The “ dirty weather ” diaries of Reverend Richard Davis : insights about early colonial-era meteorology and climate variability for northern New Zealand , 1839 – 1851

Reverend Richard Davis (1790–1863) was a colonial-era missionary stationed in the Far North of New Zealand who was a key figure in the early efforts of the Church Mission Society. He kept meticulous meteorological records for the early settlements of Waimate North and Kaikohe, and his observations are preserved in a two-volume set in the Sir George Grey Special Collections in the Auckland Central Library. The Davis diary volumes are significant because they constitute some of the earliest land-based meteorological measurements that were continually chronicled for New Zealand. The diary measurements cover nine years within the 1839– 1851 time span that are broken into two parts: 1839–1844 and 1848–1851. Davis’ meteorological recordings include daily 9 a.m. and noon temperatures and midday pressure measurements. Qualitative comments in the diary note prevailing wind flow, wind strength, cloud cover, climate variability impacts, bio-indicators suggestive of drought, and notes on extreme weather events. “Dirty weather” comments scattered throughout the diary describe disturbed conditions with strong winds and driving rainfall. The Davis diary entries coincide with the end of the Little Ice Age (LIA) and they indicate southerly and westerly circulation influences and cooler winter temperatures were more frequent than today. A comparison of climate field reconstructions derived from the Davis diary data and tree-ring-based winter temperature reconstructions are supported by tropical coral palaeotemperature evidence. Davis’ pressure measurements were corroborated using ship log data from vessels associated with iconic Antarctic exploration voyages that were anchored in the Bay of Islands, and suggest the pressure series he recorded are robust and can be used as “station data”. The Reverend Davis meteorological data are expected to make a significant contribution to the Atmospheric Circulation Reconstructions across the Earth (ACRE) project, which feeds the major data requirements for the longest historical reanalysis – the 20th Century Reanalysis Project (20CR). Thus these new data will help extend surface pressure-based reanalysis reconstructions of past weather covering New Zealand within the data-sparse Southern Hemisphere.


Introduction
New Zealand was one of the last places permanently settled on Earth (Wilmshurst et al., 2011) and meteorological records there do not extend back in time with regularity prior to the early 1860s (Fouhy et al., 1992).Qualitative climate and weather observations for New Zealand first came from exploratory voyages that entered waters around the country (Banks, 1768(Banks, -1771)).Subsequently, the increased number of colonial settlers and supply ships arriving during the late 18th and early 19th century (Chappell and Lorrey, 2013) coincided with the earliest written accounts that documented local weather and climate conditions.These observations were often included in regular channels of communication to and from "newly found territory", and some provide the first instrumental measurements of the physical environment.Early colonial-era settlers of New Zealand were very keen to understand the character of climate and weather for agricultural purposes (Holland and Mooney, 2006;Holland et al., 2009).Despite frequent mention of weather conditions in reports or diaries, however, observations were irregularly timed, spo-radically spaced, and sometimes contained little quantitative data.
A key improvement for documenting New Zealand's weather and climate occurred in the early 1850s with the establishment of several fledgling observatories within military fortifications (Fouhy et al., 1992).Instrument-based meteorological observations were recorded by the Royal Engineers in Auckland three times daily, and they constitute some of the earliest known "modern-day" long-term data for New Zealand.The Royal Engineers meteorological observations for Auckland also temporally overlap and merge with early to mid-1860s instrumental observations (Hessell, 1988) that were initiated in an orderly fashion and overseen by James Hector as part of the Geological Survey of New Zealand (Dell, 2013).The network Hector set up is essentially the precursor to the present-day New Zealand Meteorological Service's observing stations, with the long-term observations held by the National Institute of Water and Atmospheric Research (NIWA).
Australasian weather and climate accounts prior to the mid-1850s are sparse in general (Gergis, 2008;Holland et al., 2009;Gergis et al., 2009Gergis et al., , 2010;;Ashcroft et al., 2012Ashcroft et al., , 2014)).As such, additional information that can improve our understanding of past weather and climate for the region are important.Of significance, all types of historic weather observations are being sought by the Atmospheric Circulation Reconstructions across Earth (ACRE) initiative (Allan et al., 2011), which is recovering, digitizing and feeding old synoptic pressure observations into the 20th Century Reanalysis Project (20CR), a reanalysis without data input from radiosondes, aircraft or satellites (Compo et al., 2011;Cram et al., 2015).In this regard, there is a prominent opportunity to link New Zealand historic weather observations with massive data assimilation undertaken by supercomputers to provide realistic representations of regional atmospheric circulation spanning the southwest Pacific and wider Southern Hemisphere.That effort is posed to make a significant contribution to our understanding of past weather and climate change.
As part of a search to identify early weather observations for New Zealand that could be supplied to the ACRE initiative, the National Register of Archives in New Zealand yielded a reference for an historic weather diary that was kept by Reverend Richard Davis, a missionary who lived north of Auckland (Lorrey et al., 2011a, b).In this study, we have analysed that record and we demonstrate the value of the meteorological observations that Reverend Richard Davis kept.To date, the Davis weather diary is the earliest reported quantitative meteorological account for New Zealand that was continuously kept over multiple years.We provide an analysis and modern climatological context for the Davis weather diary data (Fig. 1), and are able to quantify conditions he experienced to deduce similarities and differences in weather and climate relative to today.
2 Background on Reverend Richard Davis and the climate of Northland, New Zealand

Richard Davis biographical notes
According to his memoir, written by friend and correspondent Reverend John Coleman, Reverend Richard Davis (born 18 January 1790, Dorset, England; died 28 May 1863, Waimate North, New Zealand) was associated with the Church Mission Society (CMS) of England.He spent much of his time in northern New Zealand and was stationed for significant periods of time in the settlements of Waimate North (Figs. 1 and 2) and Kaikohe in Northland.In 1831, Davis arrived at Waimate North and established a farm.Davis was also ordained a deacon in the mid-1840s.He was a prolific writer and observer of the natural environment, evidenced by hundreds of letters sent back to England and the CMS that included commentary on physical geography and astronomy (noting the occurrences of comets and the aurora australis).Davis also documented social perspectives of colonial-era settlers and interactions of Europeans with Māori, as well as general activities that occurred near the settlements of Russell, Marsden Vale, Kawakawa and Paihia (Coleman, 1865).

Physical geography and climate of Northland
Northland is a long peninsula of land that extends southeast to northwest ( ∼ 34.425-36.325• S) from north of the Auckland Isthmus to the most northern extent of New Zealand (Fig. 2).The region contains multiple deep-water harbours that intersect the coastline which were prized (though treacherous at times) during the colonial era for anchorage, including Hokianga and north Kaipara in the west, and Whangarei, Bay of Islands, and Whangaroa in the east.In general, the Northland peninsula varies in breadth from 35 to 95 km, and most of the densely settled locations are positioned at low elevations in close proximity to the sea.Topography can be variable, and local relief in some areas can exceed 500 m over a 1 km horizontal span, though in most cases it is only of the order of a couple hundred metres (Orange, 2012).Chappell (2013) recently updated the climatology for the Northland region, and basic information contained therein is similar to Moir et al. (1986).In summary, the region has a mild, humid, and windy climate.Austral summers are warm and humid and winters are mild, with only a few sites receiving a couple of light frosts per year.Mean annual temperatures range between 14 and 16 • C (Fig. 2), with eastern and northern locales being generally warmer than western and southern sites.The prevailing atmospheric circulation over Northland is from the southwest, particularly in winter and spring, but during summer the winds increase from the easterly quarter, especially in eastern districts, to equal that from the southwest.This seasonal wind flow change arises from the changing location of the subtropical ridge (high-pressure belt), which shifts further south in summer and early autumn relative to winter and spring (Fig. 2).In addition, sea breezes add to the proportion of easterlies in eastern areas in summer and early autumn.Spring is generally the windiest season, except in exposed places such as Cape Reinga, where winter tends to be the windiest period.Summer and autumn usually have the greatest number of days with light wind (with mean daily wind speed < 31 km h −1 ).
Rainfall is typically plentiful all year round in Northland, with sporadic very heavy falls.Annual rainfall totals range from 1200 mm in low-lying coastal areas to 2000 mm at  higher elevations.Areas north of Kaitaia receive considerably less rainfall than further south.Dry spells may occur in summer and autumn, but they are generally not long-lived (average dry spell duration is 20 days).Rainfall in Northland predominantly occurs when there is a stationary anticyclone to the east or southeast of New Zealand, and humid northeasterly winds cause significant rain over Northland.Also, extra-tropical depressions or ex-tropical cyclones that pass over Northland on average once or twice per year may cause torrential rain and damaging winds (Lorrey et al., 2014b).Cold, showery weather occurs in Northland with southwesterly and southerly winds, following the passage of a depression from the northwest or west.Easterly winds associated with an anticyclone to the south of Northland may also cause showery weather.Fine weather in Northland mainly occurs when an anticyclone moves slowly over the North Island, and during phases of anticyclone replacement (which typically last two to three weeks during summer).Most parts of Northland receive about 2000 h of sunshine per year, with northern and eastern areas recording more sunshine hours than western and southern areas.It can be very windy in exposed areas, and occasionally Northland experiences gales, sometimes in association with the passage of depressions of tropical origin (Chappell, 2013).
3 Data and Methods  rero).This source was considered as an important prospect to follow through on because the entry for the Davis diary was one of only a few search items that mentioned meteorological tables.Details for the Davis diary showed it was held by the Auckland Central Library (ACL), and a viewing to assess the quality of the meteorological measurements (in terms of physical state of the document, temporal completeness, legibility, and content) was undertaken.The collective components of the Davis meteorological diary numbered in the thousands in terms of entries and comments, and these are outlined in the Results section.We describe the scanning and transcription procedure in the Supplement.

Corroborating Davis' observations and comparative information
To examine the validity of the barometric pressure observations made by Davis, we also corroborated his measurements during days when available ship log data from the Bay of Islands were available.Three voyages from the "heroic" era overlapped Davis' observations for short time spans: the HMS Erebus (Capt.Ross, Great Britain), the USS Vincennes (part of the US Exploring Expedition 1838-1842 lead by Capt.Wilkes) and two corvettes from a French expeditionthe Astrolabe and the Zelee (Capt.Dumont d'Urville).Pressure data for times when these ships were anchored in the Bay of Islands and verification of historic ship tracks were supplied by ACRE through Dr. Rob Allan and Dr. Philip Brohan at the UK Met Office (UKMO).We consider the shipboard measurements were reliable because the barometers onboard would have been calibrated to the highest institutional standard.While no metadata exist about how the barometric measurements may have been regularly checked, it is likely that Reverend Richard Davis took the opportunity to periodically compare his observations with those from ships in port at Russell, Bay of Islands.For the comparison between the pressure series, we show the data in native format (keyed; and in inches of mercury) and then discuss differences relative to measurement site elevations.We also include pressure data 1 day prior to and after departure from port.For comparison to present day, temperature measurements were converted from Fahrenheit to Celsius and pressure measurements recorded in inches of mercury were converted to hectopascals.The Davis pressure measurements are not corrected for temperature, altitude or gravity.
It is evident that the temperature data from the Reverend Davis diary have the least amount of associated metadata.As such, an assessment of those data in their native format was warranted prior to undertaking a correction that could introduce additional errors or biases to the pressure series.We are still considering the most appropriate way to undertake a correction -one way is to obtain enough overlapping data to be able to develop an informed correction using associated local temperature data, but this should only be done with full knowledge of the potential biases those temperature observations might include, in addition to any inherent technique errors.In terms of the altitude and gravity corrections for pressure observations, this can be applied directly on submission of the observations to the International Surface Pressure Databank, which accepts different formats of pressure observations (some native, some corrected, some not).
Comparative daily meteorological records from the NIWA climate database for Kaikohe and Waimate North come from sites that are positioned close to where Davis resided between 1839 and 1851.The closest high-quality daily meteorological observations for the modern period that correspond to the site Davis was located at come from the Virtual Climate Station Network (VCSN; Tait et al., 2006), which is a 5 km 2 gridded field that includes 13 variables from interpolated from station data (see Supplement for more details).The VCSN data set provides 9 a.m.pressure, daily maximum temperature (T max ) and daily minimum temperature (T min ), amongst other variables.Hourly meteorological measurements for the Far North are relatively sparse; however, some do exist for Kaikohe, which overlaps one of Davis' observation locations, and it is very close to the Waimate North site.In order to extract added value from the Davis weather diary aside from describing his twice-daily temperature series, both of Davis' temperature recordings were transformed to be equivalent to VCSN T max and T min using an established relationship between the VCSN daily extremes and 9 a.m. and noon temperature measurements from Kaikohe (established using all available data between 1972 and 2012).T max and T min were then derived from the Davis diary recordings, and were subsequently used to derive T mean .So as to not introduce an interdependence element to the derived VCSN reconstruction, we were also able to produce a time series of 9 a.m.temperatures independently for the VCSN grid us-ing 9 a.m.vapour pressure and the Antoine equation1 .We also used monthly mean pressure measurements from nearby sites (Whangarei and Kerikeri) for comparative purposes (see Supplement for regression equations).
The Davis reconstructed temperatures were compared to extant tree ring proxy data sourced from the Past Global Changes (PAGES) Australasia database.These data have recently been collated for the purpose of undertaking global temperature reconstructions and were already standardized (Neukom and Gergis, 2012) using five different standardization techniques.We have used the "signal-free" (Melvin and Briffa, 2008) chronology produced by Neukom and Gergis (2012) for three cedar (Libocedrus bidwillii) tree ring records to establish new, significant correlations to austral cool season (and winter) temperatures (Lorrey, unpublished) from Takapari, Moa Park and Flanagan's Hut (original chronologies from Xiong and Palmer, 2000) to corroborate the Davis diary winter observations.The relationship between cedar tree rings and temperature was established via correlating the standardized signal-free chronologies to the closest VCSN grid at a monthly level, then aggregating monthly temperatures into seasonal and longer composite averages and rerunning the correlations to achieve the strongest correlation.This exercise clearly indicated that the cedar tree ring growth is sensitive to austral cold season and winter temperatures.The regression equations from these correlations allowed the standardized index values to be transformed to a quantitative temperature, which was then converted to an anomaly relative to the modern period .
The collective temperature anomaly reconstructions from the Davis diary and the tree-ring-based temperature conditions for 1839-1843 and 1848-1851 were fed into the Past Interpretation of Climate Tool (PICT) to derive local, southwest Pacific and Southern Hemisphere climate fields, following the approach used by Lorrey et al. (2014a).The PICT reconstruction approach is essentially a modern analogue spatial field method that uses detrended gridded local and global data (Tait et al., 2006;Kalnay et al., 1996) to assess what the local atmospheric circulation would have been like based on terrestrial palaeoclimate data.A reconstructed temperature anomaly for a proxy site is first compared directly to detrended climatological temperature quintiles for a corresponding grid point.All of the analogue seasons that fall within each quintile are then selected and composited with equal weighting to produce mean geopotential height patterns, which are based on detrended daily NCEP1 reanalysis data (Kalnay et al., 1996).The fact that several sites can then be compiled into an ensemble, and that each of the proxies will have different analogues selected, helps to provide weighting toward the most commonly selected analogue seasons.The synoptic types are classified according to Kidson (2000) and later Renwick (2011) based on the daily output, and relies on the assumption of stationarity for local climatic responses to incident circulation in the maritime climate of New Zealand (i.e. when it is more southerly, it is cooler than normal, and vice versa for more northerly atmospheric circulation conditions).Full details of the PICT method, the significance testing of the synoptic type changes and differences of the mean geopotential height patterns relative to modern are described further in Lorrey et al. (2014a).This approach was used to (a) provide a comparative national-scale context for the temperature anomalies recorded by Davis and (b) provide a wider atmospheric regime context for the observed temperatures.These results are brought to bear in the discussion to contextualize the mean climate conditions recorded by Reverend Davis.

Components of the Davis diary and "dirty weather" comments
Reverend Richard Davis' weather diary consists of two parts: 1839-1844 and 1848-1851.A partial year of weather observations were made by Davis for both 1844 and 1851 and we have transcribed them; 1844 is not considered further in this study because it constitutes less than half a year of observations.The temporal break in the diary corresponds to the time when Davis was ordained as a Deacon and left Te Waimate Mission Station to establish Kaikohe Mission Station.The diary break also marks a period when tumultuous activity occurred in Northland that relates to the onset of the Māori Land Wars (King, 2003), now referred to as the New Zealand Wars, which were fought between the colonial government and Maori tribes over sovereignty and land.There is mention by Davis in his personal diary of an insurrection in Kaikohe being "crushed" in January 1846.To our knowledge, the collective observations and measurements made by Davis comprise the earliest historic land-based meteorological register for New Zealand that has survived to date.It significantly pre-dates other informal weather observations for New Zealand that come from personal diaries as noted by previous researchers (Holland and Mooney, 2006).However, it is possible that earlier missionaries (i.e.Samuel Marsden, who resided in New Zealand from 1816), military personnel, or people involved in agriculture and viticulture (i.e. the viticulturist James Busby, who is mentioned by Davis as having provided him with 50 grape plants on 8 December 1835) could have kept similar quantitative records that are even older.
The two Davis diary components collectively contain > 13 000 meteorological measurements and local environmental observations.Quantitative instrumental observations include 9 a.m. and 12 noon temperature and noon pressure recordings.Qualitative observations include daily wind direction, which are divided into eight basic compass bearings relative to true north, and an additional category termed "variable" (where multi-directional wind flow was noted).Climatology for the instrumental measurements and qualitative observations (both temporal intervals integrated) are presented below (Fig. 4).The comments column within the meteorological register includes mention of frost, ice, hail, wind strength, relative rainfall, cloud, snowfall, thunder, lightning, sunsets, and wildlife behaviour (including bioindicators about migratory waterfowl and insect life).
The Davis diary also includes 67 remarks about "dirty weather" spread throughout the two-volume meteorological register.Davis commonly associated dirty weather with atmospheric circulation from northern and eastern quadrants and in connection to southerly quadrant flow.Rainfall was common during days characterized as having dirty weather, with strong, blustery winds and low cloud cover.The general indication is that the dirty weather remarks made by Davis were indicative of generally gloomy conditions.

Davis' barometer
Analysis of Davis' personal diary entries (Davis, Richard: Letters and Journals, 1824-1863, MS-1211, sourced from Hocken Heritage Collections, Dunedin, New Zealand) was undertaken to try and gain knowledge about the type of barometer he used, where it was purchased, how he received it and how it may have been calibrated.A mention of the word "barometer" is made five times in Davis' personal diary.The following are two of the entries associated with that word: -9 February 1836: in a letter to Rev. W. Jowett in London (clerical secretary of the CMS), a request was made for Davis' friend Nicholas Broughton to obtain a barometer and send it to New Zealand (MS-1211, Vol. 1, p. 118).
-11 April 1839: a comment is made by Davis about inclusion of 3 months of barometer and thermometer data with the letter to Rev. W. Jowett (MS-1211, Vol. 2, p. 9).
Contact with archivists at the CMS of England did not yield any leads about the purchase of the barometer Davis used.
We have also made an enquiry with the Clarke family in Northland (George Clarke was a fellow missionary with Davis at Te Waimate), as well as Heritage New Zealand, who are the curators of the mission house that Davis was based at (to no avail).We do know that a friend of Davis who is mentioned in his letters, Mr Nicholas Broughton, lived at "Swanyard in Holbourn Bridge", London.A census from that era indicates many skilled tradesmen who participated in the manufacture of chronometers, timepieces and ship instruments circa 1835 (Horological Foundation, 2015) resided in Holbourn Bridge, which included a hive of barometer makers who were based locally.It seems likely that Mr Broughton would have purchased equipment there.We recognize that observers in the early to mid-1800s had access to multiple types of barometers (see Jones et al., 1997, for an example); however, metadata about calibration and correction of the Davis barometer are lacking.A common type of barometer made in the mid-1830s that was highly portable was a mercury "wheel" barometer of "banjo" morphology.Davis mentions a "screw" as part of the metadata associated with his observations, which is consistent with that type of equipment.As such, the Davis pressure series and the shipboard observations comprise a measurement pair (n = 29) that can be examined (a) to see how inland/upland station pressure and "near-sea-level" pressure compare and (b) to determine how the Davis pressure measurements (see Fig. 3 top panel) compare in general to other reference series.The common pattern of variability for the aggregated ship data and Davis measurements and their correlation are significant (r = 0.93; Fig. 3, middle panel).The Davis daily pressure observations are consistently offset lower than those reported by all of the shipboard observations (by an average of −0.64 ± 0.10 inches of mercury).This negative pressure measurement offset of −0.64 inches of mercury corresponds to the altitude increase from the harbour where the ships were anchored to the altitude of the site where Davis' land-based measurements were made (Fig. 1).The variance for the Davis and shipboard pressure measurements is also similar (0.19 and 0.25, respectively).As such, we consider the pressure measurements recorded by Reverend Davis to be a robust indication of surface pressure at both sites where he was located, and note that these measurements can be employed as station data which are not corrected for temperature, gravity or reduced to sea level.

Climatology of pressure measurements
The monthly climatology for noon pressure indicates an annual cycle with lower pressure in austral winter and spring and the highest average pressure for late summer and autumn (Fig. 4).Davis's pressure measurements indicate an annual mean value of 1016.47 hPa (when adjusted to sea level), which is similar to average annual values for modern measurements recorded at nearby stations (Kerikeri Aero, 1016.85 hPa; Whangarei Aero, 1016.81 hPa) of equivalent latitude.Across the year, Davis' meteorological diary indicates the highest pressures were most frequent from January to April, with a decrease to the lowest values in winter (Fig. 4; Table 1).Seasonal average pressures recorded by Davis also compare similarly to modern pressure values for autumn, but suggest summer pressures in the early-mid-1800s were higher than present for summer, and lower than present for winter and spring.There are significant intraseasonal and inter-annual variations in the pressure observations recorded by Davis (Fig. 3), which can be attributed to the wide range of synoptic weather systems he witnessed (supported by qualitative descriptions of clouds, precipitation, wind direction and wind strength).Davis also notes some key occurrences of unusually low pressures associated with specific storms (see Fig. 3), which are discussed below along with other observations of weather extremes.Davis also made comments about unusually high pressures during the first 5 months of 1848, and he suggested that the screw on the bottom of the barometer might have been adjusted without his knowledge, causing an artificial inflation of pressure observations by 4/10ths of an inch.This particular period corresponds to the re-initiation of observations being made after a key temporal break in his meteorological diary.We discuss the context of these "high" pressure anomalies noted by Davis in the discussion.-4 November 1833: "Today the thermometer stood at 80 in the shade; this I have never known it to do before since I have been in the country" (MS-1211, Vol. 3, p. 70).
-9 November 1833: "In the shed the thermometer stood at 78; plunged into the garden soil in the sun it stood at 110" (MS-1211, Vol. 3, p. 70).
We note here that there could be some issues with regard to radiation errors (Nakamura and Mahrt, 2005) for these temperature measurements in the absence of metadata about where the thermometer was positioned in the shed, which is not a standard type of enclosure (Parker, 1994), and we also assume Davis used a mercury-in-glass instrument.

Climatology and extremes from 9 a.m. and noon temperature
The 9 a.m. and noon temperatures recorded by Davis (Fig. 4, Table 1) ranged from a maximum in January to a minimum in July (19.3 • C to 8.9  The comparison of reconstructed T mean , T max , T min , and diurnal range from the Davis diary relative to VCSN statistics are presented in Table 3.We note specific occurrences when more than ±0.5 • C difference exists between the reconstructed Davis monthly temperature values and the VCSN, but do not attach any significance to these differences due to the large discrepancy in sample size for the individual monthly correlation functions, because of the associated errors in this style of reconstruction, and because of the limitations on the metadata for the thermometer Davis used.Nevertheless, T max , T min and T mean for December, January and March (and T max and T mean for November) appear warmer in the Davis record relative to present day, while May-August are categorically cooler.Diurnal temperatures were only relatively different (warmer) for January in the Davis record.Qualitative observations made by Davis about extremes related to temperature, such as snowfall, ice, and frost are brought to bear in the discussion about the realism of these differences.

Rainfall
Qualitative comments by Reverend Davis about rainfall were summed from the daily observations and indicate ∼ 34 % of all days had some form of precipitation (Fig. 5).Comments about fine, dry and/or calm conditions were aggregated and tallied and indicate 38 % of the time constituted absence of rain.Consecutive dry day stretches (as noted by no mention of significant precipitation) documented by Davis topped out at 18 days duration (days 207-224) during late July-mid-August 1839.That is slightly longer than the maximum in-terval of 13 consecutive dry days that occurred during August 1987 as indicated by rain data from the VCSN grid point that corresponds to Davis' site.Overall, the climatology of rainfall (derived from aggregating days with all rain keyword indicators) shows December and January were the driest months, while June, July and August were the wettest months that Davis experienced (Fig. 5).This is very similar to what the VCSN data indicate for the grid point that corresponds to Davis' site (with January and February being the driest months, and June-August being the wettest).The opposing annual trends for wet vs. dry days also lends to the same assertion.By proportion, "dirty weather" was most frequent during July, and least frequent in December.Comments about cloud cover suggest greater frequency of cloudy skies from January-May, and less so during July-December; however, this general pattern (Fig. 5) may be skewed by the fact that clouds may have not been mentioned during rainy days.

Winds
The general wind direction recorded by Davis was used to develop a wind climatology (Fig. 4; Table 4) that can be used to gauge the local conditions he experienced, including how incident atmospheric circulation changed through the seasons.This analysis can also be used to determine whether there are differences in the frequency of general prevailing winds during Davis' time relative to present day.Davis mentions "variable" or squally/disturbed conditions ∼ 11 % of the time, with almost twice as frequent occurrence during summer than other seasons (Fig. 6).On an annual basis via percentage, southerly, southwesterly, and westerly winds were most common (constituting ∼ 50 % of all entries).Grouped by direction quarter, westerly winds were  thunder are poorly correlated in the Davis diary, typically because remarks about thunder were commonly made when it was "off in the distance".Commentary related to "rivers in flood" that are mentioned in the Davis diary indicates that December was the most common month when floods happened, followed by February and November (Fig. 6).Davis also makes mention of "gale" winds, which are interpreted here as blustery stronger-than-normal winds that lasted for a substantial time during the day.The climatology of those comments (Fig. 6) indicates a general rise in frequency beginning at the end of summer, culminating in October.

Ex-tropical cyclone of 1 March 1840
A significant commentary about an extreme weather event was made by Reverend Richard Davis at the end of the February 1840 meteorological diary register and also in his personal diary.Davis wrote about sustained strong winds with heavy rain that wrought damage to a fence he had recently installed on his farmland.The personal diary entry mentions "a hurricane", and the meteorological diary comment section specifically indicates that anomalous low pressure influenced the Waimate site, with a minimum pressure in native format of 28.09 in.(28.73 in.when adjusted to sea level) recorded close to midnight on 1 March 1840.Davis remarks that the "mercury rebounded rapidly to 29.22 inches (29.86 inches when adjusted to sea level) by noon the following day" as the storm passed.When the adjusted sea-level pressure recordings are converted to hectopascals, the antecedent and follow-on conditions from the low-pressure anomalies are close to 1011 hPa, which are reasonable values for late summer-early autumn when compared to present-day values for early March.The adjusted low-pressure anomaly of 28.73 in.(973 hPa) recorded at midnight on 1 March 1840 by Davis is significant in that it, along with preceding and following high pressures and general wind direction changes, is akin to a signature of an ex-tropical cyclone interaction,

Extreme temperatures
A comparison of the monthly average T mean , T max and T min , 9 a.m.average, minimum single-day and maximum singleday extreme temperatures are shown for the Davis diary with reference to the VCSN for the same location (Tables 2 and 3).
For the mean extreme high monthly values, the Davis diary is categorically cooler across all months relative to present day by an average of −2 • C. For single-day maximum 9 a.m.temperatures, none of the extreme values from the Davis diary exceed extreme temperatures for the modern era.On average across the year, the VCSN 9 a.m.single-day extreme high temperatures for each month are 2.9 ± 1.6 • C higher than those Davis recorded, with significantly larger differences in the monthly 9 a.m.extreme relative to the modern era in March-May and July-September (Table 3).However, it is interesting to note that, for extreme 9 a.m.temperatures, the modern period has some occurrences of colder mornings for February-April, June and October relative to the time Davis was residing in the Far North.

Frost, ice, hail and snow
Several qualitative remarks related to cold temperatures and frost can be found in Davis's personal and meteorological diaries.Davis's sent a letter to John Coleman dated 21 June 1834 (Coleman, 1865: 180): "Last night was our first night of frost this year.The ice this morning was the thickness of a shilling" (approximately 1.2 mm thick).Davis again mentions ice in the meteorological diary on 15 July 1839, indicating "ice 1/4 inch thick" (6.35 mm), presumably observed on the surface of the millpond at the Waimate North site.Frost is noted 106 times by Davis over the span of 9 years.His observations suggest no frosts occurred during November-March and that the frostiest month was July, with more than half of the frost events occurring in winter (June-August).Hail was also observed by Richard Davis for all seasons except summer (Fig. 6), with a peak occurrence in winter (July), dropping away to no hail accounts in December.Snowfall was also mentioned in the Davis meteorological diary once as an isolated event spanning two days for 30-31 July 1849.For the 2 days of snow that were mentioned, Davis's meteorological diary comments are as follows: -30 July 1849: "Hail storms.This morning the southern hills and Poutahi [sic] covered with snow." -31 July 1849: "This morning the hills were again covered with snow." In a personal letter to a friend in England, Davis also affords a parallel description (Coleman, 1865: 350): -30 July 1849: "The hills were covered with snow, the first ever seen by the natives inhabiting this part of New Zealand.The Putahi was also covered." -31 July 1849: "This morning the hills were again white with snow." Contrary to widely held belief that it never snows in northern New Zealand, there are six historic accounts of frozen precipitation for Auckland/Northland (Fig. 7a and b) that can be brought to bear for reference.Two of the events (1939 and 2011) are noted as having delivered at least some light snow to high elevations.Geographic coverage of eyewitness accounts for the occurrence of frozen precipitation including snow (and/or sleet and/or graupel) related to six historic Auckland/Northland events (Fig. 7a) suggests the 1849 snow seen by Davis may have been akin to the 1939 event, which saw snowfall on isolated ridges as far north as Cape Reinga, with the next closest analogue being the 2011 event.The similarities and diagnostics for these analogues are brought to bear in the next section.

Discussion
5.1 How similar or different are Reverend Davis' weather observations of the early colonial era from today?
The Reverend Davis meteorological diary from Waimate North and Kaikohe contains years of continuous daily instrumental and qualitative observations for several key variables.The most notable components of this diary are quantitative measurements of temperature and barometric pressure (Fig. 3).A comparison of the barometric pressure from Waimate North to reference series from ships (Fig. 3) suggests that Davis' pressure measurements can be used as station data.When compiled into a climatology and compared to reference data series derived from the VCSN, there are elements of the Davis meteorological register that undeniably indicate he was making faithful measurements of local conditions.The annual cycle pattern is evident in all three instrumental data sets, and their patterns are phase-locked in terms of the timing of the peaks and troughs seen in modern climatology records.The relative temperature changes for the 9 a.m. and noon temperature climatology (Table 2; Fig. 4) between summer and winter are also quite similar to the modern era, with a change of ∼ 10 • C between summer and winter.The distinctions of the Davis diary observations with respect to modern times, however, are observed for the overall offsets in mean monthly temperatures and some of the daily temperature extremes (Tables 1-3).
5.2 Can we corroborate the general indications of past temperature anomalies noted in the Davis diary and determine their cause?
Recent work of the Australasia palaeoclimate research community has gathered high-resolution climate proxy data (Neukom and Gergis, 2012) and made it available in a centralized database (Kaufman et al., 2014).There is thus an opportunity to examine some of those proxy data, which in the case of New Zealand constitute tree ring chronologies, alongside the reconstructed temperatures for 1839-1851 (exclusive of the missing years between the diary components) based on Reverend Davis's observations.Collectively, the Davis diary anomalies and corresponding tree ring reconstructed anomalies for winter temperatures can be integrated in PICT (see Lorrey, 2014, and http://pict.niwa.co.nz for details of the reconstruction technique and prior application) to provide greater context for the local conditions Reverend Davis experienced in his lifetime.The Davis diary mean winter temperature anomalies (−0.9 • C) are comparable to anomalies for Libocedrus treering-based reconstructions from Takapari (−1.9 • C), Moa Park (−0.36 • C) and Flanagan's Hut (−0.90 • C) (see Xiong and Palmer, 2000, for chronology details).The resulting synoptic type changes that would have caused colder winter temperatures for all sites, as indicated by the PICT-based reconstructed climate fields, would have been driven by an increase in "trough" types (Kidson, 2000), a reduction in "highs" over the country (the "H" zonal type of Kidson, 2000) and a reduction of "blocking" synoptic types that typically are known for increasing the frequency of northerly quarter flow (Kidson, 2000).There are clear "differences in opinion" amongst the proxy data with regard to what the spe-cific change in frequency of occurrence for individual synoptic weather types may have been for 1839-1851 (Fig. 8).However, the integration of all sites together shows confidence in the reconstructed regional atmospheric circulation field (z1000) in the New Zealand sector, with increased lows to the east of the country and over the Chatham Islands (Fig. 9).This regional atmospheric circulation pattern would have produced more frequent S and SW winds with coolerthan-normal temperatures for New Zealand (Fig. 9).
Moreover, a projection of anomalous temperatures for the southwest Pacific, which is a result of the integrated New Zealand tree ring reconstructions with the Davis instrumental temperature observations, suggests an El Niño-like pattern existed for the mean winter climate state during 1839-1851 (Fig. 9).Those signals are corroborated against existing coral palaeotemperature reconstructions (albeit annually resolved; see Delong et al., 2012, andDunbar et al., 1994) that indicate the integration of the Davis temperatures with the tree ring data and their collective "opinion" about the tropical Pacific mean climate state is robust.Looking further afield at the wider Southern Hemisphere z1000 field (Fig. 10) the atmospheric circulation is characterized by an anomalous high pressure in the Bellingshausen Sea paired with a low pressure east of the Drake Passage.This configuration has a spatial pattern similar to what is observed for the Pacific-South  American mode (PSA; Mo and Paegle, 2001).Overall, the indications from the PICT spatial field projections are that at least two key teleconnections and climate drivers may have had an important influence on the "dirty weather" that Rev-erend Davis observed during 1839-1851.Some parts of the observed pattern (Figs. 9 and 10) are similar to what has been implicated for mean summer conditions based on 22 equilibrium line altitude temperature reconstructions for the LIA (Lorrey et al., 2014a).The integration with tree ring evidence also lends to an interpretation that the Davis meteorological diaries provide a crucial eyewitness account for the end of this recent but (locally) poorly understood climate episode.
5.3 How different are the mean and extreme conditions observed by Davis relative to today?

Temperatures and the presence of ice
The direct 9 a.m.temperature comparison of the VCSN and the Davis recordings suggest that, categorically, the 9 a.m.average temperature and the most extreme 9 a.m.temperature that Davis experienced was colder than the modern era (Table 2).The transformation of the Davis diary 9 a.m. and 12 noon temperature recordings to be directly comparable to the VCSN modern climatology of daily mean temperature and temperature extremes (T max , T min and T mean ) indicates the most significant differences were colder daily mean and daily extreme temperatures for May-August.These anomalies are congruent with wider climate change syntheses that have recognized long-term warming trends in minimum temperatures (Pittock and Wratt, 2001).The Davis diary also suggests that average monthly temperatures were relatively warmer for November-March, with the clearest signature of warm anomalies for December and January (Table 3).However, we recognize that some of the climatological results for summer and winter appear consistent with findings related to poor thermometer ventilation and/or exposure (Nicholls et al., 1996).In the context of climate driver associations, proxy evidence of past El Niño-Southern Oscillation (ENSO) activity indicates swings occurred between El Niño and La Niña episodes in the early to mid-1800s when Davis was making observations (Gergis and Fowler, 2009).This probably means the climatological mean values presented here "blend" successive ENSO events (and anomalies for Northland) via the averaging process.It may be likely that one particularly strong El Niño and/or a protracted event could skew this perception.While we have opted to not analyse the individual seasonal climate anomalies from the Davis diary in this study, future work looking further afield using Australian weather diary records could prove fruitful for integration, corroboration and delineation of past ENSO teleconnections and activity.
The documentation of surface ice on two separate occasions by Reverend Davis appears unusual.The 15 July 1839 ice event indicates temperatures at 9 a.m. were 4.4 • C.This is not the coldest 9 a.m.temperature noted by Davis.Omission of other ice comments may indicate something to the effect that observations of ice as a phenomenon may have been sporadic, infrequent, confounded with frost, or only noted for highly significant events.The alternative is that the conditions for ice formation and/or persistence into the early morning may have only been amenable during the days when Davis noted its presence where he was living.The 9 a.m.temperatures from the Davis diary indicate an extreme low value of 1.7 • C occurred on 8 July 1850, which is clearly colder than the temperature on 15 July 1839.In consideration of the fact that early morning temperatures are typically colder than those at 9 a.m., our VCSN-based T min reconstructed temperature of −1.4 • C for 8 July 1850 suggests that freezing tem-peratures at night-time (and associated surface ice formation) probably occurred episodically during the early colonial era in Northland.While little photo-documentary information about freezing cold and past ice presence in Northland exists, evidence from other undiscovered historic weather journals might shed more light on this phenomenon.Moreover, traditional Māori knowledge has suggested surface ice formation in the recent past that coincided with the early part of the instrumental observation period may have been more frequent than the present day (King et al., 2008), and that sentiment is congruent with palaeoclimate proxy interpretations.

Snowfall (frozen precipitation)
It is difficult to compare the atmospheric conditions related to the historic Northland snowfall events.Extended reanalysis integrations that are meaningful for New Zealand are not available yet for 1868, and in general past daily weather depictions are data-sparse within the 20th Century Reanalysis for the pre-1950 interval (Cram et al., 2015).The 1904 snowfall analogue cannot be fairly compared to the other analogues due to data sparseness (and this sentiment is probably applicable to the 1939 analogue because of high-latitude data sparseness).However, there are similarities in terms of the geopotential spatial field signatures of the 1939, 1976, and 2011 events (Fig. 7b).A significant "low" anchored south of the Chatham Islands extending to the fringe of the Ross Sea (which was potentially blocked to the east) and a strong "high" over southeast Australia and Tasmania are common to Circles represent values that are 1.5 to 3 times the interquartile range away from the middle 50 % of all of the data, while stars represent extremes that are more than 3 times the interquartile range away.
those three analogues.The general atmospheric circulation pattern for each of the snowfall events facilitated a corridor of strong southerly air drawn off of the Antarctic continent fringe that was transmitted to northern New Zealand.The connection of modern-day events that overlap the satelliteobservation period which have a similar depiction in reanalysis data indicate that the 30-31 July 1849 event was probably of similar origin.

Pressure observation metadata
In the Davis diary, there is a written note underneath January 1844 (before January 1848): -"Note: in the following pages, from Jan. 1 1848 to August the 1st 1848 the barometer was caused to range 40 parts of an inch higher than usual from an alteration having been made in the bottom stopper screw by some unknown hand.This was not found until August 2 1848.The month of July was arranged in copying." The range of pressure observations that were made during the January-June interval in question appear higher than normal relative to the rest of the record (Fig. 11).We have no reason to not trust the metadata comment by Reverend Davis found in the diary.As such, we have corrected the first 6 months of data in 1848 by subtracting 4/10ths of an inch of pressure prior to converting the measurements to hectopascals and analysed these data according to the corrected version.Future work that will see those measurements integrated into the International Surface Pressure Databank (ISPD) (Cram et al., 2015) will mean the scale of the pressure adjustments can be tested in subsequent reanalyses, and this will afford an additional opportunity to examine the Davis pressure series (including means, variability and extremes) in more detail.

Conclusions
The observations in Reverend Richard Davis's two-volume meteorological diary represent some of the oldest surviving instrumental observations from the colonial era in New Zealand.The data in this historical register are not as comprehensive as the observations subsequently taken by the Royal Engineers in Auckland during the early to mid-1850s (thrice daily), or those from James Hector's fledgling meteorological service network of the late 1860s.However, it is fitting that Davis should be recognized as having made some of the most significant and earliest contributions to New Zealand meteorology and climatology.The extent and breadth of the observations as well as their general antiquity suggest Reverend Richard Davis probably deserves the title of New Zealand's first meteorologist.When Davis' temperature observations are transformed to be comparable to modern-day VCSN T mean , T max and T min observations, it appears as though temperatures were categorically cooler during winter when he was resident in the Far North.The wind observations that are provided by Reverend Davis also suggest southerly-quadrant flow was more frequent than present day.The timing and descriptions of monthly and seasonal climate anomalies, when compared to tree ring and coral proxy data , suggest a connection to ENSO and potentially the PSA existed for New Zealand climate during the mid-19th century.It is likely that these two climate drivers guided some of the local anomalies and synoptic variability that Reverend Davis observed.With the addition of new data fed into an extended reanalysis, the depiction of past conditions will be clearer, and these hypotheses can be tested more rigorously.
Extreme temperature values, potentially linked to a subtly different mean climate state (Mann et al., 2009), suggest Davis experienced a relatively higher proportion of what are normally uncommon occurrences of frost and rare events (freezing, ice, snow) that do not typify the modern climate and weather of Northland.Overall, the "dirty weather" comments Davis penned with his extensive instrumental observations provide an eyewitness account of the Little Ice Age conclusion in New Zealand.The LIA culmination is notoriously indicated by historic photos and paintings of ice margin positions with juxtaposed moraines along the Southern Alps margin to the south of where Davis lived that unequivocally show glaciers were much more extensive relative to today (Chinn et al., 2012).Extended evidence from the Southern Alps using equilibrium line altitude-based summer temperature reconstructions (Lorrey et al., 2014a) similarly suggests generally cooler conditions existed during Davis's time in the Far North, with other proxy evidence demonstrating seasonal variability -including both cold and warm temperatures -was associated with enhanced ENSO activity (Fowler et al., 2012).As such, the anomalies of colder winters and warmer summers on average during Davis' time are not unexpected, and this evidence enriches our understanding that early settlers may have faced significant climate anomalies (such as drought and deluge) that New Zealanders continue to grapple with today.
The "discovery" of this meteorological gem in a local archive raises the interesting point that future prospects for historic climate work in New Zealand are numerous.There are clear indications that historical documents contain instrumental weather observations and some of these observations overlap and even antedate the Davis diary, based on initial investigations about ships that transited into New Zealand waters during the colonial era (Chappell and Lorrey, 2013).Our expectation is that extension of historic climate work utilizing a range of documentary archives will enrich the knowledge about the range of natural weather and climate variations that are possible, and this endeavour is requisite for contextualizing past-to-present historic trends and for making adequate preparations for future changes.
The Supplement related to this article is available online at doi:10.5194/cp-12-553-2016-supplement.Central Library for her assistance in allowing access to the Reverend Richard Davis diary in the rare manuscripts collection and for her invaluable help in facilitating permissions for the scanning the diary.Rob Allan, Philip Brohan, and Gil Compo are thanked for fruitful discussions about this work.Gregor Macara is thanked for supplying VCSN data from the NIWA climate database.Comparison pressure series from ship logs was obtained via the French Pacific Fund project "Recovery, safeguarding and dissemination of historic French meteorological and climatological data: a Pacific component of the international ACRE programme of Atmospheric Circulation Reconstructions over the Earth (Récupération, sauvegarde et diffusion des données historiques françaises de météorologie et de climatologie: une composante Pacifique du programme international ACRE de reconstitution de la circulation atmosphérique terrestre)" This work was supported by the NIWA core funded project "Climate Present and Past" CAOA1501.

Figure 1 .
Figure 1.Top left: print of a photomechanical portrait of Reverend Richard Davis taken ca.1860, from the file print collection, Box 16.Ref: PAColl-7344-97, Alexander Turnbull Library, Wellington, New Zealand, sourced from http://natlib.govt.nz/records/23073407. Top right: a digital scan of the Davis meteorological diary for July 1849 which also includes commentary about dirty weather and snow.Bottom: the Waimate North mission house in the Far North of New Zealand where Davis lived.

Figure 2 .
Figure 2. Map of Northland, New Zealand, including major points of interest in Reverend Richard Davis' meteorological diary.The inset map shows New Zealand's location in the southwest Pacific and a box around the Northland region.The base map displays the median annual temperature for the region, based on the 1981-2010 climatology period (temperature legend on right).The top bar plot shows monthly rainfall (1985-2010 period) and the bottom bar plot shows monthly temperature (1985-2010 period) for Kaikohe, with frost day occurrences (triangles) inset on the temperature plot.

Figure 3 .
Figure 3. Top: monthly pressure observations from the Reverend Richard Davis (RRD) meteorological diary for 1839-1843.Number on x axis denotes month of each year.Circles represent values that are 1.5 to 3 times the interquartile range away from the middle 50 % of all of the data, while stars represent extremes that are more than 3 times the interquartile range away.Middle: comparison of pressure observation in inches mercury from RRD relative to ship data in the Bay of Islands for the same day.Bottom: RRD pressure observation vs. expedition measurements (leader noted in parentheses) from USS Vincennes (Wilkes), the corvettes Astrolabe and Zelee (d'Urville) and the HMS Erebus (Ross).There are 29 pairs of daily observations and so the x axis simply shows the comparisons of Davis' record to the three ships in a sequence with the specific intervals noted.

4. 3
Temperature 4.3.1 Temperature recordings and thermometer metadata Davis recorded twice-daily (9 a.m. and noon) temperature at the Te Waimate mission house grounds and Kaikohe (Figs. 1 and 4), and several comments related to temperature recordings are made by Davis in his writings to others and in his personal diary.Davis also made sporadic observations about soil temperature and contrasted temperature measurements in the direct sunlight as well as in the shade.The general commentary from Davis (below) suggests that the thermometer was kept in a ventilated shed in the shade.

Figure 4 .
Figure 4. Top: climatology of 9 a.m.temperature and noon temperature and pressure measured by Reverend Richard Davis at Waimate North and Kaikohe (means for 1839-1843 and 1848-1851 inclusive).Bottom: seasonal wind climatology (% frequency observation) for the same sites and interval.

Figure 5 .
Figure 5. Climatology of qualitative observation for "dirty weather" rain days and "dry" days (left-hand scale) vs. "dirty weather" and cloud (right-hand scale) percentage of days per month in the Reverend Davis' meteorological diary.

Figure 6 .
Figure 6.Climatology of qualitative observation for "gale" winds and frost days (left-hand scale) with flood and hail (right-hand scale) percentage of days per month in the Reverend Davis' meteorological diary.

Figure 7 .
Figure 7. (a) (top) Distribution of historic frozen precipitation events (snowfall, sleet and graupel) for northern New Zealand and (bottom) reported elevations for the eyewitness accounts above plotted by latitude, with demarcation lines separating the minimum estimated settling elevation for frozen precipitation for each event.The diamond colours note evidence for distinct events: red -2008; green -1976; orange -1868 and 1904; grey -2011; blue -1939.The maroon (encircled) diamond indicates the 30-31 July 1839 event recorded by Reverend Davis for the Putahi volcanic cone when he was living at Waimate North.(b) 500 hPa wind strength and streamlines for the aforementioned snowfall events, courtesy of the 20th Century Reanalysis v2.

Figure 8 .
Figure 8. Top: frequency of New Zealand synoptic types (x axis) during austral winter as determined by an ensemble composite of reconstructions from three tree ring proxy sites and the Reverend Richard Davis weather diary for 1839-1851 CE.Grey bars indicate climatological frequencies in terms of percentage (y axis, left); box-and-whisker plots indicate distribution of anomalies in terms of change in frequency (y axis, right) indicated by the ensemble reconstruction.The black horizontal line in each box is the median bound by the 25th and 75th percentile, while whiskers are 5th and 95th percentile.Synoptic type abbreviations followKidson (2000;   see Supplement for full details).Bottom: heat map of New Zealand synoptic type (x axis) frequency changes with respect to climatology for individual site members (y axis) of the ensemble composite for 1839-1851 CE.Significance of synoptic type frequency changes was assessed using a Monte Carlo approach.A total of 10 000 simulations of synoptic type evolution were realized based on Markov chains constrained by the observed frequency and transition probabilities betweenKidson's (2000) synoptic types observed during the modern reanalysis era.Circles and stars represent anomalies significant at the 90th and 95th level, respectively.Figures generated using the Past Interpretation of Climate Tool (PICT) courtesy of National Institute of Water and Atmospheric Research (NIWA).See Lorrey (2014) and http://pict.niwa.co.nz for details.

Figure 9 .
Figure 9. Top left: June-August (JJA) geopotential height anomaly at 1000 hPa (z1000) over the New Zealand region for 1839-1851 CE determined by an ensemble composite of reconstructions from three tree ring proxy sites (Moa Park, Takapari and Flanagan's Hut) and the Reverend Richard Davis weather diary.Anomaly height is in metres.Reanalysis data are courtesy of the National Centers for Environmental Prediction (NCEP).Confidence intervals (90th and 95th) are noted with black (dashed and solid) contour lines.Top right: temperature anomalies for JJA as reconstructed using the selected analogue circulation patterns from four sites for 1839-1851 CE.Temperature anomalies are degrees Celsius.Bottom: JJA sea surface temperature (SST) anomaly over the southwest Pacific region for 1839-1851 CE determined by an ensemble composite of reconstructions from four proxy sites.Temperature anomaly is in degrees Celcius.SST reanalysis data is courtesy of the Hadley Centre (HADSSTa v3).Confidence intervals (90th and 95th) are noted with black (dashed and solid) contour lines.Supporting temperature reconstructions for years corresponding to the New Zealand data and associated errors are shown as purple symbols on the map to denote locations of reconstructions and alongside the SSTa scale with associated 1 standard deviation errors.The SSTa reconstructions are based on coral Sr/Ca from the Great Barrier Reef (triangle), New Caledonia (circle) and Fiji (square) in Delong et al. (2012) and from d18O for the equatorial Pacific at the Galapagos Islands (hexagon) after Dunbar et al. (1994).Base figures were generated using the Past Interpretation of Climate Tool (PICT) courtesy of National Institute of Water and Atmospheric Research (NIWA).See Lorrey (2014) and http://pict.niwa.co.nz for details.

Figure 10 .
Figure 10.July-August geopotential height anomaly at 1000 hPa (z1000) over the Southern Hemisphere for 1839-1851 CE determined by an ensemble composite of reconstructions from the Reverend Davis diary temperatures and three tree-ring proxy data series (same as Fig. 9).Anomaly height is in metres.Reanalysis data are courtesy of the National Centers for Environmental Prediction (NCEP).Confidence intervals (90th and 95th) are noted with black (dashed and solid) contour lines.Figure generated using the Past Interpretation of Climate Tool (PICT) courtesy of National Institute of Water and Atmospheric Research (NIWA).See Lorrey (2014) and http://pict.niwa.co.nz for details.

Figure 11 .
Figure 11.Pressure series for the second half of the Reverend Richard Davis meteorological record showing adjusted and unadjusted (clear/white boxes) pressure series for January-June 1848.Circles represent values that are 1.5 to 3 times the interquartile range away from the middle 50 % of all of the data, while stars represent extremes that are more than 3 times the interquartile range away.
There are no other entries that indicate what instrument he had and how the instrument was calibrated.Prior to discussing the observed climatology and extreme pressure values, we outline a corroboration of Davis pressure measurements.Several ships of exploration that transited through New Zealand waters or were based in New Zealand on military operations report being anchored east of Waimate North and Kaikohe in the Bay of Islands (Fig.2).
Three separate occasions in 1840 are used to compare the Davis pressure measurements to parallel observations made on British, American and French vessels (the HMS Erebus, the USS Vincennes, and the Astrolabe (and Zelee), respectively; data provided from ACRE by Rob Allan, UKMO).
• C for 9 a.m.; 22.2 to 11.4 • C for noon).Mean 9 a.m.vapour pressure and the Antoine equation were used to derive a local 9 a.m.temperature from the VCSN relative humidity values (instantaneous) to compare to climatic means calculated from the Davis diary 2 .Mean annual 9 a.m.This was done because the VCSN temperature data include minimum and maximum values that can occur at any time during a day rather than a set time.T mean can be calculated from those categories; however, use of T mean , T max or T min to compare to Davis • C, which is 2 • C lower when compared to a VCSN average 9 a.m.temperature of 16.4 • C (Table2).Monthly 9 a.m.temperature variance was greatest for December and lowest for March in the Davis record.The 9 a.m.temperature derived from the VCSN grid closest to the Waimate and Kaikohe sites also indicates that Davis' measurements of maximum extreme monthly 9 a.m.temperature were categorically cooler than those observed during the modern era.In addition, many of the 9 a.m.minimum extreme temperatures appear cooler than present day, with the exception of February-April, June and October (Table2).4.3.3Tmax , T min and T mean derived from Davis temperature measurementsComparisons between local high-resolution hourly temperature measurements at Kaikohe and the corresponding Kaikohe VCSN grid were used to generate correlation functions for T max and T min , where use of noon and 9 a.m.temperatures as measured by Davis were converted to T max and T min respectively.This was done so the Davis diary measurements could be directly compared to a modern VCSN-based climatology representative of the Waimate North and Kaikohe sites where Davis took temperature measurements.The fidelity of the correlation functions (and therefore the VCSN reconstructed temperatures from the Davis diary) are better for noon temperature and T 2 max than for 9 a.m.temperature and T min .In addition, correlations are strongest for the austral cool season (T max vs. noon r>0.75 for April-October inclusive; T min vs. 9 a.m.r>0.65 for April-August inclusive) than

Table 2 .
VCSN-equivalent temperatures from the Davis diary for 9 a.m.mean, 9 a.m.extreme minimum and 9 a.m.extreme maximum values with reference to VCSN 9 a.m.temperature data for 1972-2012.
for the warm season (T max vs. noon r< 0.75 for November-March inclusive; T min vs. 9 a.m.r< 0.53 September-March inclusive; see Supplement for more details).

Table 3 .
VCSN-equivalent average monthly T min , T max , T mean and diurnal temperature range based on Reverend Davis 9 a.m. and noon temperatures compared to modern climatology(for 1972-2012).Bold (italic) highlighting indicates warmer (colder) differences of more than 0.5 • C for Davis observations relative to the present.

Table 4 .
Percentage frequency per month (and averaged by season) for qualitative wind direction observations by Reverend Richard Davis for the entire span of his observations.VRB: variable.
weather, including thunder and lightning, hail, frost, ice and floods.Comments about thunder are greatest in October and January and least frequent in March.There is no mention of lightning during August-November, with highest frequency of comments in March and June.In general, lightning and