Parametric Study of Prompt Methane Release Impacts II: Eﬀect of a Dynamic Ocean on Model Results

,


INTRODUCTION / BACKGROUND
An important event in 2020 was the clear lack of ice cover over the shallow Kara Sea and Laptev Sea continental shelves at the end of October [1,6] -a necessary antecedent to a hypothesized continental-shelf clathrate methane-release runawaywarming feedback (https://www.mdpi.com/2076-3263/9/6/251/htm)[2,13,14] (not well-addressed in AR5).Although the current state of coupled Atmosphere/Ocean General Circulation Models (AOGCMs) is such that they probably cannot reliably predict resulting weather-pattern changes in detail, a potential lower-bound for the overall impact of such releases can be explored on a decadal scale; impact can be a dominant factor in risk-assessment, in turn important for governance.
- ------------------1.A PREVIOUS ESTIMATE Previous work (Figure 1., below) [3] with a prescribed-ocean, atmospheric circulation model suggested a ~0.01 C global land surface air temperature increase per gigaton of additional atmospheric methane burden.The inclusion of a coupled dynamic ocean model in the present work yielded a tripling of that estimate to about 0.03 C global warming per gigaton increase in atmospheric methane burden.Additionally, much larger Arctic responses were observed when specifying model changes chosen to approximate increased cloud brightness over the Arctic ocean.In this work, the AR5 RCP8.5 GHG scenario was chosen as the baseline in order to most faithfully represent current and anticipated trends over the relatively short time period investigated here (i.e., 2020 -2040).
After an extended model spin-up, the model global mean temperature was constant for two decades of integration using a constant model-year (2019), and closely reproduced the mean warming rate reported for 2020 -2040 in the AR5 data when run in baseline RCP8.5 transient-mode over the subject time period.As is typical for the AOGCM used in this work (AR5 GISS ModelE2.0 07.50.01), the modeled Arctic sea-ice losses were less than is currently observed in the real-world; hence, the results presented here are regarded as no more than a lower bound to the magnitude of expected Arctic response to such a release.

2.A IMPACT / PROBABILITY
This work seeks to more clearly define parameters and impacts of a hypothesized methane release to support rigorous assessment of Risk (i.e., Probability x Impact).Colloquially, it is desired to know whether the hypothesized methane release is of Probability-class suggested for supervolcano eruptions and large asteroid/comet strikes (or a magnitude-9 earthquake off the coast of Sendai, Japan); or if the probability is more akin to the likelihood of the large family of climate events well-anticipated as the Anthropocene deepens.However, if the Impact of the hypothesized methane release is sufficiently large, then it warrants more careful study and consideration (i.e., IPCC WG-I) than it has received to date, independent of Probability estimates.This work studies only Impact, and briefly examines one proposed mitigation, Arctic marine cloud brightening (https://royalsocietypublishing.org/doi/pdf/10.1098/rsta.2014.0053)[4][5][6][7], for the purpose of maintaining the equatorial/polar pressure gradient force (mitigating Arctic sea-ice melt).This work models the years 2020-2040, with the a primary methane release occurring in the years 2025-2030.[8,9] The model was spun-up for 300 years with 1880 climatology (E_AR5_V2_NINT_oR.R template, master_yr=1880), then was restarted as a dynamic simulation from 1880 to 2020.The resulting 1JAN2020 dataset was then used as initial conditions (IC) for subsequent runs.To assess spinup, this IC was restarted for a steady 20 year run (master_yr=2020) to quantify the modeled global temperature "drift" over the time period examined, and then restarted as a dynamic case under RCP8.5 conditions to validate against the mean AR5 modeled temperature increase trends from 2020-2040.(see Figure 3.) Repeat integrations of runs following large changes in methane or insolation essentially retrace.The results of this validation suggest that the model numerics [15] are running well enough for estimation of planetary response to a large-scale methane release.
Atmospheric methane is herein presumed to have a mean half-life of 9 years, converting mole-for-mole to CO (moles of H O thus produced are ignored).

It must be noted that the modeled methane-release scenarios are only approximate because, among other reasons, uncertainties of methane dynamics
in the real-world, and because the methane here is always well-mixed and globally dispersed -there is no explicit modeling of the real-world time needed to disperse methane globally from regional sources, temperature-and concentration-dependent reaction rates, solar radiation effects, etc.This compares favorably with the model result for this time period in this work (Figure 3.).

2.C SCENARIO METHANE INPUTS
The methane GHG input temporal profiles for the data presented here are shown in Figure 5. Obviously, there is no precedent to know, a-priori, the temporal profile of such a release, and it depends on many factors; e.g., associated with as-yet uncharacterized reinforcing feedback loops.A possible feedback could facilitate a transition to new methane emission rates on the order of 1 gigatons of methane per year in the near-term, e.g., 45NR and 20NR, from unspecified sources.A methane-pulse-only scenario (51P) was included, which suddenly halts after 2031 and decays back toward RCP8.5.The latter may seem somewhat unlikely given the magnitude of such an event.
Note that these scenarios are based on atmospheric methane burden, not on quantities of methane released, as-measured at the Arctic continental shelf

3.A DATA SUMMARY
Global modeled annual mean (2032 -2040) surface air temperature increase above modeled RCP8.5 scenario, as a function of atmospheric methane burden: ✲ ~0.03 C mean increase per gigaton methane burden.
✲ ~0.2 C/year typical initial increase rate.The annual mean global surface air temperature data for all scenarios are shown in Figure 7., and warming continues after the methane has stopped increasing, or decreases strongly.In previous work with an atmosphere-only climate model (https://doi.org/10.1002/essoar.10503094.1)(prescribed oceans) it was found that land surface air temperature rise nearly halted when methane release halted.[3] Although not shown, the land surface air temperature increases in this work were significantly greater than in the previous study.These results suggest a strong role for the dynamic oceans in the current model by magnifying and introducing inertia into the land surface and global temperature rise.

ARCTIC SCENARIO RESULTS / CONCLUSIONS
Arctic mean surface air temperature increase.Y-axis 1-σ error bars represent combined annual model variation and the observed upward drift of each scenario over that time period.
The single X-axis 1-σ error bar represents the exponential decrease in methane burden as the 51P scenario decays back toward RCP8.5 (see Figure 5.).Temperature rise is roughly double that of the mean global surface air temperature rise.
The mean annual Arctic surface air temperature data are significantly more scattered than the mean global data, with roughly double the mean increase.ModelE2.0 over-predicts Arctic ice volume (RCP8.5)by about 50%; this ice-bias suggests that these results likely constitute a lower-bound to scenario-associated Arctic temperature increases.

DISCUSSION/CONCLUSIONS
The spin-up of the AOGCM used here was validated with a steady-state drift test and with the RCP8.5 scenario global mean surface air temperature change over the time period of interest.The scenarios remain stable and appear to produce numericallyconsistent integrations.[15] The temporal profile of a methane release from shallow Arctic clathrates is unknown, so two types of release were investigated and produced similar results: Global and Arctic temperature changes are somewhat crude indicators of climate change impacts on global civilization; however, these are all that can be presented at this time.Work is ongoing to study mesoscale impacts of these scenarios (e.g., regional changes in precipitation and jets), but no strong trends have yet emerged.Some regional trends could be inferred from these data in conjunction with AR5/6/paleo temperature results, and in the absence of verified models of our scenarios, this currently seems the only option.However, this should likely be done with some care.
Comparable simulations with other AR5/6 codes should be done for consensus and increased confidence.Important AOGCMdependent unknowns include: (1) methane sensitivity and impact of finite dispersion and destruction rates, (2) the dynamic response to methane-release timing with respect to the observed low-frequency (multi-year) global cycles, and (3) the transition from Arctic ice-loss to ice-gain conditions associated with Arctic cloud brightness.

ACKNOWLEDGMENT
The decades of study and effort by the many scientists and contributors, whose hard work and persistence has made it possible to model the weather and climate processes of Spaceship Earth in some detail, is gratefully acknowledged.12/18/20, 11:48 AM Cloud processes are among the least well-simulated aspects of current-generation AOGCMs, especially considering extreme events such as modeled here.The impacts of Arctic cloud brightness are explored as a sensitive aspect for the modeling of diminishing summer sea-ice coverage and similar runaway feedbacks in the Arctic, such as snowline retreat (http://eprints.whiterose.ac.uk/138040/).[6] On the other hand, considered from the point of view of Marine Cloud Brightening, [6,7] these observations on marine clouds are confined to the Arctic, and the Arctic only; strongly invoking a Precautionary Principle (https://www.mdpi.com/2071-1050/12/21/8858/pdf), while still aiming for a key impact in the imminent climate emergency (https://report.ipcc.ch/sr15/pdf/sr15_spm_final.pdf).[10,12] Of course, such an effort is probably misguided unless the IPCC SR-15 [2018] recommendations are implemented; i.e., large scale-scale halting of carbon emissions during the time period modeled here.[12] 4.A EXPERIMENTAL Arctic cloud brightness is studied here in an approximate and minimally-intrusive way by noting that, although dependent upon cloud microphysics, the dominant energy-balance effect of brighter cloud cover is the rejection of visible insolation.The AOGCM's cloud-process algorithms themselves are not modified, only top-of-atmosphere (TOA) insolation as proxy for cloud brightness.The region of insolation-modification is illustrated in Figure 8.

4.B SPATIAL DATA
A substantial increase in area with mean annual surface air temperature at -20C (Figure 9

4.C TEMPORAL DATA
If cloud brightness is parametrically increased beginning in the year 2030 for scenario 45NR, the temporally-averaged, mean annual Arctic ocean surface air temperature responds by decreasing (Figure 10).Temporal trends in the results are shown in Figure 11.Significant changes occur if the fraction of insolation removed increases above about 0.2, after which Arctic air temperature may continue to decrease beyond the time period investigated.Although not shown, removing 20% of the Arctic ocean insolation in scenario 45NR produces modeled Arctic ocean annual ice-volume gains -yielding volumes above baseline RCP8.5 ice volumes for 2030-2040.However, it must be pointed out that ModelE2.0 over-predicts RCP8.5 Arctic ice volume by about 50%, so a quantitative estimate of the transition from ice-loss to ice-gain is not yet possible.12/18/20, 11:48 AM

Figure 3 .
Figure 3. Mean annual surface air temperature data.Blue/Diamonds: constant 2020 model run showing annual data and smoothed data (dashed line), with low global mean temperature drift; Red/Squares: transient model run (AR5;RCP8.5GHG scenario) showing annual data and smoothed data; showing a linear-fit mean global temperature rise of 0.272 C per decade.This compares well with estimates from the ensemble-mean AR5 (Figure 4.) data, which was estimated at 0.31 C per decade for this time period.

Figure 6 .
Figure 6.Mean global annual surface air temperature increase above mean RCP8.5, for given approximate methane burdens, averaged over the years 2032-2040.Y-axis 1-σ error bars represent combined annual model variation and the observed upward drift of each scenario over that time period.The single X-axis 1-σ error bar represents the exponential decrease in methane burden as the 51P scenario decays back toward RCP8.5 (see Figure 5.).Other X-axis error bars are not shown due to uncertainty in real-world methane dynamics.The 2-σ lines are fitted to the combined Y-axis 2-σ points of the dataset.Scenarios show continuing upward temperature drift (see Figure 7.).

Figure
Figure 7. Detrended annual mean global surface air temperature data points for all scenarios (with the RCP8.5 trend removed, since it is
increases in mean global surface air temperature are seen when rapid, large methane increases are modeled: ≥0.03 C per gigaton of methane.~1.3 C decrease in mean annual Arctic air temperature for each 10% increase in "cloud brightness" (4.A; scenario 45NR) Mean surface air temperatures due to a CH burst are likely to continue to rise beyond 2040..comhttps://agu2020fallmeeting-agu.ipostersessions.com/Default.asp... 9 of 15 12/18/20, 11:48 AM The modeled methane-responses probably represent a lower bound for real-world changes, at least for the Arctic.The remarkable Arctic ocean increase in ice volume associated with 20% brighter Arctic clouds deserves further investigation.

4 .Figure 8 .
Figure 8. Illustration of the region of reduced annual insolation over Arctic ocean (see text).It is assumed that changes in the Arcticocean surface lead to increased cloudiness (or else intentional cloud brightening efforts), confined to the Arctic ocean and 68 N − 90 N.

Figure 10 .
Figure 10.Mean Arctic annual surface air temperatures, scenario 45NR, averaged from 2032 -2040 as a function of insolation fraction (model proxy for increased cloud brightness; i.e., decreased insolation = increased cloud brightness).Error bars include both trend and scatter of data points in Figure 11.
15 of1512/18/20, 11:48 AM Figure 11.Arctic mean annual surface air temperature data trends, scenario 45NR, with insolation fractions as-noted on the curves.Evidently, for insolation fractions less than about 0.8, the trend changes sign from increasing to decreasing, over the time period modeled here.14 of 15 12/18/20, 11:48 AM