005-0029-0, 2005. der, L. V., Rowell, D. P., Kent, E. C., and Kaplan, A.: Global Watanabe, S., Hajima, T., Sudo, K., Nagashima, T., Takemura, T., analyses of SST, sea ice and night marine air temperature Okajima, H., Nozawa, T., Kawase, H., Abe, M., Yokohata, T., since the late nineteenth century, J. Geophys. Res., 108, 4407, Ise, T., Sato, H., Kato, E., Takata, K., Emori, S., and Kawamiya, doi:10.1029/2002JD002670, 2003. M.: MIROC-ESM 2010: model description and basic results of Rutherford, S., Mann, M. E., Osborn, T. J., Briffa, K. R., Jones, CMIP5-20c3m experiments, Geosci. Model Devel., 4, 845–872, P. D., Bradley, R. S., and Hughes, M. K.: Proxy-Based Northern doi:10.5194/gmd-4-845-2011, 2010. Hemisphere Surface Temperature Reconstructions: Sensitivity to Wheeler, D., Herrera, R. G., Koek, F., Wilkinson, C., Konnen, G., Method, Predictor Network, Target Season, and Target Domain, del Rosario Prieto, M., Jones, P. D., and Casale, R.: CLIWOC, J. Climate, 18, 2308–2329, doi:10.1175/JCLI3351.1, 2005. Climatological database for the world’s oceans, European Com- Schmidt, G. A., Ruedy, R., Hansen, J. E., Aleinov, I., Bell, N., mission, 2006. Bauer, M., Baue, S., Cairns, B., Canut, V., Cheng, Y., Genio, Woodruff, S.: Archival of data other than in IMMT format: The In- A. D., Faluvegi, G., Friend, A. D., Hall, T. M., Hu, Y., Kelley, ternational Maritime Meteorological Archive (IMMA) format, in M., Kiang, N. Y., Koch, D., Lacis, A. A., Lerner, J., Lo, K. K., Expert Team on Marine Climatology, Second Session, Geneva, Miller, R. L., Nazarenko, L., Oinas, V., Perlwitz, J., Perlwitz, J., Switzerland, 26–27 March 2007 (Annex VII), JCOMM Techni- Rind, D., Romanou, A., Russell, G. L., Sato, M., Shindell, D. T., cal Report, No. 40, 68–101, 2007. Stone, P. H., Sun, S., Tausnev, N., Thresher, D., and Yao, M.-S.: Wu, T., Li, W., Ji, J., Xin, X., Li, L., Wang, Z., Zhang, Y., Li, J., Present day atmospheric simulations using GISS Model E: Com- Zhang, F., Wei, M., Shi, X., Wu, F., Zhang, L., Chu, M., Jie, W., parison to in-situ, satellite and reanalysis data, J. Climate, 19, Liu, Y., Wang, F., Liu, X., Li, Q., Dong, M., Liu, Q., and Zhang, 153–192, 2006. J.: Global carbon budgets simulated by the Beijing Climate Cen- Schmidt, G. A., Jungclaus, J. H., Ammann, C. M., Bard, E., Bra- ter Climate System Model for the last century, J. Climate, sub- connot, P., Crowley, T. J., Delaygue, G., Joos, F., Krivova, mitted, 2012. N. A., Muscheler, R., Otto-Bliesner, B. L., Pongratz, J., Shin- Zhou, T., Wu, B., Wen, X., Li, L., and Wang, B.: A Fast Version dell, D. T., Solanki, S. K., Steinhilber, F., and Vieira, L. E. A.: of LASG/IAP Climate System Model and Its 1000-year Control Climate forcing reconstructions for use in PMIP simulations Integration, Adv. Atmos. Sci., 25, 655–672, 2008. of the last millennium (v1.0), Geosci. Model Devel., 4, 33–45, doi:10.5194/gmd-4-33-2011, 2011.

www.clim-past.net/8/1551/2012/ Clim. Past, 8, 1551–1563, 2012