Temperature, carbon dioxide and methane

7 1) Globally-representative monthly rates of change of atmospheric carbon dioxide and methane are compared with 8 global rates of change of sea ice and with Arctic and Antarctic air temperatures. 2) Carbon dioxide is very strongly 9 correlated with sea ice dynamics, with the carbon dioxide rate at Mauna Loa lagging sea ice extent rate by 7 months. 10 3) Methane is very strongly correlated with sea ice dynamics, with the global (and Mauna Loa) methane rate lagging 11 sea ice extent rate by 5 months. 4) Sea ice melt rate peaks in very tight synchrony with temperature in each 12 Hemisphere. 5) The very high synchrony of the two gases is most parsimoniously explained by a common causality 13 acting in both Hemispheres. 6) Time lags between variables indicate primary drivers of the gas dynamics are due to 14 solar action on the polar regions, not mid-latitudes as is conventionally believed. 7) Results are consistent with a 15 proposed role of a high-latitude temperature-dependent abiotic variable such as sea ice in the annual cycles of carbon 16 dioxide and methane. 8) If sea ice does not drive the net flux of these gases, it is a highly precise proxy for whatever 17 does. 9) Potential mechanisms should be investigated urgently.


23
The atmospheric levels of carbon dioxide have risen during the instrumental record (IPCC 2013

30
The seasonal cycle of carbon dioxide is typically ascribed to the cycles of terrestrial productivity on the large land

42
Temperature drives ice melt and should thus be very highly correlated with the monthly rate of change of these 43 greenhouse gasses -whether or not sea ice is involved in the cycles. We test this prediction here.

45
Temperature is conventionally believed to drive the annual cycles of methane and carbon dioxide through changes in 46 vegetation and microbial productivity, including agriculture (IPCC 2013). Yet despite great efforts, there is substantial uncertainty in the locations and magnitudes of sources and sinks for these gases (Kort et

53
The locations of sources and sinks of carbon dioxide have traditionally been estimated using 'inversions' and 54 'atmospheric transport' models which rely on climate models to reverse-engineer from observed gas levels where with the global rate -with least temporal lag between timeseries of the gas rate and its local driver and with co-62 varying annual amplitudes. Of course, any seasonal variables such as livestock activity or wetland productivity or 63 sea ice extent will have correlations with methane and carbon dioxide seasonality, but the spatial pattern of lags 64 between timeseries can help identify the more likely causes and locations. For example, a polar causal variable 65 should have a relatively high correlation and low lag with a positive gas flux near the pole.

67
Terrestrial productivity in the Northern Hemisphere is typically measured by NDVI (Keeling et    4 (driven mainly by the annual cycle of solar elevation). We hypothesize high-latitude air temperatures drive sea ice 80 dynamics and snow dynamics and thence might influence greenhouse gas dynamics. Such strong relationships are 81 not presented in the review of the carbon cycle that informs international climate policy (IPCC 2013) and could focus 82 greater attention on high latitude sites and fluxes.

84
Stable isotope ratios in carbon dioxide have been used to attempt to explain the seasonal variation in carbon dioxide

86
The monthly 13 C/ 12 C ratio co-varies closely and inversely with carbon dioxide in many recording stations, typically  crystal formation which causes a lighter isotopic composition in expelled carbon dioxide whilst 13 C is preferentially 95 included in precipitated carbonate (Niles et al 2007). We therefore examine the relationship between sea ice melt 96 and freeze rates and the proportion of the heavy isotope in the atmosphere (δ 13 C).

99
We use the datasets in Table 1

111
Pole to some sites at lower latitudes.

113
We do not use models of atmospheric transport of gas but instead make the minimalistic assumption that gas from a 114 polar region will take longer to reach or cross the equator than to reach nearby sites. We assume the shape of the 115 seasonal gas flux curve will be most similar to the curve of the causal variable near the site of the causal variable 116 (due to mixing).

125
Recording of the OH radical in the atmosphere is very difficult, so is usually done indirectly using methyl 126 chloroform (CH3CCl3) which OH reacts with and hence lowers the atmospheric concentration (Ravishankara &

289
Air temperature at or near the Poles peaks in very close synchrony with regional peaks in sea ice melt (Figs. 10 and 290 15). It will also be correlated with a range of other abiotic and biotic variables with various lags, such as Northern

291
Hemisphere snow (Fig. 9) and Greenland terrestrial ice melt. Air temperature at high latitude sites leads the global carbon dioxide rate with a greater lag of carbon dioxide behind the Antarctic than the Arctic temperature ( Fig. 2 and 293 Fig. 3).

295
The rates of change of globally representative levels of carbon dioxide and methane are very strongly correlated with 296 the rate of change of global ('Arctic plus Antarctic') sea ice (Figs. 1 and 4) on the timescales examined. The rate of 297 change of methane at Mauna Loa has similar phenology but greater amplitude (Fig. 5). At the South Pole, methane 298 rates are very highly synchronous with Antarctic sea ice extent rates (Fig. 16), as are other regional methane rates 299 (Hambler & Henderson 2020b). The lag of 5 -7 months between the peak Antarctic temperature (and sea ice melt) 300 and the fastest decline of global methane and global carbon dioxide suggest a strong Antarctic influence on these 301 gases ( Fig. 1 and Fig. 4). It may take months for the effects of temperature on gas flux in the Antarctic to reach the 302 Northern Hemisphere.

304
The extremely strong predictive power of global total sea ice for carbon dioxide and methane is notable -revealing 305 possible causality or high predictive power for the actual cause. The two peaks in global sea ice rate result from the 306 peak temperatures in the two Hemispheres. Global carbon dioxide and methane rates also have twin peaks which are 307 similarly separated (Fig. 6). We propose that whatever dominates the fluxes of these gases makes strong

312
Temperature in at least one Arctic recording site has a close synchrony with carbon dioxide (Fig. 7) and methane atmospheric temperature in the Arctic summer (July, Fig. 7). This is also synchronous with peak decline in Arctic 317 ice extent (Fig. 10). However, peak negative methane flux at Alert (Fig. 8) occurs about one month earlier than peak 318 temperature and peak sea ice melt in the whole Arctic, which we suggest results from an influence of the biota or 319 other abiotic factors on methane dynamics in the Arctic. Arctic sea ice as a whole can not be the dominant causal 320 variable in this region at least, but there are regional differences in sea ice phenology, and Alert methane peak decline is more closely synchronous with the Barents Sea ice rate (Hambler & Henderson, unpublished). Peak rate 322 of decline of Arctic methane is also closely synchronous with peak snow extent decline in the Northern Hemisphere,

323
with Alert lagging snow melt rate by about a month (Fig. 9), consistent with putative terrestrial influences such as 324 increased methanogenic microbial activity. Peak methane emission from Arctic mires can occur near peak summer 325 air temperature (Jackowicz-Korczyński et al 2010).

327
Peak negative methane flux at the South Pole is synchronous with peak temperature at the South Pole ( Fig. 13) but 328 carbon dioxide rate at the South Pole lags one month behind the peak temperature which occurs December / January 329 (Fig. 12). Similarly, methane rates slightly lead carbon dioxide rates globally and at Mauna Loa (Figs. 4 -6).

338
The synchronous decline and rise in carbon dioxide and methane at many sites would most parsimoniously be 339 explained by a single mechanism. These results are broadly consistent with our proposals that sea ice is either 340 involved in the decline of atmospheric carbon dioxide and methane or is extremely strongly corelated with an 341 unknown variable causing fluxes of the gases (Hambler & Henderson 2020a, b). We argue the extremely high 342 correlations between sea ice and fluxes of both gases are more plausibly due to simple physical or chemical 343 processes than to ecological ones (Hambler & Henderson 2020a, b). In particular, we suggest the peak negative gas

372
Measured by NDVI, terrestrial productivity has relatively weak synchrony and curve shape similarity with carbon 373 dioxide rates, in any large region, even with lags ( Fig.18

407
rather than temperature, and carbon dioxide levels, rather than rates.

409
Isotope ratio rate clearly co-varies closely with sea ice rate (Figs. 19 -21) although these are not analysed statistically 410 since the isotopic time series is available for only part of the timeframe that we consider for carbon dioxide and 411 methane. During sea ice formation in either hemisphere the ratio of 13 C to 12 C declines in the atmosphere; this is 412 consistent with degassing of carbon dioxide enriched in 12 C, as demonstrated experimentally (Niles et al 2015).

413
Notably, the transition dates between positive and negative rates are near synchronous for δ 13 C and Arctic sea ice 414 volume (Fig. 19). The visual similarity between isotope rate and sea ice rate is closer for Barrow (Fig. 20)

421
In contrast to these near-simultaneous changes of rate of sea ice and isotopes, the northern photosynthetic rate of

451
The monthly timeseries of sea ice extent we use (Table 1)

483
The current paradigm for the carbon cycle is supported by weaker correlations than the paradigm we propose. In