A conservative outlook on 21st century climate change
Summary: For the past decade anthropogenic emissions have slowed down, and continuation of current trends suggests a peak in emissions by 2050. Atmospheric CO2levels should reach 500 ppm but might stabilize soon afterwards, as sinks increase their CO2uptake. Solar activity is expected to continue increasing after the present minimum, as the millennial cycle works its way towards a late 21st century peak. The reduction in the rate of warming might continue until ~ 2035 followed by renewed warming, and temperature stabilization at about +1.5°C above pre-industrial. The pause in summer Arctic sea ice melting might also continue until ~ 2035. Renewed melting is probable afterwards, but it is unlikely that Arctic summers will become consistently ice free even by 2100.
“The future is unknowable, but the past should give us hope.”
Winston Churchill. 1958.
Karl Popper’s falsifiability criterion for science requires that hypotheses not only explain known evidence, but also must be testable by evidence still unknown. However, a problem arises when any failed ex-ante prediction made by a hypothesis, can be post-hoc explained in multiple ways leaving the hypothesis nearly intact. An example is the pause in global warming that took place between 2001-2013, while accelerated warming was the CO2-hypothesis outstanding prediction for the 21st century (IPCC-FAR, 1990). The pause was explained in multiple ways (see: Nature Climate Changevol. 4, issue 3, 2014, and Nature’s“Focus: Recent slowdown in global warming“). To meet Karl Popper’s scientific criterion, the CO2hypothesis of climate change must make predictions that cannot be post-hoc explained when they fail. When demanding urgent action on CO2emissions, the predictions that can falsify the hypothesis are being made for a period ending in 2100, more than 80 years away. Its falsifiability is being removed until it no longer matters for present policy decisions.
This article deals with scientific forecasting of future climate change and its consequences. As with any other activity, forecasting has been the subject of systematic studies, and three of the foremost experts in forecasting principles have established the golden rule of forecasting: “be conservative by adhering to cumulative knowledge about the situation and about forecasting methods”(Armstrong et al., 2015). Research has shown that ignoring the guidelines deduced from the golden rule greatly increases forecasting error. However, climate forecasting is dominated by radical predictions, many of which are absurd, yet they are given disproportionate positive attention. Two of the authors (Green & Armstrong, 2007) analyzed the IPCC-Fourth Assessment Report, concluding that its forecasts were not the outcome of scientific procedures, but “the opinions of scientists transformed by mathematics and obscured by complex writing,” and warned that research on forecasting has shown that experts’ predictions are not useful in situations involving uncertainty and complexity. Previous research by Philip E. Tetlock had already demonstrated that expert forecasting is usually worse than basic extrapolation algorithms, and that there is a perverse inverse relationship between fame and accuracy in forecasting (Tetlock, 2005). J. Scott Armstrong went further and in 2007 challenged the IPCC prediction of 3°C/century (IPCC-TAR, 2001) with a no-change forecast for the next 10 years (2008-2017). Using the UAH dataset and judging by cumulative absolute error, the no-change forecast reduced forecast errors by 12% compared to the IPCC projection, showing that the IPCC projection had no value, since it was beaten by a no-change forecast (Climate tipping alarm vs scientific forecasting).
The first eight articles in this series analyzed the cumulative knowledge about climate change necessary for conservative forecasting.
- The Glacial Cycleis necessary to understand our interglacial evolution and the role of Milankovitch forcing.
- The Dansgaard-Oeschger Cycleresearches the causes and consequences of the most abrupt climatic changes in the past.
- Holocene Climate Variability (Aand B) analyzes climate change during our interglacial.
- The 2400-year Bray Cycle (A,B, and C) describes the major solar cycle that has determined important climatic shifts in the past, impacting human societies.
- The 1500-year Cycleadds a much studied and little understood non-solar climate cycle, of a proposed tidal-oceanic origin, and its relationship to the Dansgaard-Oeschger cycle.
- Centennial to Millennial Solar Cyclesfills the gap of shorter solar cycles, of which the millennial cycle shows a big climatic effect on the Early and Late Holocene.
- Climate Change Mechanismsanalyzes the way the planet responds to different climate forcings and internal variability, including the 60-year oceanic oscillation that has characterized climate change for the past centuries.
- Modern Global Warmingdescribes the features of climate change since the Little Ice Age and the perceptible effect of anthropogenic forcing for the past seven decades.
With that knowledge we can attempt to conservatively forecast future climate change for the next decades. After all, if a no-change forecast can beat IPCC forecasts, it should be a lot easier for knowledge-based conservative forecasts. The starting premise for the conservative forecast presented here, derived from past and present climate change evidence, is that greenhouse gases (mainly CO2), solar variability, and oceanic oscillations, are all significant climate variables in the centennial timeframe considered. It is important to remark that forecasts only consider a very limited number of variables and the rest is assumed asinvariant. This necessary simplification means that with increasing time the chance of a forecast being correct decreases even if the variables considered were correctly projected. The future is, after all, unknowable.
Changes in CO2emissions and atmospheric levels.
Atmospheric CO2levels are very likely to continue increasing for the next 30 years, even if changes in the rate of emissions take place. Due to the large size of natural stores, sinks, and sources, the trend in atmospheric CO2levels responds slowly to changes in emissions. A 2009 decrease in emissions due to the financial crisis is not perceptible in atmospheric CO2levels. A conservative forecast by extrapolating the average increase in CO2values for the past 10 years gives 491 ppm of CO2by 2050. Slightly higher values could be reached if the observed acceleration in the rate of increase of CO2values is maintained (figure 111), but this acceleration has been decreasing with time and is currently very small.
Regarding CO2emissions, the failure of past projections shows how difficult it is to forecast future emissions. It is very easy to extrapolate fossil fuel consumption that has experienced continuous growth for over a century, but several factors are very likely to have a significant impact in fossil fuel production for the 2018-2050 period, making a simple extrapolation a non-realistic forecast. The UN population forecast shows that there is a profound and inevitable demographic change taking place (United Nations, 2017). The UN medium-variant projection shows every region except Africa reaching peak population by 2050. The aged >60 population is the fastest growing group and by 2050 all regions of the world except Africa will have at least a quarter of its population above age 60. Population demographics suggests a growing negative pressure on per capita energy use, whose increase has been driven in the 21st century by the growing Asian middle class. For countries with high dependency ratios of old people (% >64 / working age), what it is observed is a decrease in per capita energy use with time (figure 116). China’s former one-child policy is going to turn its demographic dividendinto a demographic burden very fast. China’s working population has already peaked in 2010, and by 2040 it should have a dependency rate, old, similar to Japan.
Figure 116. Declining energy per capita for countries with aged population. Primary energy consumption (tons of oil equivalent) per person, for the world (black line) and the three countries with highest dependency rate, old (number of people above 64-years old per 100 people in working age. Italy, dark grey; Japan, medium grey; Finland, light grey). Japan’s population has been declining since 2010, Italy’s population since 2014, while Finland’s population is still growing. Source of data: BP Statistical Energy Review 2018, and World Bank.
Besides an increasing fraction of older people, population decrease should also reduce the total energy demand. From the supply side, coal production is showing an unexpected lack of growth, and oil production is generally expected to peak within the 2018-2050 period for a variety of reasons, including reducing energy return on energy invested (EROEI, manifested in increasing costs of production), energy transition mainly to natural gas, but also to alternative energy sources, and active global policies to reduce oil and coal burning.
Because of these and other economic factors, our CO2emissions have been growing more slowly for the past six years, having already reached its lowest 5-year average value since the early-1990s, over 20 years ago (figure 117, purple line). Our emissions are growing now at a rate like RCP4.5 (figure 117, black line). If the present trend slowing continues, a decrease in CO2emissions should start before 2050 and should continue for the foreseeable future, unless demographic trends change. The slowing in emissions rate started years before the Paris Agreement, but there is little doubt that its continuation will be fully attributed to its success.
Figure 117. Global CO2emissions have almost stalled. Yearly percentage rate of increase in global CO2emissions from fossil fuels and industry (blue bars, LHS), and its 5-year average (purple line, LHS). Global CO2emissions from fossil fuels and industry, in gigatons of CO2(black line, RHS). CO2emissions considered by the four IPCC emissions scenarios in the four representative concentration pathways (red, orange, light and dark blue lines, RHS). Since 2011 our emissions have been growing at a similar rate to RCP4.5. Source: Boden, T.A., Marland, G., & Andres, R.J. (2017). U.S. Department of Energy. Data for 2015-17 from BP Statistical Energy Review 2018, adding cement and flaring contribution estimates. RCP from IPCC AR5.
What would happen to atmospheric CO2levels under a likely decreasing emissions scenario from 2050? This scenario is similar to RCP4.5 that shows stabilizing atmospheric CO2at ~ 500 ppm (van Vuuren et al., 2011; table 2). However, carbon sinks have been a considerable source of positive surprises to climate researchers. First, it was the “missing sink”(Schindler, 1993), since it could not be explained where the emitted CO2that did not remain in the atmosphere was going. Environmentalists were slow to accept that the biosphere was expanding and greening in response to increasing CO2and warming, despite the opposite effect being well-documented during glacial periods. Then climate scientists became worried that the land (Canadell et al., 2007) and ocean (Schuster & Watson, 2007) carbon sinks were saturating. However, the opposite has been found, and sinks are actually increasing their rate of uptake (Keenan et al., 2016). If in the 1960’s they were taking up ~ 40% of our CO2emissions, they are now taking up ~ 55% of our much larger current emissions (figure 118; Hansen et al., 2013).
Figure 118. The decreasing airborne fraction. The airborne fraction (light blue) is the fraction of our CO2emissions (red, in gigatons of carbon) that remains in the atmosphere each year. The 7-year mean (dark blue) shows how over time a smaller part of our much larger emissions remains in the atmosphere. Source: J. Hansen, et al. 2013. Environ. Res. Lett. 8, 011006. Updated by J. Hansen.
The reason why sinks are taking up more CO2from the atmosphere is that we are farther from equilibrium. Since atmospheric CO2changed very slowly before anthropogenic emissions from fossil fuels, it can be assumed that sinks (K) and sources (S) were at equilibrium at 280 ppm (ΔK = ΔS). Due to warming the oceans release ~ 16 ppm/°C, so current equilibrium is ~ 290 ppm. Since the current level (~ 400 ppm) is above equilibrium level, sinks are larger than sources (ΔK > ΔS), and the the farther we are from equilibrium, the larger the difference between sinks and sources (ΔK–ΔS). If we stabilize emissions (E) near present levels, as current trend suggests, the difference between sinks and sources will continue increasing until it matches emissions (ΔK–ΔS = E), reaching a new equilibrium for constant emissions. Since we are ~ 120 ppm above equilibrium and sinks are absorbing 55% of our emissions (ΔK–ΔS = 0.55E), it can be calculated that for constant current emissions the new equilibrium lies at 220 ppm (120/0.55) above the present equilibrium value of 290 ppm, or 510 ppm.
Given constant emissions at present levels, atmospheric CO2should increase logarithmically towards 510 ppm, at which point sinks should match sources plus emissions (ΔK = ΔS + E). One of the biggest mistakes in the climate change debate is assuming that we need zero emissions to stabilize CO2levels. Deep ocean carbon stores are so large that carbon sinks can be considered unlimited in terms of anthropogenic emissions. The planet has dealt with much higher perturbations of CO2atmospheric levels in the past, as supported by the large δ13C excursions associated with the formation of large igneous provinces that formed over tens of thousands of years. The IPCC hypothesis predicts that under constant emissions there should be a constant increase in atmospheric CO2levels and the airborne fraction of anthropogenic CO2should increase as sinks saturate. However, after 10 years of stabilizing CO2emissions it should become apparent that the airborne fraction of fossil fuel CO2is decreasing and the rate of increase in total atmospheric CO2is slowing down. Once more we are poised for another positive surprise by carbon sinks.
Fossil fuel CO2emissions are growing more slowly (figure 117) and there is the possibility that they will decrease in a few decades. Once our emissions decrease, atmospheric CO2will start slowly decreasing, as sinks and sources equilibrate to our decreasing emissions.
Fossil fuel changes.
The biggest part of our CO2emissions is due to the burning of fossil fuels, and a cursory analysis of future fossil fuel production is required for a more accurate forecasting of future CO2levels.
Coal production reached a maximum in 2013 and although it is increasing again (BP Energy Review 2018; figure 119 a), it is still below 2013 levels. The decline in coal production has been completely unexpected. It is probably helped by an increasing trend in coal plant retirements reaching 30 GW/year (Shearer et al., 2017; figure 119 b), as many coal plants are very old. There are plans to increase the number and capacity of coal plants worldwide and coal production should increase again, but coal plant implementation rate has been low lately (37% for the period 2010-2016), with most plant projects halted, cancelled, or shelved. From January 2016 to January 2017 the amount of coal power capacity in pre-construction planning decreased from 1,090 to 570 GW (Shearer et al., 2017). We cannot discard a higher future coal production, because coal reserves are abundant, but as coal is being increasingly substituted by gas and other energy sources, it is unlikely that coal production will return to the vigorous growth of the 2002-2011 period. The unexpected drop in coal production from 2013 to 2016 increases the chances that Peak Coal could take place several decades earlier than forecasted. ExxonMobil Outlook for Energy 2018places Peak Coal in 2025.
Figure 119. Peak Coal in 2013.a)Coal production 1981-2017 in million tons for the world (black line), China (orange line), OECD (blue line), and the rest of the world (grey line). Source: BP Statistical Energy Review 2018. b)Annual coal plant retirement 2000-2016 in megawatts. Source: Shearer et al., 2017. Boom and Bust 2017 Report.
Oil is the least abundant fossil fuel. Several signs indicate we are approaching the end of oil growth (Peak Oil). Oil is categorized by its density (specific gravity), and with time the proportion of light oil from tight formations, liquid condensate from natural gas, and heavy oil, has been growing at the expense of more desirable intermediate density oil. It is also clear to anybody that we would not have to resort to fracturing shale rocks with high pressured water to obtain low-producing wells that decline by 75% in just three years, if we could still get sufficient oil by more conventional methods.
The impression that peak oil is approaching is confirmed by analyzing the oil growth curve. For the past 30 years oil growth has been declining from ~ 2% to ~ 1% (figure 120). This is a period when oil production has not been constrained, and more oil could have been produced if more demand for it existed. The decline can be attributed to a multitude of factors, including global economy growth rate, increasing oil use efficiency, economic changes that reduce energy intensity, demographic changes, and active policies to reduce oil consumption. If this long-term trend continues, peak oil is expected to be reached ~ 2065 when oil growth should cease, but linear trend extension is a poor way of forecasting. While it is hard to imagine realistic scenarios that would invert a 34-year trend towards slower oil growth, there are several scenarios that might accelerate Peak Oil. Obtaining oil from more difficult geologic formations leads to a higher cost oil that, to be sustainable over the long term, must be adequately reflected in oil price, and should promote oil substitution. The net energy yield of our global oil operations is decreasing, becoming a less efficient, less competitive process. Concerns over CO2emissions are also driving oil substitution with policies that for example support the increase in electric vehicles.
Figure 120. Decrease in the rate of change of world oil production. Ten-year average of the percentage change in oil production between 1983 and 2016 with its linear trend. The time of the peak in conventional oil and the shale revolution are indicated. Source of data: BP 2018 Energy Review.
The conservative forecast for oil production proposed here is in stark contrast to every single official projection by the International Energy Agency, the U.S. Energy Information Administration, British Petroleum Statistical Review of World Energy, or ExxonMobil Outlook for Energy, as none of them projects a decline in world oil production for the next decades. Therefore, it is fair to ask if it is really conservative to predict a Peak Oil within the next three decades. After all the business-as-usual projection has proven superior so far despite repeated claims of impending Peak Oil in the past. Each one must decide about that and I have exposed the reasons that lead me to believe Peak Oil is a conservative prediction. I am also certain that no official projection will ever anticipate a decline in production as they cannot afford to be wrong on that. They will only see a decline after a decline is taking place, and therefore have no predictive value. The decline in coal production is an example. It was never predicted before it took place, but it is predicted in some scenarios afterwards.
If Peak Oil does take place before 2050, the lack of oil growth will have to be compensated by other energy sources, if global energy consumption is to continue growing unaffected. While our response to Peak Oil might be to increase our coal consumption, it is reasonable to assume that we will also substitute it by gas and other energy sources, leading to a decrease in CO2emissions. Climate change scenarios that consider GHG emissions must factor in the almost inevitable reduction in our CO2emissions during the 21st century to avoid being unrealistic.
Changes in solar activity.
Forecasting solar activity has proven difficult. There is currently no known mechanism that can explain long term solar-variability, and accurate prediction beyond the next cycle has not been demonstrated so far. Cycle forecasting takes advantage of the presence of repeating features like solar extended minima every ~ 100 years (centennial lows) identified from past activity, despite not knowing what causes them. Humans predicted seasons thousands of years before they could explain them.
Solar activity has been increasing for the past 300 years according to sunspot observations and solar proxies (figure 121, trendline). The centennial solar cycle defines three oscillations (C1-C3, figure 121) delimited by the lowest sunspot minima every ~ 100 years. The last oscillation, C3, has the highest average sunspot number (93.4 ss/year). This period, and particularly between 1935-1995, has been termed the modern solar maximum (Kobashi et al., 2015). We are currently in a centennial extended minimum in solar activity between C3 and C4, that has been proposed to be named the Eddy solar minimum. The cycle-based forecast indicates it should affect mainly SC24 and SC25 with increasing solar activity afterwards. SC25 has already been forecast to be intermediate between SC24 and SC20 by the reliable polar fields method (Svalgaard, 2018), suggesting cycle forecasting is correct. My cycle-based solar activity forecast was made in 2016.
A cycle-based forecast for 2018-2050 must consider the centennial periodicity. Analogues for SC25-27 are SC6-8, and SC15-17. Also, since long-term solar activity is increasing within the millennial periodicity towards its 2050-2100 maximum, SC25-27 should have higher activity than SC15-17, as that period had higher activity than SC6-8 (figure 121). The forecast therefore is:
– SC25 should be slightly above SC24, but below SC23.
– SC26 should also be above SC24, and probably above SC23.
– SC27 should be similar to SC22.
Figure 121. Sunspot forecasting based on solar activity cycles. International annual sunspot number 1700-2016 in black, with rising linear trend. The centennial periodicity represented as a sinusoidal curve with minima at the times of lowest sunspot numbers, defining the centennial periods C1-C3, with their span indicated by the dates below. Horizontal bars mark the average sunspot number for each period. C3 shows the highest solar activity of the three. C2 was affected by the presence of a bicentennial (de Vries) cycle low at SC12-13. Forecasted solar activity for 2017-2090 is shown in blue. C4 is set to coincide with a peak in the millennial Eddy cycle identified from Holocene solar proxy records, and likely to have more sunspots than C3 despite another de Vries cycle low expected for SC31-32. SC1, SC10, SC20, and SC29 constitute lows in the pentadecadal solar periodicity, that reduces sunspot numbers at the peak of the centennial periodicity. Source of data: SILSO, Royal Observatory of Belgium, Brussels.
From 2018-2035 solar activity is forecasted to be below average. From 2035-2055 activity should increase inversely to the 1980-2000 decline (figure 121). SC29 should have somewhat reduced solar activity due to the pentadecadal periodicity that also affected SC20 and SC10. From 2080 solar activity should decrease due to the bicentennial (de Vries) periodicity that affects solar activity a few decades in advance of the centennial minima.
In summary, 21st century solar activity should be a little higher than 20th century solar activity, due to being at the peak of the millennial Eddy solar cycle. This level of solar activity corresponds to the Holocene highest 25%, and no doubt is contributing to the present warm period. Lower than average solar activity should only take place in the 2006-2035 period. The Sun should promote warming during the 2035-2100 period but should reach maximal millennial activity during the 2050-2080 period.
A mid-21st century solar grand minimum is highly improbable
A mid-21st century solar grand minimum (21stC-SGM) prediction breaks the golden rule of forecasting, as it is a non-conservative prediction. Surprisingly a high number of well-known authors that question the CO2hypothesis have embraced the 21stC-SGM hypothesis. In the famous for the wrong reasons first issue of the Pattern Recognition in Physicsjournal, N.-A. Mörner and eighteen more authors signed a letter (Mörner et al., 2013), stating a conclusion and two implications that challenged IPCC interpretation of climate change and triggered the termination of the journal by its owners. Of interest here is the second implication that served as a basis to doubt IPCC claims:
“Several papers have addressed the question about the evolution of climate during the 21st century. Obviously, we are on our way into a grand solar minimum.”
I am sorry to disagree with such a long list of prominent scientists, but I just can’t find any convincing evidence for a 21stC-SGM, that would justify such a non-conservative forecast. Abdussamatov (2013), probably the first to write about this issue in Russian in 2007, mistakes the current centennial low with a bicentennial one, and ignores the modulating effect of the 2400-year Bray solar cycle on the bicentennial (de Vries) cycle amplitude. But the consensus dissolves upon scrutiny, as many of those authors have not published on the issue, and Scafetta (2014), and Charvátová & Hejda (2014) actually don’t predict a 21stC-SGM, but a tame run-of-the-mill centennial low. Charvátová foresees nearly identical activity for SC24-26 as for SC12-14, very far from SGM low values. Mörner, the first signatory, is so convinced of the coming 21stC-SGM that he doesn’t present any evidence in his numerous articles about the issue (Mörner, 2011). Salvador (2013), and Shepherd et al. (2014) rely on different models for their 21stC-SGM prediction, the first a tidal torque model, and the second a solar dynamo model. Salvador’s model disputes the approaching millennial peak in solar activity, with a projection of 160 years of very low solar activity, while the highly publicized by the UK press dynamo model from Zharkova’s group doesn’t adequately hindcast past solar activity. Together with Steinhilber & Beer (2013) evidence-based forecast, they all have the problem that the very reliable polar fields method has already forecasted a SC25 with more activity than SC24 (Svaalgard, 2018), contradicting their proposed continuous decline in solar activity towards the predicted SGM.
A conservative forecast that a SGM is not going to take place doesn’t need supporting evidence, as the Sun only expends ~ 17% of its time in SGM conditions (Usoskin et al., 2007), so the chances are skewed against it. However, an analysis of what is known about SGM builds a strong case against the 21stC-SGM hypothesis. 30 SGM have been identified during the 11,700-year Holocene based on a very high rate of cosmogenic isotopes production (Usoskin et al., 2007; Inceoglu et al., 2015; Usoskin et al., 2016; figure 122). The average is one SGM every 400 years, but SGM show a tendency to cluster. 17 SGM (57%) are at around two centuries of another SGM, and 7 clusters of 2 or more SGM can be recognized. That is why the de Vries bicentennial cycle is so important for SGM as it is a very favored spacing. Usoskin et al (2016) have shown that SGM have a statistically significant tendency to cluster at the lows of the 2400-year Bray solar cycle, challenging random-probability based analyses (Lockwood, 2010). If we also consider the 1000-year Eddy solar cycle, we can see that 26 SGM (87%) fall at or right next to the periods when one of these two cycles is at its lowest 20% (red areas in figure 122, 54% of time). The conclusion is clear, SGM tend to occur nine out of ten times when solar activity is at its lowest coinciding with the lows of the ~ 2400 and ~ 1000-year solar cycles. In periods like the present, outside the lows of both solar cycles, the Sun expends only 7.5% of its time in a SGM, and with a frequency of ~ 1 SGM in 1000 years.
Figure 122. Solar Grand Minima distribution during the Holocene. Thirty Solar Grand Minima (SGM) from solar proxy records during the Holocene, identified in the literature, are indicated by black boxes of thickness proportional to their duration. The ~ 2450-year Bray solar cycle (black sinusoidal) and the ~ 980-year Eddy cycle (red sinusoidal) identified in solar proxy records are displayed at their proposed time-evolution that best matches both solar activity and climate changes consistent with their periodicity. Periods where any of the cycles are at the lowest 20% of their relative sinusoidal function are marked in red, and comprise 54% of the Holocene. Present position is indicated by a dashed line. SGM show a bias towards clustering at the red areas. RWP: Roman Warm Period. DACP: Dark Ages Cold Period. MWP: Medieval Warm Period. LIA: Little Ice Age. MGW: Modern Global Warming. Source for SGM data: Usoskin et al., 2007; Inceoglu et al., 2015; Usoskin et al., 2016.
Forecasting a SGM for the mid-21st century is really a low-probability non-conservative proposition. I would expect the probability for a next SGM to become high again ~ 2450 AD.
Changes in global surface average temperature anomaly.
Over periods of a few years climate variability appears to be dominated by ENSO variability (figure 123), that so far has resisted forecasting attempts. The failure of the 2014 and 2017 El Niño forecasts (figure 124), shows the difficulty of forecasting ENSO and global temperatures even for a few months.
Figure 123. ENSO-Global temperature relation. June 2013-January 2018 Niño 3.4 region sea surface temperature anomaly (red, LHS), and monthly global surface average temperature anomaly (black, RHS). Sources: Australian Bureau of Meteorology and UK HadCRUT4.6 dataset.
Figure 124. February 2017 failed El Niño forecast. Niño 3.4 region sea surface temperature anomaly forecast from 1 February 2017 by the European Centre for Medium-range Weather Forecast. El Niño conditions are considered above +0.5°C anomaly. After only 7 months the outcome was completely out of the anomaly plume even though it extended over a range of 2.5°C. Source: ECMWF Seasonal System 5 Public Nino plumes.
2018 is already poised to be less warm than 2017 given winter conditions and the ENSO situation. The Pacific Decadal Oscillation (PDO) index is in decline and back to predominantly negative values since July 2017. The conditions that drive a negative PDO make a La Niña more likely in the near future. If a La Niña finally develops in 2018-2020 we should see a continuation of the short-term cooling trend that started in February 2016. By 2018-20 the solar minimum between cycles 24 and 25 should have some effect on global temperature. A decrease of ~ 0.2°C in global average surface temperature has been measured from solar maximum to solar minimum (Tung & Camp, 2008), so a similar effect is expected from 2014 to 2019 due to solar forcing alone. The combined effect of ENSO, oceanic oscillations, and lower solar forcing from the 11-year cycle suggest that the temperature decrease should continue at least until 2020. By then the global average anomaly should reach values close to the 2003-2013 average (the Pause), and below the linear trend from 1950 (figure 125).
The effect of multiple solar cycles with lower than average activity on temperatures is not well known, but previous similar periods, known as solar extended minima, coincide with cooling periods (Gleissberg, Dalton, and Maunder minima). A lag of ~ 10-20 years has been found between the decrease in solar activity and its effect on tree-ring growth and ice core temperatures in several reconstructions (Eichler et al., 2009; Breitenmoser et al., 2012; Anchukaitis et al., 2017). A longer lag has been found on the maximum effect of solar activity reduction on the increase in heat transport, causing cooling in low latitudes and warming in high latitudes (Kobashi et al., 2015). The present solar extended minimum, known as the Eddy minimum, includes SC24 and SC25. A conservative forecast on the effect of the Eddy minimum on temperatures indicates no additional global warming before 2035, and perhaps even a slight cooling. This forecast is also supported by the position of the 60-year oscillation (figure 102), that also indicates no warming for the first 3 decades of the 21st century.
Figure 125. Global temperature change 1950-2018: comparing observations to models. Global surface temperature anomaly (black curve; °C; monthly HadCRUT4 13-month averaged) with its linear trend (thin continuous line). Source: UK MetOffice. Coupled Model Intercomparison Project Phase 5 (CMIP5) multi-model mean temperature anomaly 1950-2045 under RCP4.5 conditions (fine blue line; 13-month averaged) with 25-75%, and 5-95% values (medium and light blue areas) for the 42 models used. Source: KNMI climate explorer. CMIP5 models were initialized in 2006 (vertical line) and reproducing historical climate to that point was a prerequisite.
It is remarkable that a knowledge-based conservative forecasting for the next 15 years as the one presented here, agrees so well with the naive no-change 2007 forecast by Armstrong for the past 10 years, that showed to be superior to IPCC early forecast. It is important to emphasize that although very variable in the short term at different places, Earth temperature is extraordinarily constant over the long term. 0.2°C is a small variation in temperature at temporal scales of less than one year, but a significant variation on yearly averages, an important variation in decadal temperature scales, and a huge variation in millennial scales. The Neoglacial trend that has driven glacier expansions all over the globe, culminating at the LIA, was just 0.2°C/millennium or less over the past five millennia (0.38°C/millennium in Greenland; Kobashi et al., 2015), due to Milankovitch forcing. The planet lost about one degree from the Holocene Climatic Optimum to the second millennium AD average, and this amount caused considerable glacier expansion and biome changes, reducing the tropical forests and expanding the tundra. Higher temperature changes are observed on shorter timeframes, but they also have lower and shorter effects. From July 2013 to February 2016 the global surface average temperature anomaly increased by 0.4°C but has since lost most of it (figure 125). Multidecadal to centennial temperature forecasting, to be conservative, must strongly constrain the amount of temperature change that it allows. That is the reason why Armstrong no-change forecast was better than IPCC 0.3°C/decade forecast. Frequent claims that we are on course to +2°C above pre-industrial average temperature by 2100, included in some official climate scenarios, require sustained long-term warming rates above what it has been observed over the past seven decades, and thus are highly unlikely to be correct in any case.
For the past 120 years, the global temperature anomaly has been changing by a long-term trend and a 60-year oscillation that have not been affected much by the increase in CO2. The long term temperature trend is ~ +0.12°C/decade in global temperature anomaly, while the 60-year oscillation departs from trend by ~ ±0.2°C (figure 125). After 2035 a likely increase in solar activity, and an expected change in phase in the 60-year oscillation, suggest a forecast for resumed warming during the ~ 2035-2065 period. Therefore, for the entire period 2018-2065 we should expect a continuation of the observed linear increase in temperature since 1950 (figure 125) of ~ 0.12°C/decade. Any small deviation from this linear trend should be towards lower values if emissions decrease or if solar activity is lower than expected. An increase in any of them beyond what is calculated is unlikely as they are already considered a high scenario.
Of the two most important external forcings contributing to Modern Global Warming, stabilizing emissions suggest that maximum CO2levels could be reached by ~ 2075, while maximum solar forcing is forecasted for the 2050-2080 period. If those forecasts are correct, it follows that maximum temperatures should be reached in the 2050-2100 period at ~ +1.5°C above pre-industrial values, and remain essentially stable, with variability due to natural oscillations and a small declining trend, for the rest of the present warm period (figure 126). This warm period, currently unofficially known as the Anthropocene, could last 2-3 centuries more, until ~ 2250-2350, if the Medieval Warm Period is a good analog.
Figure 126. Conservative temperature, CO2level, and emissions forecast to 2200 AD. CO2emissions from fossil fuels forecast (brown continuous line) based on a peak in oil consumption by 2030-40 (dashed brown line), a second peak in coal consumption by 2050-60 (dotted brown line) and increasing gas consumption to 2100 (dash-dotted brown line), producing a peak in CO2emissions from fossil fuels at ~ 35 Gtons by 2050. Historical CO2emissions from fossil fuels are from T. Boden, G. Marland, and B. Andres, archived at CDIACand updated with BP Statistical Review 2018. Atmospheric CO2levels (blue line) should therefore stabilize at ~ 500 ppm by 2080 before starting to decrease slowly as sinks remove more than is added. Historical atmospheric CO2levels are fromLaw Dometo 1958 and from NOAAafterwards. An idealized millennial cycle in solar activity is represented by the orange line, peaking at 2050-2080, temporarily reduced by centennial and bicentennial lows indicated with their names. CMIP5 model-mean temperature anomaly (red line) projects reaching +1.5°C above pre-industrial by the 2030’s, and +2.0°C by the 2050’s. Temperature anomaly should stabilize until 2030s, increasing afterwards and peaking ~ 2070’s at ~ +1.5°C above pre-industrial, due to CO2and solar activity forecasts, and oceanic oscillations. Afterwards temperature anomaly could enter a slightly declining undulating plateau as both CO2and solar activity slowly decline. Historic temperature anomaly is from UK MetOffice HadCRUT4.
This conservative forecast of essentially no change in temperature for the next 200 years, allowing for ±0.5°C multidecadal changes, rests on the following assumptions:
– A continuation of the CO2emissions stabilizing trend observed over the past 7 years with a small declining trend of 0.3%/year after 2050.
– An increase in solar activity peaking ~ 2080 and a decrease afterwards.
– Unsaturating carbon sinks for the period and amounts considered.
– A trend to equilibrate carbon sinks and sources plus emissions at an airborne fraction close to zero.
The forecast does not depend on any change in policies or heroic reductions in emissions. Policies already being implemented, limitations in fossil fuel availability, and natural demographic changes set the course for future reductions in emissions. Faster reductions should not affect the forecast very much, as atmospheric levels should react slowly to them, and the effect of CO2on temperature appears to be lower than estimated by the IPCC (see: Modern Global Warming).
Consequences for Arctic sea ice
The 30% decline in Arctic sea ice extent that took place between 1995 and 2007 led to numerous radical forecasts, predicting in some cases a summer ice-free Arctic by 2016 (Maslowski et al., 2012) due mainly to albedo feedback leading sea ice into a death spiral (Serreze, 2008). Of course, radical forecasts are seldom true, and the albedo effect on sea ice has turned out to be lower than expected, because summer Arctic sea ice extent has refused to decline any further for the past 10 years. Green and Armstrong (2007) are proven correct in their assessment that experts’ predictions are not useful in situations involving uncertainty and complexity, when biases tend to go unchecked.
A knowledge-based Arctic sea ice forecast must take into account the known 60 and 20-year periodicities in sea ice (Polyakov et al., 2004; Divine & Dick, 2006; Wyatt & Curry, 2014; figure 126) probably responsible for the present Arctic summer melting pause. These oscillations are likely to produce no change to slight growth in Arctic summer ice until ~ 2035, when significant melting is more likely to renew. A conservative forecast is that Arctic summer sea ice will decrease at a slower rate for the period 2018-2050. By 2050 there should still be close to 4 million km2of summer sea ice in the Arctic (figure 127; table 2). A return to a warming, melting phase around 2040 might further reduce Arctic sea ice that could be down to ~ 2.5 million km2of summer sea ice (table 2) by 2100. With such low levels it cannot be ruled out that some summer might see an ice-free condition (< 1 million km2). This forecast is not too far from the IPCC RCP4.5 projection (figure 127; table 2), probably because the cryosphere (except Antarctica) is showing a strong response to increased CO2and soot levels.
Figure 127. Projected Arctic sea ice decline. Model simulations (colored lines), historical simulation (black dotted line), and observations (black line) of Arctic sea ice extent for September (1890-2090). Colored lines for RCP scenarios are model averages (CMIP5) and lighter shades of the line colors denote ranges among models for each scenario. Source for original figure: Walsh, J.D. et al. 2014. Modifications to the original figure: Brown line is a model based on the known 60 and 20-year periodicities in Arctic sea ice. Source: Javier 2017 WUWT. Black continuous line is NSIDC September Arctic sea ice extent for the satellite window (1979-2017), while 1935-1978 September Arctic sea ice extent data is from Cea Piron & Cano Pasalodos 2016 reconstruction. An extrapolation of the observed trend to 2012 (exponential fit) triggered multiple predictions of an essentially ice-free Arctic in summer (< 1 million km2) before mid-century, likely to be wrong. The conservative projection (brown line), explains the pause in Arctic sea ice melting since 2007 and suggests over 2 million km2of Arctic sea ice remaining by summer 2100.
The conservative forecast however is in stark contrast to the many alarmist projections from polar scientists that believe Arctic sea ice is past a tipping point and only accelerated rates of melting are possible now. Those projections that see an Arctic free of ice every summer before 2100 are very likely to be wrong. Lack of significant melting progress for the next decade and a half might clarify the issue.
Consequences for sea-level rise
In 2007 the IPCC made public its Fourth Assessment Report (AR4). Among AR4 emissions scenarios was B1, that contemplates slow growth in CO2emissions to 2050 followed by moderate decrease in emissions afterwards. This scenario is the one that best agrees with the conservative projection outlined above, and projects a 300 mm increase in sea levels for 2000-2100 (central estimate; figure 128). Seven years later the IPCC published its Fifth Assessment Report (AR5), and among the new scenarios RCP4.5 is the most like B1. However, the IPCC sea-level model is now a lot more aggressive and projects 525 mm for similar emissions (table 2). Such a strong upward revision responded to claims that models used in the 4th report substantially underestimated the observed past sea-level rise, although no acceleration has been observed since 1993. Despite the 60% increase due to a change of assumptions, the IPCC was severely criticized for producing estimates of sea-level rise that were too conservative. To provide a view that satisfied the consensus, Horton et al. (2014) conducted an expert elicitation (poll) on sea-level rise among authors of articles related to sea-level rise. Although they were only asked for a low and high scenarios, a mean projection can be obtained by averaging both (figure 128). This intermediate scenario derived from Horton et al. (2014) projects a rise of ~ 800 mm for 2000-2100. In 2017 NOAA published their updated global sea-level rise scenarios where the intermediate scenario, that is most consistent with RCP4.5, forecasts one meter of sea-level rise for 2000-2100 (Sweet et al., 2017; figure 128). Surprisingly, and despite lack of acceleration in sea-level rise since 1993, projections are becoming significantly more pessimistic with time.
Figure 128. Sea-level rise intermediate scenarios for 2100. Red curve, sea-level rise measured since 1993 and zeroed in 2000. Source: NASA. Dashed curves, sea-level rise projections for the 2000-2100 period under intermediate emissions scenarios from different sources. 2007 IPCC AR4 B1 scenario (dashed black); 2014 IPCC AR5 RCP4.5 scenario (dashed dark grey); 2014 Horton et al., survey intermediate scenario (average of the high and low scenarios; dashed medium grey); 2017 NOAA intermediate scenario (dashed light grey, Sweet et al., 2017). Black curve, Conservative oscillatory forecast based on 65-year oscillation and observed acceleration. All models were started at zero in 2000 but shown since 2017. The conservative oscillatory forecast is so far the closest to the observed value.
A conservative sea-level rise forecast considering only past sea-level increases, that for the last 70 years have taken place under rapidly increasing emissions, must consider the 65-year oscillation in sea levels (Jevrejeva et al., 2008), and the small 0.01 mm/yr2acceleration detected by most researchers (Church & White, 2011; Jevrejeva et al., 2014; Hogarth, 2014; figure 114). Such a forecast would see a 115 mm increase in the first half of the 21st century, and 175 mm increase in the second half, for a total 290 mm increase (figure 128; table 2). This increase is too small to constitute a problem on a global basis but might add to the problem of local sea-level rise in areas where land subsidence or lack of sufficient sedimentation are going to require adapting measures.
Other climate change consequences for the 21st century.
A conservative forecast is that most extreme weather phenomena should continue occurring in an unpredictable manner without a significant change in frequency. Storm data from the last 6500 years shows clearly that storms increase in frequency and strength with cooling, and decrease with warming (figure 75 b). The reasons are that warming reduces heat transport due to a decrease in the latitudinal temperature gradient, and that the atmospheric heat engine has a reduced ability to generate work due to an increase in the power required by the intensification of the hydrological cycle (Laliberté et al., 2015).
The only extreme weather phenomenon that is credibly projected to increase is the frequency and intensity of heat waves. However, the change could be smaller than anticipated as Modern Global Warming is having more effect on minimum, rather than on maximum, temperatures producing generally warmer winters. From a societal point of view, adaptation to increased heat-waves requires cheap, abundant energy.
The effect over the biosphere is more difficult to forecast, as it has shown very high adaptability through much bigger climate changes in the past. If we accept that the world in 2017 was ~ 0.95°C above the pre-industrial average, the conservative forecast indicates it might only increase a further 0.55°C before stabilizing. Therefore, we might have already seen over 60% of the total expected warming. The negative effects strictly from climate change over the biosphere are very limited, while the positive effects are abundant and profound. Most biomes, but particularly semi-arid ones, have responded to warming and CO2increase through an increase in leaf area (also known as greening; Zhu et al., 2016). These three factors appear to have caused an increase in global terrestrial net primary production of 12% between 1961-2010 (Li et al., 2017). The effect of this increased energy flux through ecosystems is beneficial to nearly all species. The net effect of the warming and increased CO2is clearly positive for the biosphere. It is reasonable to think that too much of a good thing should reach a point when the net effect starts being negative, but there is no evidence that we are close to that point or that it should be reached within the 21st century.
Table 2. Twenty-first century climate projections. The 2017 values for some climate indexes are given in the first data column. The rest of the columns give the projected or assumed corresponding values for 2050 (white background) and 2100 (grey background), for a conservative prediction and IPCC RCP4.5 emissions scenario. The biggest differences are for sea-level rise and temperature increase, where the conservative prediction is closer to observed changes for the past decades.
The loss of Arctic sea ice has been proposed to be a clear risk to polar bears, and the species was included in the US endangered species list solely on those grounds. However polar bears might not be very sensitive to summer ice reductions as their ice-dependent hunting takes place in spring and is negatively affected by too much ice. The species has survived very reduced or even absent summer Arctic sea ice during the Holocene Climatic Optimum and the past warmer interglacial. The main danger to polar bears has historically been human hunting, and since the international hunting limitation by the Oslo agreement of 1973, polar bear population estimates have been increasing, apparently unaffected by the loss of 30% of summer Arctic sea ice in the 1995-2007 period (Crockford, 2018). At present there is no evidence that polar bears are threatened during the 21st century from climate change, even if the projected summer ice loss in the Arctic takes place (figure 127).
Regarding other possible consequences, our knowledge is too limited to say much. Claims of sinking nations, hordes of climate refugees, and a new normal every time there is an extreme weather event, are wildly exaggerated and agenda-driven. The highest return for our limited resources is very likely to come from adaptation policies, and no-regrets policies. Policies to prevent or reduce climate change are destined to be highly ineffectual given the strong natural component of climate change, as the past demonstrates.
1) Human CO2emissions are stabilizing. Peak coal and oil, and current trends make a decrease in emissions very likely before 2050. Atmospheric CO2levels should reach 500 ppm but might stabilize soon afterwards.
2) According to solar cycles, solar activity should increase after the present extended solar minimum, and 21st century solar activity should be as high or higher than 20th century. A mid-21st century solar grand minimum is highly improbable.
3) Global warming might stall or slightly reverse for the period 2000-2035. Cyclic factors suggest renewed warming for the 2035-2065 period at a similar rate to the last half of the 20th century. Afterwards global warming could end, with temperatures stabilized around +1.5°C above pre-industrial, and a very slow decline for the last part of 21st century and beyond.
4) The present summer Arctic sea ice melting pause might continue until ~ 2035. Renewed melting is probable afterwards, but it is unlikely that the Arctic summer will become consistently ice free even by 2100.
5) The rate of sea-level rise can be conservatively projected to a 290 mm increase by 2100 over 2000 levels. Most rates published are extremely non-conservative and very unlikely to take place.
6) Climate change should remain subdued and net positive for the biosphere for the 21st century. Adaptation is likely to be the best strategy, as it has always been.
I thank Andy May for reviewing the manuscript, and providing useful comments towards improving its content and language.
Moderation note: As with all guest posts, please keep your comments civil and relevant.