Southern Ocean sea ice, icebergs, and meteorological data from maritime sources for the period 1929 to 1940

Abstract

Maritime historical documentary sources of weather and state of sea surface including sea ice can aid in filling a known climate knowledge gap for the Southern Ocean and Antarctica for the first half of the 20th century. This study presents a data set of marine climate, sea ice and icebergs recovered from a collection of logbooks from mainly Norwegian whaling factory ships that operated in the Southern Ocean during 1929–1940. The data set comprises some 8000 weather and 4000 sea ice/open sea records from austral summers of the study period. This paper further discusses the structure and content of most common Norwegian maritime documentary sources of the period along with the practices of logging information relevant for the study, such as time keeping, positioning and making weather observations. An emphasis was made on recovery of notes on sea ice and icebergs and their interpretation in terms of WMO categories of sea ice concentration. Data, including ship-related metadata from all individual documents are homogenized and structured to a common machine-readable format that simplifies its ingestion into relevant climate data depositories.

1 INTRODUCTION

There is a substantial gap in our knowledge of the climate in the Southern Ocean (SO) and Antarctica before the IGY-1957 when the first observational network in the region was established. Compared with a relative abundance of instrumental climate data from the Northern Hemisphere, the Antarctic region is still largely underrepresented, suffering from paucity and shortness of available instrumental series (e.g. Turner et al., 2005). The higher quality regular sea ice data were not available until the start of the satellite-based monitoring in 1979 (Cavalieri et al., 1999) with ongoing effort to extend it back in time based on earlier passive microwave (Cavalieri et al., 2003) and optical/infrared satellite observations (Gallaher et al., 2014). Such shortness of instrumental records hampers the understanding of the causes of recent climatic tendencies and variability in the high-latitude Southern Hemisphere (e.g. Jones et al., 2016). It also limits the skills of climate reanalysis products for the region in the pre-1980 period with implications for understanding the signature, causes, and future projections of anthropogenic warming in the Antarctic region (Bromwich & Fogt, 2004; Rayner et al., 2003).

Complementary data sources such as ships' logbooks have proven to be a successful tool in reconstructing past marine climate. The value of these data sources in advancing our understanding of the climate of the industrial era and hence the effects of anthropogenic pressure on the global climate system is well-recognized by the scientific community (see e.g. a review of Luterbacher et al., 2024 and references therein).

During navigation, the information on weather and the state of the sea surface, including sea ice and icebergs when sailing in polar areas would be routinely registered in several types of logbooks. The recent years have seen significant efforts in finding, recovery, and quality control of historical marine weather observations, implemented via a number of dedicated programmes such as the international ACRE initiative (www.met-acre.net, Allan et al., 2011) which coordinates various data-rescue efforts and communities for both terrestrial and ocean observations. The derived data sets are a major source of instrumental data for one of the largest marine climate data archive, ICOADS Release 3.0 (International Comprehensive Ocean–Atmosphere Data Set, Freeman et al., 2017). Despite this considerable progress, data from the Antarctic region and the Southern Ocean are largely yet to be recovered and analysed. In contrast to logbooks from infrequent Antarctic expeditions (e.g. Edinburgh & Day, 2016), accounts from commercial vessels are much more abundant and were shown to represent a promising source of valuable climate information for the region (Teleti et al., 2019).

Early well-documented attempts of economic activity in the region are dated to the late 19th century, and since the turn of the 20th century, a growing number of vessels mainly from different national whaling fleets have been active in the area during Austral late spring to early fall (Tønnessen & Johnsen, 1982). Harvesting of blue whales—main target for commercial whaling – in the Antarctic started in the 1904/05 season expanding gradually and reaching its peak in 1930/31 when staggering 29,409 blue whales were taken over a single season (Branch et al., 2004). With a temporal halt of whaling in the region in 1940 after the outbreak of World War Two, whaling in the region resumed already in 1945 and continued throughout most of the 20th century until the introduction of restrictions on the Southern Ocean whaling in the late 1960s, followed by the International Whaling Commission's global whaling moratorium in 1986.

This study contributes to the ongoing effort to recovery of climate information from commercial whaling activity in the Southern Ocean by presenting a new collection of historical weather and sea state data including sea ice observations for the region. We focused mainly on documents from Norwegian whaling companies and accounts of vessels from companies with UK or Irish registrations that were operated by crews of Norwegian origin (Tønnessen & Johnsen, 1982). The whaling vessels owned by Norwegian companies operated in the region since the late 19th century. Starting from the late 1920s, tens of whalers/harpoon boats and factory ships were annually present in the whaling grounds of the Southern Ocean (Tønnessen & Johnsen, 1982). This activity left behind a considerable stratum of documents which was proven to be valuable sources of climate-relevant information for the region yet largely underrepresented in current climate data archives.

An important scope of the activity was an outlook for future efforts in reconstructing past sea ice extent variability in the Southern Ocean for the pre-satellite period of the 20th century. Though successful attempts on that subject were made earlier and covered the period of 1931–1987 (de La Mare, 1997; de la Mare, 2009; Cotté and Guinet, 2007), these studies relied exclusively on whale catch positions as proxies for sea ice extent rather than direct reports on sea ice, implying a large potential for biases in the reconstructions associated with changing whaling practices as well as species harvested.

The data presented in this study mainly covers the Atlantic and Indian sectors of the Southern Ocean, with some observations acquired in the Pacific sector or while vessels were in transit to or from the whaling area.

The article is organized as follows. In Section 2, we present the documents used in the present study. Section 3 provides details on the time and date, weather, and sea ice information recovered. Section 4 summarizes the recovered data. Note that Supplementary materials provide example images of the analysed documents with details on their structure and English translation of individual elements.

2 MARITIME DOCUMENTARY SOURCES OF WEATHER, STATE OF SEA SURFACE, AND SEA ICE

A substantial part of documents analysed in this work were identified, digitized, and made available for this study by the RECLAIM project (Wilkinson, 2016a). A collection of various documents from the Southern Ocean voyages presently includes some 2700 images originating from various national and university archives (Wilkinson, 2016a, 2016b; Wilkinson & Wilkinson, 2018). This study focuses on the period of 1929–1940 that was associated with a phase of intense whaling activity in the area (Jackson, 1978; Tønnessen & Johnsen, 1982) and which left behind an extensive number of study-relevant documentary sources. This includes a collection of logbooks from whaling companies ‘Thor Dahl AS’ and ‘Pelagos AS’ from Tønsberg, Norway archived in the Vestfold archive and The Whaling museum in Sandefjord, Norway, which in the 20th century became a major centre of Norwegian whaling industry. Other major sources of documents used in this study are the accounts of whaling companies ‘Hektoria Ltd’, ‘Star Whaling Co Ltd’. and ‘United Whalers Ltd.’. of London, UK, ‘The South Georgia Co Ltd’, ‘Sevilla Whaling Co Ltd’ of Dublin, Ireland. The latter documents despite their association with British/Irish companies were, with a few exceptions, of Norwegian layout/standard and written in Norwegian as vessel crews would be hired in Norway. This collection of documents is stored in the archive of Sea Mammals Research Units (SMRU) at the University of St. Andrews in the UK. A few relevant documents were also recovered in the library of the Norwegian Polar Institute, Tromsø, Norway. Some documents comprising this database were published earlier as data sets (Divine, 2019; Divine et al., 2021, 2022). Documents from the company ‘Thor Dahl AS’ were keyed and analysed earlier (Bjørge, 2014) and the data for that study are published in (Bjørge et al., 2023).

The four types of documents were processed and analysed throughout this study.

1. Chief officer/Deck logbook (Norw. ‘Dæksdagbok’).
2. Catch logbook (Norw. ‘Fangstdagbok’).
3. Inspectors logbook/Notebook
4. Meteorological journal

Table 1 presents the list of documents analysed with the general details on the recovered data. All recovered vessel positions with meteorological and sea ice observations are shown in Figure 1. The whaling season in the Southern Ocean would typically start between mid-October and early November and continue until March to mid-April of the following year. The length of the season and hence data coverage in this part of the Southern Ocean varied therefore between about 150 and 180 days. The total number of meteorological and sea ice observations for the 14 whaling seasons recovered from these data sources and pooled for each vessel is further summarized in Table 2.

TABLE 1. Logbooks and documents from various vessels used in recovering the information on sea ice, icebergs, and weather conditions in the Southern Ocean for the spring to early fall periods of 1929–1940.

Doc. N	Vessel name	Keyed period of navigation in the SO	Document type	Document source	Data recovered
1	RV Norvegia*	05.10.1929–15.03.1930	LB	NPI	SI
2	RV Norvegia**	20.10.1930–09.02.1931	NB	NPI	SI
3	RV Norvegia**	05.10.1930–11.04.1931	LB	VA	SI
4	FS Antarctic	09.01.1930–16.03.1930	LB	VA	W, SI
5	FS Antarctic	29.09.1930–11.04.1931	LB	VA	W, SI
6	FS Solglimt	23.10.1932–19.03.1933	CB	VA	W, SI
7	FS Solglimt	26.11.1935–08.03.1936	CB	VA	W, SST
8	FS Solglimt	29.11.1936–07.03.1937	CB	VA	W, SI
9	FS Solglimt	19.11.1937–16.03.1938	CB	VA	W, SI
10	FS Ole Wegger	23.11.1935–27.02.1936	CB	VA	W, SI
11	FS Ole Wegger	08.12.1936–08.03.1937	CB	VA	W, SI
12	FS Ole Wegger	16.11.1937–15.03.1938	CB	VA	W, SI
13	FS Ole Wegger	18.11.1938–07.03.1939	CB	VA	W, SI
14	FS Thorshammer	05.11.1934–01.04.1935	CB	SMRU	W, SI
15	FS Thorshammer	28.11.1935–05.03.1936	CB	VA	W, SI
16	FS Thorshammer	02.12.1936–08.03.1937	CB	VA	W, SI
17	FS Thorshammer	02.12.1937–15.03.1938	CB	VA	W, SI
18	FS Thorshammer	17.11.1938–07.03.1939	CB	VA	W, SI
19	FS Svend Foyn	10.10.1932–21.04.1933	MJ	NPI	W, SST, SI
20	FS Svend Foyn	16.12.1935–29.03.1936	NB	SMRU	W, SI
21	FS Svend Foyn	10.11.1936–27.03.1937	NB	SMRU	W, SST
22	FS Svend Foyn	13.11.1937–26.03.1938	NB	SMRU	W, SI
23	FS Svend Foyn	01.12.1938–08.03.1939	CB	SMRU	W, SI
24	FS Svend Foyn	19.11.1938–20.03.1939	NB	SMRU	W, SI
25	FS New Sevilla	26.11.1934–31.03.1935	CB	SMRU	W, SI
26	FS New Sevilla	28.11.1936–20.03.1937	NB	SMRU	W, SST
27	FS New Sevilla	30.10.1936–06.03.1937	CB	SMRU	W, SI
28	FS Hektoria	21.11.1935–24.03.1936	NB	SMRU	W, SST
29	FS Hektoria	08.12.1936–07.03.1937	CB	SMRU	W, SI
30	FS Hektoria	25.11.1937–26.03.1938	CB	SMRU	W, SST
31	FS Hektoria	20.11.1938–07.03.1939	CB	SMRU	W, SI
32	FS Hektoria	17.12.1939–17.03.1940	CB	SMRU	W, SI
33	FS Terje Viken	12.11.1936–14.03.1937	NB	SMRU	W, SST
34	FS Terje Viken	23.11.1936–07.03.1937	CB	SMRU	W, SI
35	FS Terje Viken	24.11.1937–15.03.1938	CB	SMRU	W, SI
36	FS Terje Viken	05.12.1938–07.03.1939	CB	SMRU	W, SI
37	FS Terje Viken	20.12.1939–17.03.1940	CB	SMRU	W, SI
38	FS Tafelberg	05.12.1936–13.03.1937	NB	SMRU	W, SST
39	FS Salvestria	26.11.1936–07.03.1937	CB	SMRU	W, SI

• Table 1 indicates the logbook periods when vessels were in the waters of the SO. Notations ‘RV’ and ‘FS’ used stand for ‘Research vessel’ and ‘Whale factory ship’, respectively. Following notations used for the types of documents used. NB: Expedition notes or Inspector's notebook/log; LB: Logbook (‘Dagbok’); CB: Catch logbook (‘Fangstdagbok’); MJ—Meteorological journal. Stars ‘*’ and ‘**’ denote the third and the fourth RV Norvegia expedition to the SO, respectively. Sources of the documents: NPI—NPI library, Tromsø, Norway; VA—Vestfold Archive, Sandefjord, Norway; SMRU—Sea Mammal Research Unit—University of St. Andrews, UK. Information recovered from the logbooks denoted as ‘W’ for weather and sea state, ‘SST’ for sea surface temperature and ‘SI’ for sea ice and/or icebergs. Detailed archival references to the presented documents are found in the respective data sets.

FIGURE 1

Daily vessel positions for days with recovered observations from 1929 to 1940 grouped by months. Vessel positions for October and April are merged for convenience with the data from the neighbouring months of November and March, respectively.

TABLE 2. Number of meteorological and sea ice observations from the Southern Ocean over the period of 1929–1940 recovered from all keyed documents (‘Number docs’.) of all types.

Vessel	Number docs.	Met. Observations	Sea ice observations
RV Norvegia	3	247	247
FS Antarctic	2	1570	292
FS Solglimt	4	469	321
FS Ole Wegger	4	410	406
FS Thorshammer	5	559	463
FS Svend Foyn	6	2479	1799
FS New Sevilla	3	291	117
FS Hektoria	5	785	155
FS Terje Viken	5	653	91
FS Tafelberg	1	99	19
FS Salvestria	1	102	36
Total	39	7955	3946

• Note: ‘Sea ice observations’ also include cases of open sea/open water observations in the areas where the presence of sea ice was still possible.

The four types of accounts differ in the structure and frequency/abundance of relevant observations. Norwegian deck logbook alike nautical documents of similar type from other nations was a major onboard document intended for recording all kinds of events important for navigation, operation, and management of a ship. This would include a detailed record of weather and sea state conditions. Norwegian catch logbooks in turn were primarily intended to log the information on the capture and processing of whales. However, they also contain daily meteorological, oceanographic, and ice observations though the abundance of data recorded may vary from book to book.

The collection of documents imaged by the RECLAIM project also comprises a number of notebooks and inspector's diaries that could be filled out by, for example, company inspector onboard a factory ship. This is another valuable source of weather information, containing among other things measurements of sea surface temperature and notes on sea ice. These are essentially complementary yet important data sources. Notebooks can be used to infer, for example, if sea ice could potentially be present at the location of a factory ship, if this information was not specified in the respective catch book.

This study also presents some data from a dedicated journal of meteorological observations. This document belongs to a collection of more than 50 meteorological journals covering seven whaling seasons in during 1932–1939. These journals are a legacy of an opportunistic use of commercial whaling for establishing a meteorological observational network in Antarctica. This initiative emerged during preparations for The Second International Polar Year (1932–1933) and at the start of the initiative involved 10 Norwegian factory ships (Risting, 1932).

Typically, weather observations in the deck logbooks are reported on a 4-hourly basis, catch books provide such information once a day and the analysed meteorological journal reports 3-hourly observations. The essential set of observed parameters together with positional information and time would typically include air temperature, wind conditions, barometric pressure at or close to sea level, sea state, and a general description of weather. Some documents also contain information on sea surface temperature, sea ice, and icebergs. Catch logbooks represent the most common type of logbook in the analysed collection. While notebooks follow a free format of data logging, they were essentially used as complementary data sources for the other types of logbooks. Specific details on the format/layout of these types of documents specific to the focus period are presented in Supplementary materials.

We note that meteorological data from the third expedition of RV Norwegia in 1929–1930, Doc. 1 in Table 1, were published earlier along with other scientific results in Mosby (1933) and present in the ICOADS archive (Freeman et al., 2017). Search in the ICOADS archive also revealed that it also contains some of the meteorological data from the meteorological journal of FS Svend Foyn (Doc. 19), as a part of the data collection from Norwegian Antarctic Whaling Factory Ships (1932–1939) (Freeman et al., 2017, deck 188). These observations were digitized earlier, likely in the 1970s, and still on punch cards, and require both modern quality control and infilling of numerous data gaps. Due to time constraints, we in this study focused on core data only such as surface air and sea surface temperature (SAT and SST, respectively), barometric pressure (SLP), wind force/direction, and sea ice, while keying respective meteorological data is retained for future work.

3 OBSERVATIONS AND DATA FORMAT

3.1 Time and position

Time and position information as well as the associated uncertainties are of key importance for a correct interpretation and further use of recovered weather and sea ice data. Norwegian maritime authority (Norw. ‘Sjøfartsdirektoratt’), formerly known as ‘Sjøfartskontoret’, based on legal framework of National admiralty law (Norw. ‘Sjøfartsloven’, ‘Sjøloven’) regulated the types, number, and quality of instruments for navigation and weather observations that had to be available onboard (e.g. Sjøfartskontoret [The Maritime Office], 1915).

At least one chronometer set to a GMT time was supposed to be available onboard for timekeeping and vessel positioning. With the evolution of a network of dedicated time signal broadcasting stations the time keeping was further assisted by radiotransmitted time signals (e.g. Johnsen & Bryn, 1942). Chronometers as well as other instruments necessary for positioning were subject to regular inspections onboard as well as in authorized public offices while onshore, where their accuracy and precision would be checked and instruments calibrated if found necessary (Sjøfartskontoret [The Maritime Office], 1915).

Inspection of deck logbooks, including instructions for bookkeeping at the preface of each deck logbook, does not specify what time was used as a standard/default throughout the logbooks. Apparently, crew shifts, meal times, and other onboard activities had to, as a common practice, follow local (‘ships’ or ‘deck’) time, which in turn was linked to daylight conditions and hence the meridional position of the vessel. The associated routines for ships' clock adjustments described in Johnsen and Bryn (1942), notes made on ships' clock adjustment events together with logged offsets from the onboard chronometer found in some of the studied documents, suggest the times recorded in both deck logbooks and catch book entries followed local/deck time. Therefore, we consider that unless the time GMT was specifically indicated for a particular document or an observation (observation series), the onboard (ships/deck) time was used in the document.

By the end of the 1930s the use of zonal time on Norwegian ships was not yet ubiquitous and time adjustments for a fraction of hour depending on the actual ships longitudinal position could be applied instead (Johnsen & Bryn, 1942). We found and registered notes on this kind of fraction of an-hour time adjustments in some of the studied documents, but which of the two methods of deck timekeeping was preferred in each particular case for the remaining documents was not possible to establish with any certainty. It is therefore assumed for simplicity that the times recorded in logbook/notebook entries, unless specified, followed the local time zone corresponding to time zone, the ship was positioned at the time of the observation. This local deck time would be changed when the ship passed from one time zone to another.

For data submission, the original recorded time entries of the analysed documents were kept. Respective conversion to GMT (unless was already logged in GMT) can be conducted following the procedure outlined in Teleti et al. (2019) which uses the ship's longitudinal position and the associated time zone.

As a common practice for that period, in case of favourable weather conditions the ship's position would be observed once a day at apparent noon local time and logged with necessary corrections for a time offset at noon deck time (e.g. Johnsen & Bryn, 1942). Ships coordinates were also regularly calculated using dead reckoning of the vessel-bearing, speed, and drift over the preceding 24-h period. This would provide alternative positional information for days when solar or celestial observations were not possible. As a common practice on Norwegian vessels, the positional observations were to be taken with reference to true north rather than magnetic north. We note that regular logging of deviations from magnetic compass as well as results of control of onboard chronometers was supposed to be logged in a special ‘Compass and chronometer logbook’ (Johnsen & Bryn, 1942). The analysis of these documents was, however, outside the scope of this study.

While deck logbooks contain both types of coordinates, catch books feature only a single pair of columns for noting ship position (see Supplementary materials) without specifying if this is a position based on direct observations or a calculated one. In some of the analysed notebooks, specific remarks were made if coordinates were obtained from dead reckoning or celestial observations. As coordinates based on direct observations are expected to be more accurate, it is not unreasonable to assume that whenever observations-based coordinates were available, they would be logged in a catch logbook.

During the keying of the data set, both types of coordinates if present were logged. Typically, whenever available, the observation-based position was given a preference. Originally, latitudinal, and longitudinal position was recorded in degrees and minutes of four cardinal directions. These were further converted into the degree and decimal system to facilitate further data processing and analysis.

3.2 Meteorological observations

Logging weather information on Norwegian vessels during the period covered by this study followed a system presented on the front pages of the logbooks. This was subject to modifications/improvements throughout the studied period. The dedicated page with an explanation of notations and acronyms (Norw. ‘Tegnforklaringer’) at the preface of each logbook (see Figure 2) presents standardized tables with various scales/categories for various weather phenomena that were used for logging observations. Though these tables are not present in catch logbooks, analysis of these documents indicated that recording weather observations in this type of document followed the same system. Tables 3, 4 and 5 for the three main scales/categories of weather and sea state are shown below along with English translations. Some of the terms used are not from the modern Norwegian language, being closer to Danish, same as the language used for logging various textual information in the analysed logbooks.

FIGURE 2

Image of one of the preface pages of the deck logbook presenting notations and acronyms (Norw. ‘Tegnforklaringer’) along with standardized tables with various scales/categories used for logging weather and sea surface state observations. This image belongs to the logbook of RV Norvegia, Doc. 1, and follows the year 1926 layout/format typical for other reviewed logbooks of the analysed period.

TABLE 3. Wind force (Norwegian ‘Vindstyrke’) categories based on a 7-point wind force scale used on Norwegian vessels during the studied period (Figure 2) together with suggested conversion to the SI units following Russeltvedt (1920) and the Beaufort wind force scale.

Category (Norw. ‘Tegn’)	Notation (Norw. ‘Betydning’)	English translation	Wind speed in m/sec	Wind force, Beaufort number
0	Stille	Quiet or still	0–0.5	0
1	Svak	Gentle/Weak	0.5–5	1–3 (1–2)
2	Lett	Easy/Light	5–9	3–5 (3–4)
3	Frisk	Fresh	9–13	5–6
4	Sterk	Strong	13–17	6–7 (7)
5	Storm	Storm	17–28	8–10
6	Orkan	Hurricane	Over 28	11–12

• Note: Figures in bold in the right column indicate the most suitable Beaufort numbers to be used when conversion is ambiguous due to overlapping ranges of wind speeds between the two scale tables.

TABLE 4. Weather phenomena (Norw. ‘Været’, see Figure 2) and state of the sky indices/categories used in the Norwegian logbooks of the studied period.

Category (Norw. ‘Tegn’)	Notation (Norw. ‘Betydning’)	English translation	Cloud cover, octas
a	Klart	Clear	0/8–1/8
b	Lett-skyet	Partly cloudy	2/8–3/8
c	Halvklart	Half cloudy	4/8–5/8
d	Skyet	Cloudy	6/8–7/8
e	Overskyet	Overcast	8/8
f	Meget mørkt og truende	Very dark and threatening
g	Byget	Showers (snow, rain)
h	Disig	Hazy
i	Tåket	Foggy
k	Regn	Rain
l	Sne	Snow
m	Torden	Thunder

• Note: Approximate conversion of cloud cover to scale in octas (eights) is also provided.

TABLE 5. Sea surface state scale/height of waves (Norw. ‘Sjøgang’) used in the Norwegian logbooks of the period.

Category (Norw. ‘Tegn’)	Notation (Norw. ‘Betydning’)	English translation
0	Stille	Quiet/calm/still
1	Svak	Weak/smooth
2	Lett	Easy, slight
3	Frisk	Fresh/moderate
4	Sterk	Strong/rough/very rough
5	Svær	Severe/high
6	Vældig	Very high/phenomenal

• Note: Conversion rule to the modern sea state scale is not provided.

All relevant information had to be logged according to deck logbook keeping instructions (Norw. ‘Anvisning til utfylling av rubrikkene i dagboken’) found in the first pages of deck logbooks. The sections associated with observation and recording of meteorological parameters state explicitly that a particular observation would be logged in the line of the logbook corresponding to the time this observation was made. For analysed deck logbooks all weather observations are reported every 4 hours (04-08-12-16-20-24 LT), while for catch logbooks the air temperature and barometric pressure were registered once a day per 08:00 LT. Other weather and sea state variables registered in catch logbooks had to reflect respective average/prevalent observations registered throughout the day.

For conducting navigation-relevant weather observations each vessel had to be equipped with at least one barometer and one thermometer for measuring air pressure and air temperature (Sjøfartskontoret [The Maritime Office], 1915). Vessels that operated in waters where sea ice and icebergs could be encountered had to be equipped in addition with a dedicated thermometer for measuring sea surface temperature. All these instruments were subject to regular inspections in authorized public offices where their accuracy and precision would be checked and calibrated if found necessary (Sjøfartskontoret [The Maritime Office], 1915).

Typically, a gimbal-mounted marine mercury barometer (Norw. ‘Sjøbarometer’) equipped with a thermometer for measuring ambient air temperature and barometer readings correction would be used for air pressure observations during the focus period (Bryn, 1938). Since mercury barometers were relatively inertial, for observing rapidly changing atmospheric conditions, a parallel use of more sensitive though less accurate aneroid barometers was also recommended (Bryn, 1938). For air temperature measurements, a sling thermometer or Assmann aspiration psychrometer was typically used (Bryn, 1938).

During the focus period, no specific system of units was introduced as a standard for measuring and reporting atmospheric pressure and temperatures. The analysed documents were found to report these parameters in both inches and millimetres of the mercury column, same as temperatures could be logged in degrees Celsius, Fahrenheit, or even in the Kelvin scale. Apparently, the units used in the logbooks depended on the instruments available onboard. To bring uniformity to the observations in this dataset, a standard formula was used to convert mercury column inches and millimetres into hecto-Pascals (hPa or millibars); values of air and sea water temperatures where needed were converted to degrees Celsius.

The vessels Ole Wegger, Hektoria, Solglimt, Svend Foyn, and Thorshammer before the start of the whaling season of 1932/1933 were equipped with similar sets of instruments complying with standards for maritime meteorological observations of that period (Risting, 1932). This would comprise a mercury barometer and Assmann aspiration psychrometer and thermometer with a precision better than 0.1°C. While barometers were placed in vessels' radio rooms, thermometers were hung in/outside the chart house. The crew received the necessary training for conducting meteorological observations from the Norwegian Meteorological Institute.

Raw barometer readings require corrections for mercury cistern temperature, latitude, elevation over the sea surface, and eventually scale displacements to be introduced for the observed pressure to be representative of the local air pressure at the surface. Without these corrections deviations in the order of a few mm of mercury column are possible. Bryn (1938) (see pp. 24–26) provides detailed instructions for this relatively straightforward procedure. Moreover, some instruments were already equipped with brass slide rules for adjustments for latitude and elevation to be introduced directly on the barometer. Bryn (1938) notes, however, that corrected values at sea are not as important (or rather ‘of interest’) as observations of relative changes in raw barometer readings between the measurements. Making these corrections was only indicated necessary when measured air pressure had to be compared with pilot charts, observations transmitted by radio from/to other vessels and land stations, as well as weather forecasts.

Deck logbook keeping instructions dated to 1931 (also shown in Johnsen & Bryn, 1942) do not mention corrections to measured barometric pressure. Similar instructions from 1951 (e.g. Johnsen & Bryn, 1952) already mention this procedure explicitly which, however, remains ‘recommended’ or ‘preferred’ rather than mandatory. We conclude therefore that with a lack of information on the models/types of instruments used, and most importantly, onboard routines common in each particular case, it cannot be established with any certainty whether the logged barometric readings were corrected or not. The only exception is a meteorological journal of FS Svend Foyn (Doc.19) where dedicated columns for mercury cistern temperature and the calculated corrected sea level pressure were provided (see Supplementary materials S1.4). However, one should note that for the region of interest, the barometric pressure corrections for ambient temperature and latitudinal normal gravity changes have opposite signs and hence partly compensate each other.

Wind direction had to be logged as a point of the true compass from which wind blows (Johnsen & Bryn, 1942) and observed to the nearest true compass point. A 16- or 32-point compass rose for indicating wind direction was used throughout the analysed documents. Wind force in some of the analysed logbooks was logged using an older 0–6 wind force scale (see Figure 2, table ‘Vindstyrke’) and notated using either text or a respective windforce number. Around 1931, the standard for logging wind force on Norwegian vessels changed: instead of an older 7-point wind force scale, the 12-point Beaufort scale is introduced (see Johnsen & Bryn, 1942, pp. 134–135; compare with front pages of Documents 1, 4, and 5 that represent logbooks of the year 1926 format). However, the use of the formally obsolete 7-point wind force scale was still common throughout the analysed documents.

For conversion of the 7-point wind force scale to SI units and the Beaufort wind scale, we refer to Russeltvedt (1920) (see respective conversion table on page 22). The naming conventions are identical to wind force categories in the deck logbook preface page, except for category 2 ‘light’, (Norw. ‘lett’) which is a synonym to ‘laber’ in Norwegian. It is remarkable that the 7-point wind force scale is referred to as a ‘Land scale’ (Norw. ‘Land-skalaen’), while for maritime applications, the Beaufort wind force scale is provided. Table 3 summarizes the derived conversion rules.

Finally, for various weather phenomena and cloud cover, a code-based system was used (see Figure 2 and ‘Været’ in Table 4). Similar to wind force, information could still be logged using a text note rather than the respective weather code.

3.3 Sea surface state, temperature, sea ice, and icebergs observations

Information on sea surface state included wave/swell height and direction and occasionally the presence of sea ice and icebergs. Similar to the wind force scale, around 1931, a 10-point sea state code similar to the Douglas sea scale was introduced to replace an older 7-point sea state code. However, we note that an older scheme was used throughout some of the analysed documents (Table 5).

Several documents provide measurements of sea surface temperature. According to (Risting, 1932), the factory vessels Ole Wegger, Hektoria, Solglimt, Svend Foyn, and Thorshammer were equipped with marine thermometers by Kühler & Söhne mounted at the engine room intake, inside the sea water pumping centrifuge. This enabled measurements of the temperature of the seawater pumped to cool the engines with a precision better than 0.2°C. More details on the measurement system including the inlet depth that may have implication for the accuracy/bias of the inferred sea surface temperature (Kent et al., 2010) could not be found.

When a vessel was positioned within or near an ice pack, the respective note would typically be made in the logbook/catch book. The frequency of observations varies between the documents. For the analysed logbooks of Documents 2–5 sea ice conditions are reported every 4 h. For catch books, the observations would be logged on a daily basis. For the period considered, a system for logging sea ice conditions was not yet well established on Norwegian vessels. The notes on sea ice as well as the observational practices are therefore not standardized and tend to vary between the vessels/observers.

Areas where sea ice could have been expected but where there was open water were typically logged as ‘no ice’ or ‘open sea’. Often, the open water state could also be recovered by references to a rough sea state and a vessel rolling. For the deck logbook as shown in Figure S1, the notes on sea ice are found on the left-hand side of the double-facing page, registered after the information on weather and sea surface state. In the example image, a note ‘beliggende i isen, ingen drift’ (English ‘positioned in the ice, no drift’) followed by ‘Do’ (abbreviation for «ditto» or «det samme» meaning ‘the same’ in English) below indicates a vessel located and drifting in pack ice throughout the time period covered by the page. In catch logbooks sea ice/open water and icebergs would be reported on the right-hand side at the end of the line for weather conditions (see Figure S2), typically, in a form of a short note on both sea ice and/or icebergs. In the meteorological journal shown in Figure S4, the project-relevant sea ice and iceberg notes are logged on the right-hand side, the rightmost column of the double-facing page of the journal. For this journal, the notes have typically a standard format and can be categorized into a few states of sea ice cover. Notes found in the example page ‘Under gang i isen’ (English ‘steaming in the ice’) and ‘under gang i iskanten’ (English ‘steaming along the ice edge’) suggest disambiguously the vessel's location in ice-covered waters.

During mid- to late-1930s, a more explicit/systematic category-based system of logging the information on sea ice and icebergs emerged. Figure 3 shows sea ice and icebergs classification table (Norw. ‘Isbetegnelser’) introduced as an addendum to item 4 of the instruction to catch logbook keeping.

FIGURE 3

Image of an insert from catch logbook of FS Thorshammer for the whaling season 1938/1939 (Doc. 18) showing sea ice and icebergs classification system for logging sea ice information. The respective English translation is presented in Table 6.

Classification of sea ice in this system integrates both ice coverage (ice concentration) and stages of ice development as given in Table 6, and with some reservations enables conversion to elements of the presently used WMO sea-ice nomenclature (JCOMM Expert Team on Sea Ice, 2015). In total, 14 of the analysed documents, namely, Doc. 7–13, 15–18, 23, 36, and 37 categorized sea ice and icebergs conditions using this letter code-based system. Though the recovered information on the state of sea ice is not properly or systematically categorized for the remaining documents, one should note that taking the vessel types, dimensions as well as their operational conditions (whale hunting) into consideration, in the majority of cases the logged sea ice notes should be associated with a relatively loose/open ice pack. As hunting for whales was mainly conducted from relatively small steam-engine powered wooden vessels with displacement below 400–500 tonnes, their capability for navigation in the areas with higher fractional coverage of sea ice would be very limited. This state of ice cover would correspond to the categories of ‘very open’ to ‘open drift ice’, that is, ice concentration of 1/10–3/10 and ice concentration of 4/10–6/10, respectively, according to the WMO sea-ice nomenclature (JCOMM Expert Team on Sea Ice, 2015).

TABLE 6. Category-based system of logging the information on sea ice and icebergs (Norw. ‘Isbetegnelser’) that was introduced in mid to late 1930s on Norwegian vessels; designations used shown with their respective English translation.

Category	Notation in Norwegian	English translation
A	Isfritt	Ice free/Open sea
B	Drivis	Drift ice (loose/very open)
C	Slak pakkis	Loose/open-pack ice
D	Slak pakkis med klarer	Loose/open-pack ice with breaks/leads
E	Tett pakkis med klarer	Close pack ice with breaks/leads
F	Sammenhengende pakkis uten klarer	Very close or consolidated pack ice
G	Nyis, (sprøis eller tallerkenis)	New ice (brittle ice or pancake ice)
H	Kokeriet ligger utenfor, men i sikte av pakkis. Der tilføes s (syd), n (nord), w (vest) eller f.eks. so (sydost), peileretningen til ispakkene, idet man søker å slutte sig til hvilken hovedretning pakkene har	Ship is positioned outside, but within sight of pack ice. Main bearing direction to ice pack is indicated by S (South), North (N), West (W) or e.g. SE (South West)
I	Mange isfjell	Numerous icebergs
J	Adskillige isfjell	Several icebergs
K	Enkelte isfjell, (meget få)	Solitary icebergs (very few)

For example, the logbook from FS Antarctic from 27.03.1931 (Doc. 5) documents the situation with two whaling boats stuck in the consolidated ice: ‘…maatte la gaa 1 Finhval …da isen var tetnet saa der var vanskelig at komme ut’. (approximate English translation: ‘…let one fin whale to escape as ice became so consolidated that it would be difficult to steam out of the pack’.). Later in the afternoon, the logbook reports on the manoeuvring of the factory in ice pack to release the hunting vessel ‘Leslie’: ‘…manavrede in i isen for at hjelpe Leslie ut’. (approximate English translation: ‘…navigated in ice pack to help “Leslie” go out of the ice’). It suggests that the observed/logged sea ice can broadly fall into four categories equivalent to the modern WMO Sea-Ice Nomenclature in terms of sea ice concentration (or areal sea ice coverage) and a category of open sea. Table 7 summarizes the proposed rules that can be applied to convert the textual information as well as category-based sea ice classification scheme into sea ice concentration. Note that ice edge is associated with ice concentration of 15%, in accordance with the WMO nomenclature. A list of typical notes related with the state of ice cover or the vessels navigation in sea ice is presented in Table 8 below.

TABLE 7. WMO classification of sea ice cover with respect to fractional coverage of the sea surface.

WMO ice concentration category	Sea ice fractional coverage
Open water	No sea ice or sea ice at a concentration below 0.1 (10%)
Very open (drift) ice	Ice with concentration 0.1–0.3 (10%–30%)
Open (drift) ice	Ice with concentration 0.4–0.6 (40%–60%)
Close (pack, drift) ice	Ice with concentration >0.7 (>70%)
Sea ice edge	Ice concentration 0.15 (15%)

TABLE 8. Typical notes on sea ice/icebergs and a vessel navigation in or near sea ice pack as found in the analysed logs together with the respective English translations.

Doc. N	Sea ice note in Norwegian	English translation
19	Skibet gaar öst langs iskanten	Vessel heading East along ice edge
19	Skibet gaar öst gjennom enkelte ispakker	Vessel heading East through loose ice patches
19	Skibet gaar östover gjennom enkelte ispakker	Vessel heading East through loose ice patches
19	Gaar langs iskanten östover	Vessel heading East along ice edge
19,5	I drift ved (utenfor) iskanten	Vessel is adrift at (outside) ice edge
19	Gaar gjennom store mangda is	Vessel going through heavy/close pack ice
19	Driver i isen	Vessel adrift in the ice
M	Ligger i isen og driver	Vessel adrift in the ice
19	Drift i isen	Vessel adrift in the ice
M	Beliggende i isen/ligger i isen	Vessel (adrift) in ice pack
19	Beliggende ved iskanten	Vessel (adrift) at ice edge
19	Enkelte isfjell	Separate (solitary) icebergs
M	Gik i pakisens ytterkant	Sailed along the edge of pack ice
M	Gik i slak pakis	Sailed in very open (loose) drift ice
M	Spredte isfjeld	Scattered icebergs
M	Mange (masse) isfjeld	Numerous icebergs

• Note: The list is not comprehensive, as at that time a common routine of logging sea ice information was not yet established. ‘M’ stands for several documents with this type of notes.

4 RESULTS

4.1 Recovered data summary

The keyed data were initially logged in Excel spreadsheets of a broadly common format (see Supplementary materials for the workbook data structure) that partly reproduced the layout and content of the logbooks and notebooks. This simplified subsequent data crosschecks both between the vessels and with the original documents. Vessel- and voyage- related relevant metadata, whenever available in the document were also logged. Both the original notes in Norwegian and their English translations were logged in the files.

One should note that attempts were made to apply some of the available OCR (Optical Character Recognition) freeware to speed up the keying of the raw logbook images. Hindered by cursive writing with writing styles and abbreviations varying between the documents, results were not found satisfactory enough even with a recent progress made in developing these techniques. In view of all necessary labor efforts required for manual post-editing comparable with time required for a direct keying of the data from the imaged sources, a manual data extraction was preferred.

Figure 1 shows all daily vessels positions for days with observations for the months between October and April. In total we extracted about 8000 unique weather records and 4000 notes on sea ice that altogether makes 35 vessel-seasons between Austral fall of the 1929 and spring of the 1940. In order to highlight the temporal variability in the spatial distribution of recovered observations they were grouped by months. Due to a relatively low number of observations for October and April, they were merged for convenience with data from the neighbouring months of November and March, respectively. Majority of the recovered daily positions are for the areas of the Atlantic and Indian oceans sector of the Southern Ocean, though a few observations from the Pacific sector of the Southern Ocean are available as well. A progressive shift in vessels' positions southwards towards the continent throughout the Austral summer is indicative of a general tendency to relocate factory vessels south along with whales' migration following a gradual recession of the edge of summer ice pack. However, whale catches made outside the ice pack could still constitute a substantial part of the total seasonal catch (Bjørge, 2014).

4.2 Data quality control, homogenization, and standardization

Prior to publication data were subject to quality control. Ship tracks were inspected for the presence of anomalous positions of individual points or track segments that could emerge due to keying errors. Points of concern were further cross-checked with the original logbooks. Air, sea surface temperatures, and barometric pressure series were inspected for possible outliers. Suspicious observations were cross-checked with the original logbook images. For wind force and swell the distributions of values keyed for each particular document were checked for agreement with the inferred scales for these variables.

Recovered data were published in two formats. One format is close to the keyed original and may have features in a way of logging the project-relevant information unique for each particular document. We also retained relevant original notes and notations/abbreviations used when the observations were logged. This data format is however neither convenient for further statistical analysis nor compliant with general standards and requirements for modern electronic scientific data services and storage.

In order for the recovered data to meet the criteria for data ingestion by major climate data archives, for example, ICOADS or Copernicus Climate Data Store, the data were further homogenized and standardized to a common format where only weather/sea ice relevant information together with time and positional information were retained. Each document was allocated in an individual data file with a spreadsheet for the data and a dedicated spreadsheet with the metadata. Metadata contained the information crucial for the data interpretation and included both vessel-specific metadata, such as details on ship's identification, dimensions, etc, as well as observations-specific metadata. The latter provides details on the onboard instrumentation used for making observations as well as units of measurements. Note that for such variables as temperature, barometric pressure, and wind force we kept for most documents the original measurement units, rather providing necessary conversion factors to SI units in the metadata. This should help to reduce the potential for emergence of any data processing-associated biases later, such as, double-rounding effects (Rhines et al., 2015). Data from (Bjørge et al., 2023) were, however, already converted into SI units earlier and were kept as is after cross-checking with original imaged documents. The final data structure is presented further in Supplementary materials.

4.3 Summary of the inferred sea ice concentration from the analysed documents

Keyed textual notes on sea ice and ice conditions codes were converted into sea ice concentrations based on the proposed classification scheme (see Table 9). These inferred sea ice concentrations were included only in the files with the data in their original format. Note that the data covers the months from earliest October to latest April, thereby nearly encompassing the months of maximum and minimum sea ice extent that according to modern satellite observations may occur in the Southern Ocean during September to October (maximum) and February to March (minimum, Parkinson, 2014).

TABLE 9. Proposed conversion rules between narrative and category-based sea ice and iceberg information to sea ice categories of the WMO nomenclature.

Relevant narrative information in the logbooks	Sea ice category as in Table 6	Inferred SIC	SIC according to WMO nomenclature
(A) Vessel reported to be adrift or steam in open water; (B) No information on ice conditions provided, and/or rough sea and strong/violent vessel rolling is reported	A	0	Open water
Vessel reported to be adrift or steaming in the vicinity of the sea ice edge	H	0.15 (15%)	Very open (drift) ice
Vessel travels in waters with solitary floes or very loose/open-pack ice	B, G	0.1 (10%)	Very open (drift) ice
Vessel adrift, or reported to steam with variable courses in pack ice	C, D	0.30 (30%)	Very open (drift) ice
Vessel adrift, or reported to steam in heavy pack ice	E	0.4–0.6 (40%–60%)	Open (drift) ice
Vessel navigation was reported to be seriously hindered by pack ice (see 3.2. for an example of such report)	F	0.7 (70%) and higher	Close—Very close—Consolidated—Compact (drift) ice

• Note: SIC stands for sea ice concentration.

Due to the subjectivity of the proposed methodology, users should take into consideration potential uncertainties in the proposed values of sea ice concentration. Though providing exact uncertainty bounds in each particular case may not be possible, we suggest a typical width of the Marginal Ice Zone in the region, expressed as the area of sea ice with concentration within the 15%–80% range, to be a realistic upper bound on the uncertainty estimate. The width of this zone varies driven by changing local winds and ocean currents and can reach up to 200 km wide (Stroeve et al., 2016), that is, up to 2° latitude. This uncertainty expressed in distance translates into a vessel position and can subsequently be used when making estimates of local sea ice extent.

We use the months of minimum and maximum modern ice extent as examples to demonstrate the potential of the recovered sea ice data for the reconstruction of past sea ice conditions. However, as October data are not abundant in this study, we rather focus on November as a month when sea ice extent is still close to its seasonal maximum. In order to place the recovered sea ice and open water observations in the context of contemporary sea ice extent in the region, they are shown together with the SSMI passive microwave satellite-based monthly median ice edges for the period of 1981–2010 (Fetterer et al., 2017).

Figure 4 presents inferred November sea ice concentrations/open sea pooled from ice reports for the entire data period. A large proportion of sea ice observations were made up to 4° latitude to the north of the modern median sea ice edge position. This does not rule out a slightly more northward position of winter sea ice extent in the region for the period analysed here. However, comparison with satellite data, in particular, an absolute observed seasonal sea ice maximum of September 2014, also suggests that positions of past sea ice observations, with the only exception found in the Indian sector of the Southern Ocean, are still located within the margins of the contemporary seasonal ice zone.

FIGURE 4

Compilation of positions for sea ice observations and sea ice edge for November, the month close to the seasonal sea ice maximum in the Southern Ocean, with inferred sea ice concentrations for all analysed documents over 1929–1940. Circles indicate vessels' positions and observed (inferred) sea ice concentrations. Light blue is for the 15% ice concentration (interpreted/reported as ice edge), while darker blue is used for the 30% ice concentration and higher. White dots indicate reported open sea conditions. Green line indicates median November sea ice edge positions (based on 15% ice concentration) from NSIDC SSMI passive microwave satellite-based observations for the period 1981–2010. The monthly mean sea ice concentration during the sea ice maximum of September 2014 registered over the period of satellite observations is shown by blue shading.

Figure 5 shows all recovered February observations of sea ice for the studied period. Compilation of sea ice reports also suggests a potentially higher past summer sea ice extent of that period in the region relative to the median February sea ice edge position during 1981–2010 and certainly when compared with the recent SO sea ice minimum of February 2023 (Purich & Doddridge, 2023). The difference is most apparent in the Weddell Sea area where ice pack was present far away from the Antarctic shore which, on average, is clear of ice in February at present. In the Dronning Maud Land sector of the Southern Ocean sea ice was also observed up to 4° latitude to the north of its present median position. Nevertheless, similar to winter sea ice maximum, past sea ice observations for February still lie within the margins of seasonal sea ice extent over the satellite period and more data supported by a quantitative analysis of past and modern observations would be required for a more robust inference.

FIGURE 5

Compilation of positions for sea ice observations and sea ice edge for February, the month of seasonal sea ice minimum in the Southern Ocean, with inferred sea ice concentrations for all analysed documents over 1929–1940. The circles indicate the vessels position and observed (inferred) ice concentration. Light blue is for the 15% ice concentration (interpreted/reported as ice edge), while darker blue is used for the 30% ice concentration and higher. White dots indicate reported open sea conditions. Red line indicates median February sea ice edge position (based on 15% ice concentration) from SSMI passive microwave satellite-based observations for the period 1981–2010. Green highlights monthly mean sea ice extent during the recently observed February 2023 absolute regional sea ice minimum. For comparison, blue shading shows the monthly mean sea ice concentration for February 2014 when sea ice extent was closest to its seasonal maximum registered over the period of satellite observations.

We note that the recovered sea ice observations for the studied period are in line with past tendencies and variability in both total and regional sea ice extent reconstructed from a compilation and synthesis of various instrumental observations and climate indices (Fogt et al. (2022); Fogt et al. (2024)). The ice-core-based reconstructions of past sea ice in the region, though less coherent and mainly representative of regions other than the Atlantic and Indian sectors of the Southern Ocean, also hint at a possibly higher seasonal sea ice extent (e.g. Abram et al., 2010) for the period of the early 1930s.

5 CONCLUSIONS AND OUTLOOK FOR FUTURE WORK

The conducted study demonstrates the high potential of historical Norwegian maritime documentary sources such as logbooks from whaling ships and factory vessels to fill the knowledge gap on pre-satellite era sea ice and climate variability in the Southern Ocean. These results motivate us to continue exploring and exploiting available Norwegian documentary sources, further extending the study through the early 1900s and to the late 1960s. Data from the present study is published/held in the science data depository of the Norwegian Polar Institute and will also be submitted to Copernicus Climate Data Store with a perspective for their further assimilation by ICOADS and ECMWF Reanalysis.

While the importance and use of major recovered weather variables such as surface air temperature and sea level pressure for the region is discussed elsewhere, an important part of this initiative for the future would be a consistent recovery and logging of sea ice data. Sea ice is an important component of the climate system and is well-recognized as an efficient and relatively easy to observe climate indicator (Intergovernmental Panel on Climate Change [IPCC], 2022). A note on sea ice observation made at a specific location and time, even if no further information is provided, may already tell a lot about the state of the climate locally and even on the regional scale. In this study, we demonstrate that despite a lack of consistency/systematic logging of sea ice data during the first half of the 20th century, the observations when assessed critically can still be interpreted successfully in terms of modern sea ice nomenclature. Potentially, assimilation of such observations from different vessels may aid in reconstructing ‘snapshots’ of areal sea ice extent for the periods when relevant data density is sufficiently high, see for example, Divine and Dick (2006); Vinje (2001). Statistical analysis of already recovered sea ice observations in the context of modern sea ice extent and variability inferred from satellite-based sensors as well as a more specific comparison with already existing and future proxy-based reconstructions is also pending.

Extending the analysis to recover new and adding already existing weather and sea ice observations data from the 1920s and 1930s (e.g. from Teleti et al., 2019) will provide an information on decadal-scale average regional sea ice extent and seasonal climate in the region. This, in turn, may help addressing some specific scientific questions, for example, regarding the period associated with a so-called early 20th century warming (e.g. Bengtsson et al., 2004). This period had a clear manifestation in the Arctic, where it was characterized by receding sea ice and positive temperature anomalies surpassed only in the recent two decades, and in the Southern Hemisphere in general (Hegerl et al., 2018). Its manifestation in the Antarctic region, however, is controversial and yet to be revealed.

ACKNOWLEDGEMENTS

This study was partly funded by the Norwegian Polar Institute; D.V.D., S.D., E.I., H.D.J and I.S. would like to thank B. Njåstad (NPI) for supporting this initiative. A number of logbook images were made available via RECLAIM project (http://icoads.noaa.gov/reclaim/). Financial support for the preliminary archive work and photography of ships' logbooks and cataloguing of the images at the University of St. Andrews and the Vestfold Archive was provided through the UK Meteorological Office, Hadley Centre, Exeter, United Kingdom via ERA-CLIM2 project funded by EU-FP7 (Grant Agreement 607029), and the National Institute of Water and Atmospheric Research (NIWA), Auckland, New Zealand. C.W., S.W. and M.V.G. gratefully acknowledge the assistance and guidance provided by the staff of the Special Collections, University of St. Andrews. Thanks, are also due to Joanna Rae, Assistant Archivist at the British Antarctic Survey. The authors would also like to gratefully acknowledge the assistance and guidance provided by the management and staff of the Vestfold Archive, Sandefjord, Norway. O.E.B. is also deeply grateful to the late K.I. Ugland for his advice during the master's study at the University of Oslo, and the staff at Kommandør Christensens Hvalfangstmuseum (The Whaling Museum) in Sandefjord, Norway, for providing access to the original whaling logbooks of the Thor Dahl A/S company.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

Open Research

OPEN RESEARCH BADGES

This article has been awarded Open Data Badge for making publicly available the digitally-shareable data necessary to reproduce the reported results. Data is available at [https://doi.org/10.21334/npolar.2022.eb997ced] and [https://doi.org/10.21334/npolar.2023.b9f318f5].

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are openly available in NPI data repository on data.npolar.no at http://doi.org/10.21334/npolar.2022.eb997ced and http://doi.org/10.21334/npolar.2023.b9f318f5.