Natural Climate Change

   For at least half of the last 600 MY, Earth's surface has been 15-30°F warmer than now, with CO2 levels at least 5 times today's, much higher till 350 MY ago.  [See Heat section.]

     Starting ~ 375 MY ago, CO2 levels fell steeply, after plants colonized land.  Glaciers were widespread, like in the 1900's and 25,000 years ago, ~300 MY ago.
    Oxygen levels generally fell when CO2 levels rose, and vice versa.  See figure above.  That’s also true, but subtly, over the past century.

    In part, higher CO2 levels compensated for a dimmer sun in the distant past.  Our sun brightens slowly as it ages, warming Earth about 2°C more per 100 million years.  Clouds modify that.

          Our sun also brightens, then dims, by ~ (about) 0.1% (from minimum to maximum), during "11-year" sunspot cycles (9-13 years, usually 10 to 12).  More dark sunspots are outshone by more bright faculae around them, for a brighter sun, especially in ultraviolet light.  2008 saw the fewest sunspots in 111 years, so it’s no surprise that solar radiation was the lowest in the 40-year satellite era.  That means that changes in solar output generally had a slight cooling effect on Earth over the last 40 years.  Variation during longer (and more variable in length) sunspot cycles is subtler.  Solar changes do influence climate, but their effect has been overwhelmed by the much larger effects of human emissions.  The sun has cooled slightly since 1979, so humans emissions (with their feedbacks) account for over 100% of the warming observed since then.

     In the satellite era, Earth warmed as the sun brightened, and cooled as it dimmed, mostly - until 2002.  During 2002-2008 and 2014-16, Earth’s surface temperature rose while solar output fell.  Again, solar variability’s modest influence on climate has been overwhelmed by the human effect.

     Graphs that Dr. Fry created are mostly available in many of his slide shows, in editable PowerPoint slides, some shown with underlying Excel spreadsheets.  Dr. Fry’s graphs feature Arial Bold font, larger than in most other people’s graphs, usually with gridlines and a few somewhat thick data lines of various colors.  Among them are all but 4 graphs below.

     CO2 levels have varied a lot over the eons (see above for 570 million years).  Very long ago (~250 MY, etc.), vast lava eruptions lasting a million years or so, such as the ones that created the Siberian Traps and India's Deccan Traps, and even the Columbia River Basalts a few MY ago, added lots of CO2 to the air: multiples of what we have now.  20th century volcanic eruptions added far more modest amounts: roughly 1% of current annual human CO2 emissions.

     When continents collided, mountain ranges rose.  As moist air moved over them, they caught more rain and snow.  This speeded up rock weathering processes, which annually remove ~3% as much CO2 from the air as humans now put in.  In weathering, CO2 dissolved in rainwater combines chemically with minerals (mostly calcium silicates) in rock surfaces, which wash away to become carbonate sediments.  This has been so for 100s of MY.

     During the PETM's temperature spike 55 MY ago, global temperatures rose by 0.025 to 0.06°C per century.  To compare, surface warming over the past century was 1.2°C. This has speeded up.  Over the past 20 years, the warming rate was 1.7°C per century.  2.2° per over 10 years.   

     Weathering in the Himalayas has driven CO2 levels down for some 50 MY, from some 3 times current levels (see 1st figure above).  Ebbing CO2 cooled Earth's surface, enabling glaciation in Antarctica starting 34 MY ago.  As CO2 ebbed further, Greenland glaciation began 18 MY ago and hit its stride 8 MY ago.  Finally, as CO2 levels fell further, widespread glaciation began in Canada, Alaska, Scandinavia, most of Siberia and the northern US ~ 2.5 MY ago (punctuated by interglacials).  Note how glaciers ebbed in Antarctica from 27 to 15 MY ago [pointed blue bars], as temperatures rebounded by 1 to 2°C for 14 MY.
    Algae, plants, and seashells also removed CO2 from the air, to store in soil, water, and sediments, once life took hold.  When life spread to land, the process speeded up (see above, 350 to 300 MY ago).  As conditions permitted, these dead lifeforms transferred into coal, oil, gas, and limestone, which stored carbon underground.  This has also been going on for hundreds of MY.

      During ice ages over the past 2 MY, CO2  & CH4 levels were much lower than today's (~ 420 parts per million [ppm] of CO2 in the air, 1900 ppm of CH4).

     The timing of these ice ages (once general CO2 levels dwindled enough - credit the Himalayas & an isolated continent at one pole), was driven by Milankovich cycles - small variations in Earth's tilt, the roundness of its orbit, & when it's closest to our sun (in northern [land hemisphere] summer, winter, or in between).  The "beat", when these rhythms reinforce each other, comes about every 100,000 years.

     Finer time resolution for these ice ages shows warming came 1st, followed "shortly" by more CO2 and CH4 in the air.  Warming, from natural orbital factors, drove carbon out of permafrost (CO2 where it's dry, CH4 where it's wet), oceans, and soils.  Warming speeds up decay by soil microbes.

     Worldwide warming averaged .04°C per century from 20,000 to 10,000 years ago, but almost twice that at Vostok.

     The relationship of CO2 (and CH4) to temperature about 4 and 14 million years ago is essentially the same as shown in the more recent Vostok ice cores.  See graph below.

     The green line connects the patterns relating temperature change to changes in CO2 and CH4, over the last few ice ages, to those the last times (4 & 14 Mya) we had about this much CO2 (and perhaps CH4) in the air.  The purple line is just for the ice ages, since we don't know CH4 levels earlier.  The story is consistent across millions of years.  The equation including CH4, shown in green, has the best explanatory power (R2) for Vostok.  The natural logarithm (LN) form adjusts for GHG molecules getting in each other’s way as they become more numerous, so that a smaller fraction of the newer molecules actually absorb outbound infrared radiation over a given timespan.

      With current CO2 & CH4 levels, the equation yields global warming of 8.6°C.

     The purple line connects the patterns relating temperature change to changes in CO2 only.  Neglecting CH4 provides a fit to the ice core data that is not quite as good.  It implies just 5.3°C warming from today’s CO2 levels.

     (Since 1750, CH4 levels have risen far faster than CO2 levels, 170% vs 48%, especially in the 19th and early 20th centuries, also in the 2010’s.)

     What’s going on?  Over a few decades, sea ice will vanish, as will sulfates from coal power plant smokestacks.  Over many decades, snow cover keeps shrinking, arriving later and especially leaving earlier.   Antarctic and Arctic sea ice may vanish at similar times, depending on future human emissions, carbon removal from air, and physical processes yet only partly identified and understood.  Antarctic ice declined steeply in 2022 and 2023.   All these changes make Earth darker, so it absorbs more heat.
    More important, over the past 30 years, Earth’s cloud covered area has probably been decreasing by .06% per year.  That makes Earth darker, so it absorbs more heat.  This effect is actually a very large number, since Earth is large and half covered in clouds.  The effect of the observed cloud cover decrease (in 2013 AMO report) was about half as much over 30 years as the warming effect to date from all GHGs combined.
    That’s assuming that the ratio of high-altitude clouds to low-altitude ones stayed roughly the same.  However, since 2001, high altitude (net warming) clouds have increased, while since 1983, low altitude (net cooling) clouds have decreased, while middle altitude clouds have increased modestly.  (See Heat section, Clouds page.)  Finally, low clouds have been growing more opaque (Clouds section); this opaqueness cooling effect partly offsets the effects of shrinking area and the growing high-to-low cloud altiude.  The overall cloud effect, a very fast feedback, treated as a multiplier; by 2020 it added an estimated ~20% to warming from other factors.  Recent research suggests that it may be an underestimate, especially with cloud changes over 2020-2025.

     In addition, smaller albedo changes come over many decades to some millennia, as ice vanishes from almost all of Greenland and West Antarctica.  (In recent years, ice loss has accelerated 12% per year in Greenland and in Antarctica.)  Earth’s surface continues darkening.  Thus, Earth absorbs still more more sunshine, heating up more.  This is a amplifying ("positive") feedback loop.  Most of the heat absorbed goes into the oceans, but a small fraction heats soil and rock.  A smaller fraction melts ice.  The smallest fraction (about 1%) heats the air, which is mostly what we pay attention to.

     All these albedo changes will be multiplied by more water vapor in the air.  Water is the #1 greenhouse gas.  It amplifies warming from other causes: not only other GHGs, but also albedo changes.  Air holds 7% more water, at the same relative humidity, for each 1°C warming.  The difference between relative humidity and absolute humidity is quite important: more water vapor in warmer air does not mean more clouds.  We have actually observed, in recent years, increasing water vapor and decreasing cloud cover, both measiured globally.  The 7% more water vapor increases the heat effect (“radiative forcing”) by 1.5 Watts / square meter, about 2/3 as much as all greenhouse gases combined, or another 0.6-0.7°C water vapor feedback for 1°C warming.

     This amplifies the direct effect of CO2, CH4 and other GHGs - a lot.
But it also amplifies the effects of albedo changes.

  5.3°C global warming is enough to make Kansas as hot as Las Vegas - sparsely inhabited desert
8.6°C make Kansas almost as hot as Death Valley - uninhabitable.

     Earth has natural cycles that last for years.  El Niño / La Niña is the most prominent of these.

     Elevated CO2 levels have long been used in greenhouses to increase production of flowers and vegetables.  This uses the "CO2 fertilization effect", boosting yields by 6-35%.  However, plant growth is limited by many factors - temperatures too hot or cold (e.g., below freezing, or hot enough to denature key proteins), low light levels (night, etc.), CO2, water, nitrogen, phosphorus, potassium, calcium, acidity, soil depth, etc.  Greenhouses provide enough of these other sometimes limiting ingredients so that CO2 is their limiting factor.
    When CO2 is a limiting factor, adding more CO2 helps plants grow, especially C3 plants - until the plants run low on another factor, most often nitrogen.  This explains why experiments find that an initial (1-5 year) CO2 growth spurt then fades.
    C3 plants include most food crops and weeds, but corn and sugar are C4 plants.  See “Food” section, in “Summaries of Observations” and above “Future Impacts 2” for graphs of CO2 fertilization effects, most combined with temperature effects.  However, while plants with high CO2 availability grow more carbohydrates, they grow fewer proteins.  See Food section for explanations.

Human-Accelerated Climate Change

     Especially in the past century, humans have added lots of CO2 to the air - by burning coal, then oil, then also natural gas.  Cutting down forests and farming practices (tilling, cattle and feedlots, flooded rice paddies, etc.) are other major factors, including CH4 increases.

     These can, and do, speed up natural climate change.
    The last times CO2 levels were almost this high, Earth’s surface temperature was about 7°F (4° C) warmer.   When they were slightly higher than now, more like 12°F (7-8° C) hotter.

        After a Medieval Optimum and 3 Little Ice Ages (see Heat section), land surfaces warmed from 1890 to 1942, plateaued to 1976, warmed steeply to 2004, little till 2012, then shot up again, even faster.
    Sea surfaces warmed 44% as fast as land surfaces over 1880-2023.  This includes 89% over 1963-83, 39% over 1983-2003, and 79% over 2003-23.

    Q: If CO2 causes warming, why doesn't warming increase as smoothly as CO2?
    A: Sulfates.  See farther below.  Note that global land surface temperatures are ALREADY in 2023 almost 2.0°C above 1880 (5-yr moving average), according to NASA (1.9°C says NOAA).  They are more than 1.5°C above the 1880 level, a stretch target in the December 2015 Paris Accords.  And they are not headed down.

.    Earth's land surface has warmed about 1.4°C (2.5°F) since 1921. That includes 2.1°C (3.8°F) per century since 1971 and 2.4°C (3.7°F) per century since 2001.  Warming since 1920 has been 70 times as fast as it was 11,000 to 7,000 year ago, or ~35 times as fast as from 18,000 to 11,000 years ago.  (Earth's sea level rose rose 2 meters (6.6 feet) per century then.)  And Earth warmed still faster in more recent years.  Earth's surface has been warming far faster than before humans put lots of CO2 in the air.

     ~90% of Earth's heat gain went to warm the oceans (including 80% to only 700 meters deep), while only 1% went to warm the air - which is mostly the warming we care about.  The rest melted ice (2%), heated rock & soil (5%), and increased water vapor in the air.

     The rate of ocean heat gain has accelerated, from 1967 to 2016.  Since 2005, oceans added heat 25 times as fast as humans now use energy (~150 x as fast as the US does).

    Since 1969, ocean heat gain exceeds 10 x the energy humans have ever used: ~ 3,000 years of current US energy use.  The tail is wagging the dog.  (Humans are wagging Earth's oceans and air).

     Effects of sulfates are clear for major volcanoes, which put sulfates in the stratosphere for many months. (Smaller eruptions - too many to show - had much smaller effects, as most did not reach the stratosphere.)

    The bigger picture (numbers 41-123 in bottom box of the Temperature / Sulfates slide) is that, over 135 years, human sulfur emissions mostly rose, especially 1940-80, to peaks in 1973 and 1979.  That masked much warming from CO2.  But when sulfate levels fell (1930s, 1975-2005), warming was unmasked.  Then temperatures rose steeply.

     The graph above traces the 3 biggest albedo factors - other than clouds, although some cloud changes are included under SO4 - plus the two principal GHGs, over the same time period.   It shows that the close agreement between global surface temperatures and cumulative CO2 emissions.  The historical explanatory power, 95.3%, however, is not as good as 98.3% for global land surface temperatures adding SO4 and CH4, above.  Note that neither analysis includes less important, but non-negligible, influences: black carbon, N2O, CFCs, O3, some cloud changes, and solar changes.

     Above: note the Mt. Pinautubo eruption in 1992 and El Chichón in 1982.

     One result of added CO2 has been warming in America.  From 1975 thru 2015, daily summer highs in 26 places* around the US have increased, on average, by 3°C (5.4°F / century), a bit faster than Earth's land surface as a whole.  From 1975 to 1995, the 26 barely warmed.

(* Jointly, the 40-year urban heat island effects shrank (cooled) slightly for these 26 places.  They were chosen to avoid places with rapid population growth, and thus increasing urban heat island effects, as US energy use did not increase faster than population.)

From 1975 thru 2015, daily summer highs in 26 places* around the US have increased, on average, by 3°C (5.4°F / century), a bit faster than Earth's land surface as a whole.  From 1975 to 1995, the 26 barely warmed.  But since 1995, these 26 cities warmed 10°F / century.  The summers of 2011 and 2012 were especially hot.

     Maybe it's just year-to-year variability.  Maybe it it's more than that.  But if warming continues as fast as it did from 1995 to 2015, by 2100 summers in Kansas, Oklahoma, Georgia and South Carolina would be hotter than Las Vegas ones now.  That’s bad for crops.  Summers would be still hotter in Texas and Nevada.

     Still, 26 is a small sample and 20 years is not so long.  So, error bars on extrapolation are substantial.

    So, switch tracks to rising temperatures in the USA.  Focus on summer highs.

     With a larger sample of 330 places, the picture is very similar.  See the Heat page for regional graphs and the Data page for individual city graphs and data.  The results are similar, but the warming is a little faster (+5.8 and +10.5°F / century).  Some of this modest (5.4 to 5.8 and 10.0 to 10.5) difference is probably an increase in the urban heat island effect.)

    Warming varies a lot from one region to another.  It was been fastest in drier areas: Rocky Mountain and West South Central states.  Warming was also fast in South Atlantic states - except Florida.  Warming was slowest in North Central states and Alaska.

     Why was warming fastest in the Rockies and West South Central states?  The #1 explanation:  On land, sunlight absorbed evaporates water, if available; otherwise it mostly heats air.  At sea, it also heats water below the surface.  Preliminary research on US daily winter lows and daily summer lows shows only weak and ambiguous trends, lending further support to this hypothesis.

Effects

     1995-2015 trends, especially in the faster warming areas, are truly sobering.
    If these 20-year rates (which vary by place) should continue, today’s Las Vegas summer heat comes to Fresno in 2036 & Austin in 2037.
    At the state average level, Nevada, Texas and Arizona summer heat would be as severe by 2083.  Idaho, South Carolina, Kansas ("breadbasket of the world”), Oklahoma and Georgia would grow as hot shortly before 2100.

     Humid areas - such as Georgia, South Carolina and Alabama - could become uninhabitable during summers late this century.

    How long until summer highs average as hot as Las Vegas ones today,
IF recent trends continue?

Slowing and even stopping such warming is utterly important.

     Daily low temperatures, summer and winter, have not risen much or have even fallen (winter lows, 2nd half of 42 years).  But summer highs have risen fast.  Evaporation is higher then and soil moisture levels can be lower.  See Heat page for more summary of summer and winter lows and the bottom of the Data page for details on them.

     Oddly enough, since 1994, when (in any single year) Earth's land surface warmed faster than its trend, the US usually warmed slower than its trend, and vice-versa.  (Correlation = -0.39.)  Food for thought and investigation.

     2012's summer heat could become the new normal within the lifetimes of most of us, by 2030 in several states.
    In the longer term, unless carbon emission rates are cut sharply, by 2200 a few areas (e.g. Persian Gulf) could become too hot and humid for humans to survive in.  They already are for a few hours per year, absent air conditinoing.  Even Alabama, Georgia, and South Carolina sometimes in the summer.  If CO2 levels reach 650 ppm and CH4 evels rise similarly, by 2300, large sections of Earth (e.g. most of Australia, India, the Middle East, the southern US, Mediterranean lands, and parts of Latin America and China) could become too hot and humid for humans to survive.

     Since CO2 emissions are a major cause, to slow down heating or even stop it, we must do 3 things.

     1. Stop burning money (use energy efficiently), whether or not CO2 is a problem.
     2. Use low- & no-carbon energy (more wind and sun [now the cheapest energy in many areas].  Use nukes for at least a few decades.  Use less coal (and oil).  Long term, phase out coal now, gas soon, then maybe nukes.  If we burn all fossil fuels in the ground, Earth would become ice-free and seas would be more than 200 feet higher.  Most of Earth would be too hot for humans to survive.  Ditto for most mammals.

     3. Take carbon out of the air faster than we put it in.  Pay ranchers & farmers to put it back in soils, etc.  Use direct air capture, accelerate rock weathering, de-acidify the oceans, etc.

     Warming has consequences.  Perhaps the most important is faster evaporation, which helps bring intense droughts more often.  (Between droughts, we get worsening floods.)  Here is a projection from 1989, in 2 forms.  Larger views appear on the Overviews page and in the middle of the slide show (available here).

The 2 figures below changed my life.  They come from a study published in 1990 by David Rind, James Hansen, Cynthia Rosenzweig adn others.

     The figure below looks back in time.
It matches well the figure above, which peered into the future.
That is, observations up to 2002 match projections made in 1989.
Worldwide, severe droughts have been increasing.

     Current agricultural models estimate that climate change will directly reduce production of corn and wheat (and with yet more warming, rice, and eventually soybeans too).  These 4 supply 80% of human calories.  Tropics suffer first, temperate zones later (with overlap).  One model estimates a reduction up to 43% by 2100, another more.  The IPCC now estimates 17%.  With shrinking fresh water supplies, especially dwindling groundwater, further large declines loom, absent big improvements in water use efficiency.  Pressure on food supplies is not likely to stop after 2100.

Peering into the Future

Reducing Earth's population to fit the future food supply will not be pretty.

     Think Bosnia, Rwanda, Darfur, Somalia.  Boko Haram, ISIS.  Multiply.

However, many countries already face shinking populations: Japan, China, Korea, and many in Europe.  Many others, including the US, Europe's other countries, and recently India, now have fertility rates below replacement levels

     Some aspects of climate change we know much better than others.  Perhaps the greatest uncertainties are for

1. large scale changes in % cloud cover;
2. how much and how fast we can move CO2 from ambient air to carbon sinks;
3. willingness to cool Earth while we reduce our forced heating of it.
4. future human carbon emissions and the speed of that reduction;
5. how fast snow, sea ice, and land ice melt;
6. future effects on the food supply;
7. how much future warming is unmasked, as we stop emitting sulfur from coal;
8. future rates of sea level rise (see #5, Antarctic & Greenland melting);
9. future carbon escape rates from permafrost and methane hydrates;
10. future carbon escape rates from currently un-frozen soils and a warming ocean, plus changes in land biomass;
11. effects at regional (e.g., Australia) and sub-regional (e.g. Britain) levels;

     Permafrost emissions are based on MacDougall (see Overviews page).  They correspond to DEP 6.0 in the Base Case, 4.5 in the case where fossil fuels are eliminated by 2050, and 2.6 in the case that also includes very large-scale carbon capture and sequestration - from ambient air.  Higher DEP cases (than 2.6 emissions in Dr. Fry's Base Case) are used in Dr. Fry’s modeling above, to compensate for demise of cooling from sulfates and Arctic sea ice and demise of cooling from losing former clouds patterns (see Heat page, Clouds section), all of which were not part of MacDougall’s model.  Systematic changes in cloud cover are subtle.  See “Clouds” on the Heat webpage for details.

     McDougall is conservative in several other ways, mostly as he states: 1) his estimate of the permafrost carbon pool is only 54% of more recent estimates, 2) he does not include methane hydrates (which ~2010 emitted 30% as much carbon per year as permafrost), 3) he does not account for CH4 release from permafrost, 4) he does not account for thermokarst and water erosion of permafrost, 5) he does not account for permafrost warming by black soot from increasing near-Arctic forest fires, and 6) his temperature estimates for the permafrost region are at the low end of the range for the literature.

     Future temperature changes will likely be dominated by cloud cover changes, loss of snow and ice, sulfates loss (in the short term), permafrost carbon emissions (later), and increasing water vapor.  Well before 2100, elimination of sulfates (a coal by-product that increases cloud cover) and polar sea ice play substantial roles.  (See future temperature graph above from 2010 to 2070.)  They both induce positive feedbacks from permafrost and clouds.  Shrinking sea ice and snow cover play larger but similar roles, over decades to, for snow, a few centuries.  Looming above all is the water vapor warming amplification effect.

     Future ice loss rates from Greenland and Antarctica are uncertain.  But several new ice dynamics processes have been identified that were not in earlier models.  All of them speed up ice loss projections.  So, for example, DeConto and Pollard (2016) cut their estimated melt times for much of Antarctica from a few million years (2013) to a few hundred (2016).  Sea level rise shown here to 2100 is half that projected by Hansen et al. (2016), but to 2400 is similar to DeConto and Pollard (2016).

     The general temperature increase is very consistent with Snyder (2016, see Overviews page for summary).  Also, based on more detailed paleoclimate data than used in the graphs above, Snyder foresees 5°C eventually from today’s CO2 levels and 9°C with 560 ppm CO2 (double pre-industrial).  But she does not envision the warming to happen as quickly as I do; short-run warming of about 3°C.

    Below are annotated versions of the set of 4 projection graphs above.

     This all points to the utter importance of removing from the ambient air most of the carbon humans have put in it by burning fossil fuels.  The sooner the better, most of it this century.  It would be good to go carbon negative as a species before 2050.  Civilization cannot survive the temperatures indicated if we fail to do so.  The world food supply would be cut in half, or worse, at 5°C global warming.  9°C warming would be far worse.  Past major extinctions have been driven by large climate changes due to copious CO2 emissions.