Only a limited number of Arctic meteorological stations were in operation before 1910, and in few places have observations been carried out before the late 19^th century. In Fennoscandia, the oldest systematic climatic observations north of 65˚N were in Tornio (Finland) during a period from 1737 to 1749, while regular weather stations were initiated around 1850. The first station in northern Russia was at Arkhangelsk in 1813. Early meteorological observations were initiated in Greenland in the period 1866-1895. Meteorological observations in the Arctic Ocean began at the voyage headed by F. Nansen on-board the “Fram” in 1894-1896 and were again taken up during the forced drift in the ice of other ships in the 1930s. The first real meteorological station in the High Arctic was Green Harbour (Svalbard) in 1911. A new phase of the Arctic data coverage started in the mid-1930s with North Pole Ice stations that continuously operated in the Arctic Ocean up to 1991. Economic reasons later caused a disappointing reduction in the number of existing Arctic meteorological stations during the 1990s, especially in northern Russia and Canada. The available meteorological records, however, clearly conveys the impression of high Arctic temperature variability; both on an annual basis and between individual regions in the Arctic.

Long Arctic surface air temperature records (Figure 13) illustrate conditions during the last cold interval of the Little Ice Age, which was terminated by a marked increase in mean annual air temperature around 1920. This rise was mainly derived from rising winter temperatures, while the temperature increased less in other seasons, especially summer. The termination of the Little Ice Age was followed by a warm period, recorded at many Arctic meteorological stations, continuing to around 1940-1950, where after a widespread cooling commenced. In some parts of the Arctic this cooling culminated around 1970-75 and was followed by a corresponding temperature recovery towards the end of the 20^th century. In some regions, such as Greenland and Iceland, cold conditions prevailed almost to the end of the 20^th century, while in other regions (e.g. Svalbard) temperatures have been rising without interruption since the cold period around 1970 [139].

The period 1915-1940 (Figure 13) represents a period of general warming in the Arctic especially influencing northern Alaska, northern Canada, Greenland and the northern parts of Russia and Siberia. The warming was by far most pronounced during the winter season (DJF) and least during the summer (JJA). The Svalbard-northern Russia region was exposed to relatively warm autumns (SON), suggesting increased cyclonic activity, leading to retreat of sea ice in the Kara Sea. This may be explained by the onset of a prolonged weakened state in the continentally based Siberian High in this period, as suggested by Meeker and Mayewski [82]. Hanssen-Bauer and Førland [140], however, suggested that the early 20^th century temperature increase in the Norwegian Arctic cannot be explained as a result of changes in advection alone. It may therefore be suggested that increasing cyclonic activity along the northern coasts of Russia and Siberia by wind stress may have contributed to opening of the sea ice cover, releasing heat from warm water below to the atmosphere, and thereby providing an additional source of heat. In order to illustrate the potential effect of this on air-temperatures, the average heat loss over the ice covered Arctic basin is about 6 W/m², while the heat loss to the atmosphere from the ice-free north eastern Greenland Sea averages is no less than 200 W/m² [141] [142].

Whatever the reasons for this warming may be, several very mild winters were recorded in W Greenland, most notably in 1929, where no land-fast sea ice formed at all and hunters were able to use kayaks throughout the whole winter [56]. At the same time new fish species appeared in the central W Greenland region, causing a significant change in Inuit hunting habits towards open sea fishing. The most important new species in the Disko Island region was the Greenland cod, locally known as the big-headed cod, which presumably is identical to the Icelandic cod [56]. Most likely, larvae of the Icelandic cod spread to SE Greenland waters in the early 1920s, driven by an enhanced Irminger Current, and for some decades were able to reproduce and survive in SW Greenland waters [143]. Whatever the precise background, this species represented a very important resource in Greenlandic economy for about 15 - 20 years, after which a decline set in, ending with a number of disastrous fishing seasons.

The changing ice conditions after 1920 also made the first navigation of the NE Passage without wintering possible in 1932 for the Russian trawler “Sibiryakow” [144]. Ten years later, in August 1942, the reduced sea ice extension prompted the German naval high command to order the heavy cruiser “Admiral Sheer” into the Kara Sea, to intercept suspected allied convoys from US and Canada with supplies to the Red Army, making use of the extraordinary open water conditions along the Russian and Siberian coasts at that time [145]. During this operation “Wunderland”, famed “Sibiryakow” was regrettably sunk by “Admiral Sheer”. The background for the German concern was the fact that a German raider, the “Komet”, without great difficulties passed through the Northern Sea Route to the Pacific Ocean in summer 1940 [146] [147]. However, the Allies never attempted to make use of the NE Passage during the war.

Special interest has often been attached to the Svalbard region, as this part of the Arctic apparently displays a high climatic variability, as was recognized by both Ahlmann [148] and Lamb [132], and later by IPCC [149]. Significant climatic variations in the Svalbard region during the 20^th century are well documented by meteorological data since 1911 [139] [150] [151]. As an example, a marked warming around 1920 within 5 years changed the mean annual air temperature at sea level from about −9.5˚C to −4˚C. This is one of the most pronounced increases in air temperature documented anywhere during the instrumental period. Later, from 1957 to 1968 the air temperature sharply dropped about 4˚C, with a subsequent more gradual increase towards the end of the 20^th century (Figure 13). The high climatic variability of the Svalbard region probably derives from three mechanisms. Firstly, the islands are located directly in the main transport pathway for air masses into the Arctic Basin [78] [152]. Secondly, Svalbard is located near the confluence of air masses and ocean currents with very different temperature characteristics. Finally, the variability is further enhanced by rapid variations in the sea-ice extent, coupled with both atmospheric and oceanic circulation [80] [153] [154] [155] [156].

Moving on to the period 1940-1965, in contrast to 1915-1940, this represents a period of cooling in many parts of the Arctic (Figure 13). Again, the temperature change was by far most pronounced during the winter and weaker during other seasons. Especially Alaska, NW Canada, Russia and parts of Siberia were affected by this development. The North Atlantic sector with Scandinavia at that time had relatively cold autumns, suggesting an increase of northerly winds. North Atlantic sea ice also increased in this period, and a greatly increased flow of the cold East Greenland Current brought polar water far south (cf. Figure 3(c)). In several years (1965 and especially 1968 and 1969) more Arctic sea ice came to the coasts of Iceland than seen for fifty years [137]. In April-May 1968 and 1969 Iceland was half surrounded by sea ice, which has not occurred since 1888. At the same time, the Greenland cod migrated to Iceland waters, and for a few years (1967-1971) offset the declining stocks there. The 1960s also saw the abandonment of attempts at grain growing in Iceland, which were resumed during the previous relatively warm period (1920-1950) after a lapse of some hundreds of years [137]. At the beginning of the 21^st century, grain (barley) was only grown at one farm near Skoga in southernmost Iceland, and only for making hay (H. Björnsson, pers.com. 2003).

The final ca. 30 years of the 20^th century has been a period of renewed warming in many parts of the Arctic, lending observational support to the common notion of ‘Global Warming’ [93] [149] [157] [158] [159] [160]. Again, this warming has been most pronounced in the winter season and less so in other seasons. The late 20^th century warming has affected especially western Canada, Russia and Siberia. This was accompanied by retreating sea ice in both the arctic North Atlantic sector, and in the Kara Sea, after reaching a maximum extension in the early 1970s (e.g. [161] [162]). A study by Winsor [163] concluded that the mean sea ice thickness in the Arctic Ocean remained constant from 1986 to 1997. Satellite mapping showed the late summer 2012 sea ice extension in the Arctic to be the smallest since initiation of satellite mapping in 1978. However, the sea ice expanded again in 2013 and 2014, and despite the season 2018-2019 showed the next smallest ice-covered area, there is no trend in September ice-cover 2007-2019.

An apparent paradox is that warming registered in the Atlantic water layer below the sea ice in the Arctic Ocean became noticeable in the early 1990s, whereas the major ice thinning occurred prior to 1990 [163], suggesting the dynamics of sea ice extension in the Arctic Ocean to be more complex than usually thought.

Polyakov et al. [164] proposed that the Arctic Ocean has experienced an “atlantification” (i.e. increased role of Atlantic inflow) during the last decades. This seems contradicted by Klyashtorin et al. [165] who documented an increased Atlantic inflow through the Faroe-Shetland Strait from 1910 to 1970, a fall to low levels in 1980 and continued low levels at least up to 2000. Humlum [56] recorded a drop in the North Atlantic Current temperature from 2004 to 2019 with a peak-low in 2016. This can hardly be combined with an “atlantification” in the last two decades (nor is the general decreasing trend in warm water to the Arctic since 2011; Figure 10(A)).

According to Polyakov et al. [164] “the role of Atlantic water heat in sea-ice reduction is not negligible”. Observations from the 1990s and 2000s documented two warm, pulse-like Atlantic water (AW) temperature anomalies on the order of 1˚C (relative to the 1970s), entering the Arctic through Fram Strait and occupying large areas of the Arctic Ocean. The strength of the 2000s warming peaked in 2007-2008, with no analogy since the 1950s (cf. [166]). This AW warming has slowed slightly since 2008. Ocean ventilation may have had pronounced effects on the ice edge position (a question to be returned to in Part-2 [136]).

Whilst the Arctic region, in general, records a warming during the last 25 years, the Eurasian region records a cooling [159]. According to Roy [93], this is the function of the weaker sunspot cycles in the last 40 years. The influence of the Sun on the Arctic winter climate goes via changes in Arctic Oscillation (AO). During winter solar minima, there is a warming over most of the Arctic (in December-February), especially over the Kara and Barents Seas, but a cooling over Siberia and Scandinavia [93].

Due to the pioneering work of Torgny Vinje [180] on assessing the location of the summer ice edge between Svalbard and Frans Josef Land (20˚E - 45˚E), based on log books of whalers, warships, early explorers, and scientist (e.g. [181]), and later airplanes and satellites, we have a record of the sea ice cover in this region, dating back to the 1580s. Falk-Petersen et al. [178] updated this time series to the present and related it to the primary and secondary production in the Whalers Bay north of Svalbard. They hypothesized that in periods with little ice the total bio-production in the European Arctic increased due to two main factors: a) the increase of the total area of open water available for primary production during spring and summer in the Barents Sea and Nordic Seas, and b) the winter upwelling of nutrient rich Atlantic water along the shelf break in the area north of Svalbard and Frans Josef Land fuels production hotspots in Whalers Bay as soon as the Sun returns. The high primary production attracts the lipid-rich Arctic grazing zooplanktons such as Calanus glacialis, Calanus finmarchicus and Calanus hyperboreus, the key species in the high latitude lipid-based ecosystems [169] [182].

The area north of Svalbard was called the Whalers Bay by the Dutch whalers, who carried out intense whaling here for more than 100 years. The peak of the whaling period, between 1660 and 1730, coincided with a period where the summer ice edge was north of 80˚N for extended periods. Interesting, the short period with the expansion of the ice edge between 1730 to 1750 also showed a reduced catch of the bowhead whale as demonstrated in Figure 14 showing the reconstructed aggregate catch [183]. Comparing Figure 5, Figure 12 and Figure 14, we find that the maximum catch of more than 3000 bowheads happened in 1701 (i.e. at the end of the Maunder Minimum) when there was a type-c ocean circulation with cold water all the way south to Portugal (Figure 5). In historical time, it was the period with most spring ice in the North Atlantic and the sea temperature along the Norwegian coast was so low that the cod did not reproduce.

The Whalers Bay in the Fram Strait is a high productive hotspot today, and we suggest that this also was the case in the period with the historical whaling from 1670 to 1790, and that the high primary and secondary production sustaining the large stocks of whales and seals in this area [178] [183] [184] [185]. Bowheads are slow swimmers that feed by filtering zooplankton out of engulfed water masses with their 4.5 - 5 m long baleen plates. The main prey is the energy rich Calanus copepods such as Calanus glacialis and Calanus hyperboreus, but also other lipid-rich zooplankters [186] [187]. Weslawski et al. [184] estimated that bowhead whales consumed as much as 4 million tons of plankton per season from June to September in the Svalbard area at the peak of their abundance.

The documentation on the “French Naval Operations in the Spitsbergen during Louis XIVs Reign” further illustrates the productivity of this area in the Dutch whaling period. The commercial value of whaling activity was high and the commercial output of interest for several European nations [188]. In 1693, the French king sent a fleet of warships north of Svalbard to take the Dutch whaling fleet as prizes. On the 6th August 1693, the two French frigates Agil and Favori encountered and took a fleet of more than 40 ships, all flying the Dutch standard, under fire at the mouth of the Hinlopen strait (La Baye aux Ours—today Sorgbukta). The French frigates sized 13 ships and sunk a large part of the Dutch ships. The remaining Dutch whaling fleet escaped by sailing northeast in the ice-free area in the Hinlopen Strait. Despite the losses, the Dutch ship owners still were able to make a profit. The Dutch whaling stopped in some periods due to naval conflicts in Europe, for instance 1665-1667, 1691 and 1710 [183].

After the intense Dutch bowhead whaling period 1680 to 1790, the sea ice expanded rapidly, and large parts of the Nordic Sea and the Barents Sea were covered by ice during spring and summer until 1910. Then the ice retreated rapidly until the 1940s, before the ice expanded again in the 1960s and 70s. In the long period with extensive ice cover after 1790, there was a short period with less ice around the 1850s and in the decade that followed. In 1873 the cod appeared on the banks of Spitzbergen, and a fishery was established the year after. The cod appeared in dense aggregations where the cod were feasting up on sea butterflies (Limacina helicina) [189] [190]. The fishery in this period was done by hand-line from small rowboats. The intense cod fishery off Svalbard (Spitsbergen) lasted from 1874 to1882 [189]. Then there was a rapid retreat towards north by the ice edge between 1910 and 1940 (Figure 15) and already in 1931 H. U. Sverdrup [181] with the submarine Nautilus did marine research as far north as 82˚N in the Fram Strait. In the 1930s there was an intense cod fishery off Svalbard with yearly landings of several million tons [191]. The Northeast Arctic cod stock in the Nordic Seas and the Barents Sea was strong during the mid-1950s, with a stock of about 4 - 5 million tons. However, during the cold 1960s and 1970s, with an expansion of the sea ice in the Barents Sea (see Figure 15), the stock was dramatically reduced to 1 million tons by 1964 [190] [192]. The cold period lasted to approximately 1985 in the Nordic Seas [169] [193], with persistent large ice cover in the European Arctic. This period falls together with a collapse both in the herring and in cod stocks in the Norwegian Sea and Barents Sea [190] [192]).

The expansion or contraction of distributional range of cod will often coincide with long-term temperature changes [194]. Northward expansion of the cod stock with increasing temperature and reduced ice cover has previously been documented during the warm periods of the late 1800 and during the 1920s and 1930s. During the warming event in the 1920s and 1930s and associated decreasing ice cover, the cod changed its distribution from southern Greenland up to Disko Island [195-197]. The cod was present as far north as the Disko Bay until the 1970s when they disappeared following a rapid decline in seawater temperature, with a slight recovery after the 1990s [197].

After 1990 there has been a rapid increase in temperature and decrease in ice cover in the European Arctic, followed by a large increase in the commercial fish stock in the Nordic Sea and the Barents Sea. In the Whalers Bay a mesopelagic layer of fish and zooplankton was recorded with an increased presence of mammals north of Svalbard [176] [185] [198] [199].

Figure 15 shows the revised BIE position estimates for the period 1579-2018 (black dots) and an 11-point running mean. Some important biological events are marked with numbers (Figure 15, points 1 - 11). The red curve documents the presence of all the major solar maxima/minima known for the last 440 years (Figure 4); viz. little ice in the late 16^th century solar maximum, ice expanding from 1600 to 1650 corresponding to the first half of the Maunder Minimum. The retreat of BIE in the second half will be discussed in Part 2 [136]. Then we observe less ice during the 18^th century solar maximum, expanding ice from 1780 to 1800 referring to the onset of the Dalton Minimum, less ice in a short period around 1850, and rapid retreat of the ice edge 1910-1940. Finally, we observe more ice 1940-1970, and a retreat 1970-2010. Since 2010, except for the record low ice in 2012, the summer ice cover has not shown any significant decrease (http://nsidc.org/arcticseaicenews/). We may also notice the rapid expansion of the ice in the years 1614-1616, 1664-67, 1690-95, 1780-85, 1864-69, and 1960-62, which may be due to special weather events. Apart from the cycles, the dominating part of the curve is the gradual movement of the BIE northwards for more than 200 years (1800-2018).

Another interesting point is that the ice minima in 1760 and 1940 are 180 years apart. This may be related to the José cycle of 179 years [200], which is the fundamental orbital repetition period of the Jovian planets and is also synchronized with the sidereal orbital periods of Venus, Earth, Mars and Jupiter as well as the Schwabe (11.1 yrs), Hale (22.2 yrs) and Gleissberg (90 yrs) periods [201]. The mechanism and phase lock of these events will be discussed in more detail in Part 2 [136]. Since the 179 yrs period is a stable period, the José cycle may confirm an ice minimum in 1580 and predict a future ice minimum in 2120.

There are a number of ocean oscillations systems like PDO (Pacific Decadal Oscillation), AMO (Atlantic Multidecadal Oscillation) and NAO (North Atlantic Oscillation), which sets the character of regional changes in oceanic environment. They are all included in the “stadium wave wheel” of Wyatt and Curry [202], indicating coupling and feedback interaction (Section 9.5 and Appendix 1). Easterbrook [44] [203] noted that glaciers on Mt. Baker in Washington State showed a regular pattern of advances and retreats, which matched the PDO variations. This has a bearing on the effects of changes in the North Atlantic (Gulf Stream, NAO, AMO) with respect to the changes in ice-edge position on the Barents Sea region.

The NAO index is defined as the normalized pressure difference between subtropical high and Icelandic low pressure. Its connection to the Arctic Ocean oscillation (AO) is well described by D’Aleo and Easterbrook ( [204], Fig. 11.16) indicating alternations between stages of zonal jet stream and warm positive NAO conditions, and meridional jet streams and cold NAO conditions in the North Atlantic. The NAO-index was increasing from 1895 and had maxima between 1908-1926 and 1985-1997 and negative values 1955-75. It is about 70 years between its two maxima. The Earth’s rotation (LOD) is negatively correlated with the integral function of NAO (INAO), and also with global sea surface temperature (SST) [65]. They [65] modelled SST based on LOD and INAO back to 1650 and found that the cold sea was dominating in the beginning of 18^th century, as shown in Figure 12, even if there were warm land temperature periods in Europe. Another anomaly was the period 1900-1930 with an increasing positive NAO and a cold SST [205]. Lansner and Pedersen [206] showed that in the warm period 1920-1950, it was only stations with little ocean influence, which showed warming, while the warming at the end of the century took place both at ocean affected and shielded stations. The result is a slight cooling for inland stations 1900-2010 measuring −0.03˚C/century, while ocean affected stations showed a warming 0.78˚C/century. This is interpreted as a sign of a slow ocean warming since the cold LIA.

The AMO index (Figure 16) is defined as the pattern of sea surface temperature variability when the linear trend is removed [207]. It has a 65 - 70 years’ quasi-periodic variation [208]. The updated AMO series in Figure 16 has a period of 67 years. The BIE position variation is roughly in phase with the AMO index. During the period 1900-1925 the SST was cold, AMO negative, and NAO posive, which shows that AMO is dominating over NAO [205]. Both Alheit et al. [209] and Drinkwater et al. [194] discussed the AMO variations and recorded its effects on fish faunas and North Atlantic temperature and ice-edge variations. Mette et al. [156] ] record a close correlation between AMO and changes in ocean temperature in northern Norway.

According to Klyashtorin and Lyubushin [210] and Klyashtorin et al. [165], the stocks of herring and cod in the Barents Sea fluctuate with a 60-year periodicity (the cods lagging by 8 - 10 years), which correlates with the changes in Arctic temperature, the ocean water temperature and the ice cover conditions in the Barents Sea, all exhibiting a ~60-year periodicity. Mörner [43] noted that the changes in ice cover reflect the “delivery of warm Atlantic water to the region” and that “the main source of heat delivered to the Arctic basin is warm water inflow from the North Atlantic Stream” as recorded by Nikolaev and Alexeev [211]. It agrees with the observed changes in Earth’s rotation [36] and Gulf Stream beat [37]. This is illustrated in Figure 17.

This implies that the BIE position variations may be synchronized with the movements of planets in the solar system and follow the grand solar maxima/minima alternations and the 60-yrs cycle during the last 440 years. This suggests a primary planetary—solar forcing (further discussed below, and in Part 2 [136]).

The interaction between Earth, the Sun and other planets in the solar system results in variations in the Earth’s orbit on many time scales. Secular changes in eccentricity, axial tilt and precession result in long climate cycles, also called Milankovitch cycles [221] [222], which in generally are held to be the main reason for the Quaternary Ice Ages. The rapid changes of cooling and warming periods observed are in phase with ¹⁰Be variations, however, which makes it more likely that solar activity is the main reason for the alternations between Ice Ages and Interglacial, between stadials and interstadials, and short-term changes in climate [48].

Analysis of solar activity in a 9400-year long record of solar modulation potential φ determined from cosmogenic radionuclides such as ¹⁰Be and ¹⁴C shows distinct periodicities of 88, 104, 150, 208 and 506 yrs [47]. They find excellent agreement between the long-term cycles and periods in planetary torque, suggesting a physical connection (see Figure 4). It should be noted, however, that they did not detect a 60-yrs cycle in their analysis.

An analysis of the Zürich mean annual sunspot number Rz (1749-2009 AD), the group sunspot number Rh, (1610-1995 AD), the Extended time series of Solar activity Indices (ESAI) (1090-2003 AD), and the simulated extended geomagnetic index aa (1619-2003) by Komitov et al. [223] show that in all series exists a strong cycle with duration 55 - 60 years in the 18^th and 19^th centuries. It is less pronounced during the end of the 19^th and the beginning of the 20^th centuries. In all series except Rz there is also strong and stable quasi 100 - 120 years and ~200 years cycles. In the Rz series a strong ~95 years period was present. Also quasi 40 - 45 and 30 years cycles are detected, but they are unstable during long time scales and have low amplitudes. An interesting difference is that the simulated aa-index shows a period 180 years while the Rh shows 202 years. Cliver et al. [224] found a strong correlation between decadal aa and global temperature values (r = 0.90) and conclude that 50% - 100% of the temperature increase since Maunder Minimum is due to solar variability.

The solar dynamo theory [225] explains that in the beginning of a sunspot cycle, the dynamo process transforms the Sun’s polar field into an internal toroidal field that increases during the maximum phase of the solar cycle, then affecting Rz, when completing the sunspot cycle the toroidal field transforms into a poloidal affecting the aa-index. However, analysing wavelet power spectra Duhau and Chen [226] find that Rz and aa undergo cyclical oscillations whose amplitudes vanished in 1923. After that the ratio between Rz and aa increases from 5.3 to 12.4 and the phase difference decreases from 0.12 to 0.05 radians. They find that the secular Rz modulation steadily rise until the next Gleissberg cycle maximum in 1971. The temporal change in the coupling function between Rz and aa is synchronous with a loss in the relationship between temperature and sunspot number in the period 1920-1955 [227]. For the secular changes in Rz 1844 - 2000 Duhau and Chen [226] determined the following periods: 85.5, 121.1, and 171.2 years.

Finally, it may be significant to note that Blanter et al. [228] gave 4 possible explanations to the low values of the Solar Cycle 23: it may mark: 1) the beginning of a New Grand Solar Minimum, 2) the beginning of a new Gleissberg cycle, 3) the end of a great maximum episode, or 4) a random variation.

In order to forecast future variations, it is necessary to identify stationary oscillations. This was done for the first time for TSI variations and group sunspot series by Yndestad and Solheim [92], who searched for stationary components using the wavelet autocorrelation method. They found stationary components of the solar variability controlled by the 12 years Jupiter period and the 84 years Uranus period and its 3/2, 5/3 and 9/2 harmonic components giving periods 125, 210, and 373 years. The TSI record of Hoyt and Schatten [229] showed only the stable periods of Jupiter, Uranus and Neptune.

According to Kuznetsova [230], the changes of the terrestrial climate during the last millennium are driven by the interaction of Jupiter, Saturn, Venus, the Sun and the Moon.

Le Mouël et al. [231], analysed the Ri series by singular spectrum analysis (SSA) and found 4 components: a multi-secular trend with a slight acceleration around 1900, two Hale periods of 11.0 and 10.0 years, the fist increasing in amplitude, the second with secular amplitude variation with maximum around 1740 and 2010, and a clean Gleissberg period component which varied from 85 to 94 years with a decreasing amplitude—signalling a longer period beat. The multi-secular trend confirms the modern grand solar maximum from 1920 [3] [232].

Analysis of the last 4000 years of the reconstructed central Greenland surface temperature, derived from the bi-decadal 𝑑¹⁸O record from the Greenland Ice Sheet Project II (GISP2) [233] [234] revealed periods ≈3500, ≈1130, ≈790, ≈580 and ≈370 years [235]. A 3-period model based on non-linear optimization of the longest periods resulted in a good fit that clearly shows the timing of historical warm periods ( [236], their Fig. 8).

A global temperature harmonic analysis [236] based on paleotemperature reconstructions and instrumental records for the period AD 1 to 2015 gave three significant periods: ≈1000, ≈460 and ≈190 years in addition to a less significant period ≈60 years. A 190-year period was also found modulating the length of the sunspot cycles ( [237] [238]).

Harmonic analysis of the HadCRUT4 global temperature (1850-2012) [239] showed that a linear trend of 0.5˚C per century and periods 9.1, 21.6 and 66 years. A 9.3 years period is a harmonic of the 18.6 years lunar tide period. This period is also identified as a standing wave in the Atlantic Ocean by Mazimov and Smirnov [240] [241]. The 21.6 years period is close to the Sun’s magnetic period, the Hale period of 22 years.

During solar cycles with negative polarity of the Sun’s northern polar field, the Galactic Cosmic Ray (GCR) variation has a peaked form. In the other phase it has a plateau. This is an effect of the differences in cosmic ray drift in the positive and negative phases of the magnetic cycle. Integrated GCR counts are higher in plateau cycles compared with peaked cycles [242]. This may be the reason for the amplification of the 22 years component in the global temperature curve. The difference between 11 and 22 years climate response, is also seen in the latitudinal difference in the rhythm of growth in pine trees, (as shown in [239]; Fig. 6), where the 20 years period dominates north of 65 degrees latitude, while the 10 years period dominates at lower latitudes. This may be a result of differences in atmospheric circulation or effects of cosmic rays of lower energies reaching deeper at higher latitudes. For Svalbard at 78˚N, an analysis of temperature series by Humlum et al. [88] shows that periods 17 and 26 years are much stronger than those at 10 - 12 years.

Kasatkina et al. [243] showed that dendrochronological data from the Kola Peninsula and Finnish Lapland recorded the effect of low-latitude volcanic activity with “significant decrease in polar tree-ring growth over 7 years after eruption”, and “the main cycles of solar activity (11, 20 - 25 and ~100 years)”.

The 66 yr period can be explained as a harmonic solar (3 * 22 = 66 yr) or lunar (3.5 × 18.6 = 65.1 yr) period. A maximum entropy power spectra analysis by Scafetta [244] showed that 9 different global and regional temperature series contained the same 10 spectral cycles from the Sun and the Planets, in addition to a 9.1 years period from the Moon.

Scafetta [245] demonstrated multi-frequency spectral coherence between planetary and global surface temperature oscillations by advanced techniques and statistical significance tests, with large coherence peaks of 20 and 60 years quasi-periods. The latter has maxima around 1880, 1940 and 2000.

Periods mentioned above may well be present in global time series, but the sensitivity of the local (high latitude) climate is different from low latitude sensitivity, and the combination of thermal-salinity-cryospheric interactions, also involves lunar and orbital forcing, acting different at seasons. A framework for the solar-lunar-orbital interaction for high latitudes is presented by Cionco et al. [246] and demonstrated in Figure 19. The latitudinal temperature gradient (LTG) drives the (wind driven) ocean circulation between the tropics and the polar regions. Analyses of recent changes in LTG suggest that the Earth’s climate system is unusually sensitive to changes in LTG [247]. LTG varies according to seasons. In the NH summer the variation in obliquity is dominating with 41 Ka orbital and 18.6-yr lunar cycle. In NH winter the TSI is dominating with 11yr cycle [90] [246]. The LTG is clearly modulated by the TSI variations [219] and represent a channel both for solar and lunar forcing of Arctic climate.

Treloar [95] identified that lunar-solar forces produce multiple tidal ocean effects with periods of 86.795, 46.22, 13.53, 18.02 and 20.295 years. Lunar Nodal oscillation components of ~56 and ~74 years are observed in the North Sea, Kattegatt and Baltic tide gauge data [248] [249].

An analysis of time-series for the North polar position, extent of Arctic ice, sea level, sea temperature, air temperature, and the NAO winter index shows that a harmonic spectrum from the 18.6-year lunar nodal cycle is present in all time-series [86] [96]. The cycles in the harmonic spectrum have stationary periods but not stationary amplitudes and phases. A sub-harmonic cycle of 74 years may introduce a phase reversal of the 18.6-year cycle.

Correlations are also found between the lunar nodal tide effect and temperature and salinity in hydrographic time series for the Faroe-Shetland channel and the Kola Section, suggesting that the temperature and salinity variations are forced by harmonics of the nodal tide [96] [250]. Dominating periods were the 18.6 yr lunar nodal period and its 9.3 year harmonics.

Davis and Brewer [90] stated: “empirical evidence suggests that extraterrestrial forcing influences the Earth’s climate” (a notion that we certainly agree with), and they presented data in the effort of “unifying orbital, solar and lunar forcing”.

Figure 19 shows the summer insolation at 65˚N during the last 2000 years [246]. It gives the combined effect of variable TSI and orbital variations due to lunar and planetary cycles, including solar variability. The rapid decline 1350-1400 coincides with the start of the Little Ice Age (LIA), which had its maximum cold period 1687-1703, according to the longest instrumental temperature record from Central England [251]. Recovery from this cold period was first visible in the Arctic from about 1920 as demonstrated in Figure 12 and Figure 15. The summer insolation at 65˚N increased in three steps after 1700 with the recent 1920-2000 maximum being the highest. After 2005 the June 21 TSI at 65˚N is decreasing with a minimum in 2045 and a recovery at a lower level about 2085.

It is interesting to notice that a major reorganisation in atmospheric circulation took place about 1420. During the preceding Medieval Warm Period, winter storminess across the North Atlantic was at a minimum. This is confirmed in Na⁺ deposits in the GISP2 borehole in Greenland [252]. After 1420 until about 1925 cells of winter Icelandic low pressure have been strongly anchored within northern hemisphere circulation, which has resulted in increased frequency and severity of North Atlantic winter storms. This corresponds to a period of reduced June 21 solar Insolation at 65˚N (Figure 19). A reduction in winter storminess is documented for Grimsey and Stykkisholmur weather stations in Iceland after 1920 [253].

The Earth’s magnetic field is modulated by the solar wind, which varies in phase with the solar activity or sunspot cycles, with an average period of 11.06 years. At maximum solar activity the shielding capacity of the Earth’s magnetic field against cosmic rays from exploding stars is maximum, and this reduces the number of cosmic rays that hit the Earth. Since cosmic rays may be related to formation of clouds [254], this has a climate effect. We can therefore expect less clouds and higher temperatures at solar maxima. The variations in the solar wind amplify the TSI-effect on the Earth’s climate, and longer periods with reduced solar wind intensity are in general cold periods on Earth (cf. Figure 4).

A linear relation between sunspot numbers and the rotation of the Earth, or the length of the day (LOD), shows that increased solar activity or strong solar wind decelerates the Earth’s rotation (Fig. 2 in [37]). This affects the oceanic and atmospheric circulation systems (cf. Figure 9), both of which have a significant effect of terrestrial weather and climate. The relations among solar activity, irradiance, solar wind, variations in the Earth’s atmospheric shielding and variation in LOD are visualized in Figure 20.

The changes in Earth’s shielding capacity are recorded in the changes in atmospheric ¹⁴C production. Figure 4 gives the changes from AD 950 to 1950. The production is modulated by a 210-yr periodic fluctuation, which corresponds to the 208-yr period observed by Abreu et al. [47] during the last 9400 years. The cycle had its maximum in 1910 and a minimum in 2005. This signals the end of the warming period.

The shielding capacity is a function of the solar wind, which brings magnetic plasma, which strengthens the Earth’s magnetic field. Weaker solar wind means less shielding. 30 years of output from a data-constrained magneto-hydrodynamic model of the solar corona has made it possible to model solar wind variations over the last four centuries [45] as shown in Figure 21 for the equatorial component that hits the Earth. The global solar wind variations show main evolution in four zones: 1) the Maunder Minimum of lower solar wind speed values, 2) the 18^th century of generally higher values, 3) the Dalton Minimum of lower values, and 4) the last 180 years of surprisingly similar values.

Figure 22 shows solar wind measured by spacecraft in geocentric orbits, made available through the omniweb.gsfc.nasa.gov website, and observations at the Oulu neutron count detector. Neutron counts are a good measure of cosmic galactic rays (GCR) particles and are clearly anti-correlated with the solar wind. The figure shows that the solar wind has decreased more than 30% from the peak in 2004. Since 1990 the neutron counts (GCR) has increased and the solar wind speed decreased indicating that Earth’s shielding capacity has decreased considerably since 1990 (Figure 22).

Bucha [75] demonstrated that increased corpuscular radiation (as recorded by the neutron counts) implied enhanced penetration of electrons into the middle atmosphere and stratosphere, resulting in increased temperature and atmospheric pressure in the stratosphere and troposphere in the auroral oval. This leads to a change from a meridional to a zonal type of atmospheric circulation over the Arctic. Visa versa, low corpuscular magnetic activity leads to low temperatures in the stratosphere. Further effects on the Solar Wind interaction with the terrestrial system are given in Figure 9 (cf. [43] [63] [64]. Kuznetsova [230] identified a 198 yrs magnetic field cycle and a 54 yrs solar wind cycle in terrestrial temperature data.

A closer study of the geomagnetic activity over the past century of the solar wind flow speed Vsw, the interplanetary magnetic field strength B, and the open solar flux Fs, indicate that the grand maximum started in 1920, peaked around 1955 and 1986 and then declined. Extrapolating the linear decline trends indicate that the grand maximum of the three parameters will end in 2013, 2014 and 2027, respectively [232].

The equatorial current lags behind the rotation of the solid Earth, which lead to an accumulation of warm water-masses outside East Asia rising local sea level by about 30 cm. At El Niño events, 0.4 ms of the rate of rotation (LOD) is transferred from the solid Earth to the hydrosphere, which is manifested in an eastward migration of the water-masses from East Asia towards the west coasts of the two American continents. When the water hits the west American coasts, sea level rises by about 30 cm, and 0.4 ms in rotation (LOD) is transferred back to the solid Earth. This provides a significant documentation of the interaction and feedback coupling between the redistribution of mass (water) and changes in rotation with the interchange of angular momentum between the solid Earth and the hydrosphere [36] [43].

Mörner [255] noted that the Gulf Stream exhibited a pulsation of 16 cycles within the Holocene, recorded in high-resolution deep-sea cores, sea level changes in the Kattegatt-North Sea region, and temperature changes in Northwest Europe. He interpreted this in terms of a feedback coupling between water masses redistributed from low latitudes to middle-to-high latitudes, and changes in Earth’s rotation in order to compensate for this. We now know, that the primary driving force is to be found in external planetary interaction with the Sun and the Earth (Figure 9; [37] [43] [63]).

The Gulf Stream beat system proposed by Mörner [37] has a close relation to the alternations between NAO+ and NAO− stages recorded by Wanner et al. [256]. Their main illustration (Fig. 9 in [256]) has been endorsed and reprinted by Pinto and Raible [257] and Boström [258]. Whilst the Gulf Stream beat theory has a driving energy input from external forces (Figure 7 and Figure 9), the NAO theory of Wanner et al. [256] is based on internal interaction and feedback process. It seems significant that Wilson (2011) concluded that he had presented facts, indicating, “that the phases of two of the Earth’s major climate systems, the North Atlantic Oscillation (NAO) and the Pacific Decadal Oscillation (PDO), are related to changes in the Earth’s rotation rate”. As shown by Mörner [36] [37] the changes in rotation originate from planetary-solar beat as illustrated in Figure 9, i.e. from external sources (Figure 7).

In his account on “the Lunar nodal cycle influence on the Barents Sea”, Yndestad [259] [260] presented both an internal and an external view of the climate feedback loops of the Arctic; i.e. the coupling and interaction between temperature, water, ice extent and light reflection in the Arctic region of the Atlantic sector.

He noted that his first lunar view model was good to illustrate the fundamental dynamics operating with respect to feedback loops in an autonomous dynamic system. This model represents the internal interaction and feedback processes. He noted that this model was not enough, but called for external oscillating forces.

In his external forced oscillators model, he included solar/orbital influx (from the Milankovitch variables), Moon orbital forces (tides and rotation) and CO2 effects.

Today, we know that the Sun varies in quite another way, and that the planetary beat effects the Sun, the Earth and the Earth-Moon system (Figure 9).

Yndestad’s thesis [259] [260] is deeply relevant, because it at the same time as it shows the fundamental interaction and coupling between different internal parameters, it also calls for external forcing functions (today in improved understanding, however).

The polar position oscillates with periods 6, 18, and 74 yrs cycles, all related to the 18.6 lunar nodal cycle.

Yndestad [259] explains that in the cold Arctic Ocean, water accumulates in circulating layers, the Arctic Surface Water, the Atlantic Water, the Deep Water, and the Bottom Water, all of which have different densities and circulation patterns. The Atlantic Water residence time is about 25 years, while the Deep Water residence time is about 75 years, which is the same period as the polar motion. His model is that the gravity force from the 18.6 years lunar nodal cycle, influences the polar position and most likely the circulating water in the Arctic Ocean, the Arctic ice extent, and Arctic atmospheric condition. A continuous source of energy from the stationary polar position cycle is transformed into a spectrum of oscillating water circulation periods, where some harmonic cycles are preserved by resonance. This is the base for “The wine glass theory”, that the rotating Arctic Ocean behaves like liquid in a rotating wine glass, where the Deep Water oscillation has a resonance of about 74.5 years. The denser Bottom Water has a resonance of 4 × 74.5 = 290 years, while the less dense Atlantic water has a resonance 74.5/3 = 25 years.

Yndestad [259] presents the detection of lunar nodal oscillation or harmonics in all Arctic climate time series analysed (Barents Sea ice extent, Kola section ocean temperatures, Hammerfest tidal variations, Greenland Sea ice extent, Røst winter temperature, and NAO winter index). He concludes that these stationary cycles may enable us to predict the future Arctic climate.

Wyatt [261] and Wyatt and Curry [202] have presented a quite remarkable “Stadium-wave” mechanism of a special and temporal wave of 4 segments propagating across the Northern Hemisphere activating a network of synchronized climate variables. They identified four temporal groups (I - IV) changing between negative and positive phases. It offers a synchronization and feedback coupling between ocean, ice and wind factors. In Appendix 1, it is shown that the “stadium wave” mechanism of Wyatt and Curry [202] can be combined with the planetary-solar modulation, Gulf Stream beat and the external rotation forces here discussed.

In this paper, we acknowledge that the mechanism seems to imply a logical manifestation of “terrestrial changes in climate & environment” in Figure 9, but calls for external driving forces (as indicated in Figure 7 and Figure 9).

The Gulf Stream circulation system is strongly driven by the main alternations between Grand Solar Maxima and Minima (Figures 1-6, Figure 8, Figure 12, Figure 19). During the Maunder and Dalton Minima, the inflow of warm Atlantic water to the Arctic was strongly reduced, and cold Arctic water reached all the way south to central Portugal (Figure 5). In the GIN Sea region, severe climate conditions and periods of extensive ice expansion occurred (Figure 12). Both periods are well recorded in the Solar wind intensity record of Figure 21 (zones 1 and 3). The solar maximum of the 18^th century, is recorded in strongly intensified inflow of warm Atlantic water to the Arctic, high solar irradiance at 65˚N (Figure 19), high values of solar wind (Figure 21, zone 3), and generally warm climate conditions and strongly reduced ice extension in the GIN Sea region with the April maximum ice edge of year 1769 located even north of Svalbard (Figure 12).

A particular problem with this interpretation is that the maximum catch of bowhead whales took place in the period 1685-1703 (Figure 14) when the coldest climate in the millennium was recorded in Europe, and the global sea surface temperature was 1.0˚C - 1.5˚C cooler than in the 1950-1980 period [65]. From Figure 15 and Figure 21 it is clear that the BIE moved north at least 50 years before the recent high solar wind level was reached. Gulf Stream circulation type-a dominated the 18^th century ( [63], Fig. 7a in [64]). The water north of Svalbard and in the Fram Strait was apparently warm enough for rich marine life. This seems to call for a local source of heat or special wind pattern, maybe recorded in the weak positive NAO-index in 1705-1740 [263].

Another problem is that the high level of solar wind through the 19^th century did not move the BIE north. This did not happen until the start of the solar Grand Maximum in the 1920s, and was accompanied by a rapid temperature increase at many Arctic stations (Figure 13). One possible explanation will be explored in Paper 2 [136]: The Sun was stressed after the producing short sunspot-cycles resulting in the 18^th century maximum, and it relaxed by producing long cycles in the 19^th century. Only at the start of cycle 16 (1900) it was “refreshed” and able to produce short cycles [238]. This pushed the BIE north in the 20^th century, still modulated by the planetary cycles.

In the BIE position estimates (Figure 14) we find minimum ice 180 years apart, corresponding to the José period of alignment of the planets in the solar system. It may also seem significant that Yndestad and Solheim [92] were able to document the planetary beat of Jupiter, Saturn, Uranus and Neptune in TSI and sunspot datasets in addition to the recorded 210-yr (de Vries/Suess) cycle, which is the predominant cycle in the ¹⁴C production record of the last 9400 years [47].

The most prominent feature in the BIE record (Figure 15) are the two deep minima 1630-1660 and 1800-1900 and gradual move north after 1900. A series of maxima 180 years apart suggest a modulation following the José cycle of orbital synchronization. Since this is a sum of three 60-year cycles we may also find the 60-year cycle in the data (to be searched for in Paper 2 [136]).

A 60-year quasi-periodic cycle is detected in global temperature (Figure 19; [208] [239] [262] [264] [265]), in TSI [219], in LOD [36] [43] [266], in geomagnetism [267] [268], in Jet Stream changes between zonal and meriodional circulation [101] [269] [270], in sea level changes [271] [272] [273] [274] and ocean oscillations systems like PDO, AMO and NAO (Figure 18; [204]).

Scafetta (e.g. [244] [266]) and Solheim [239] showed that the 60-yr cycle had an origin in the combined 60-yr beat of Jupiter and Saturn. Consequently, we must place planetary beat—in this case the combined beat by Jupiter and Saturn—at the top of the forcing (Figure 9). This beat simultaneously affects the Sun, the Earth and the Earth-Moon system. The solar variability generates changes in luminosity (irradiance) and Solar Wind, the last one of which controls a number of fundamental Earth’s processes as illustrated in Figure 9.

The planetary beat effects on Earth’s rotation and the Earth-Moon system imply that there must also be a generation of tidal effects. It therefore seems significant that Yndestad was able to identify the 18.6 and 74.4 tidal cycles in different Norwegian Sea parameters [84] [250] [259] [260], and Hansen [248] was able to document the 55.8 yrs and 74.4 yrs tidal cycles in the North Sea and Kattegatt records.

Cyclic forcing functions imply that we would have means of predict changes in the near future. The deterministic model analyses of sunspots and TSI recorded by Yndestad and Solheim [92] suggest that we will have a new Dalton-type minimum by about 2025-2050 and 2040-2065, respectively. This fits well with other estimates of a New Grand Solar Minimum; e.g. at about 2030-2050 [37] [39], starting in 2014 and reaching its minimum in 2055-2060 [275]), in 2030-2040 [244] [276], and around 2040 [165].

The 60-yr cycle (Figure 19) and the lunar cycles must interact with the Gleissberg (~84 years) and the de Vries/Suess (208 years) cycles as assumed in the concept of a New Grand Solar Minimum peaking some 20 - 30 years in the future (above). The effect of the positive trend of a millennium long cycle [235] should also be considered. It seems significant, however, that the combined Gleisssberg-de Vries cycles provides a cyclic trend which agrees well with the observed changes in ¹⁴C prediction and the Gulf Stream beat between type-a and type-c circulation patterns ( [38] [39]; Appendix 2).