I’ll get right to the results, which are pretty straightforward.

As seen in the accompanying plot, 50-year (1973-2022) summer (June/July/August) temperature trends for the contiguous 48 U.S. states from 36 CMIP-6 climate model experiments average nearly twice the warming rate as observed by the NOAA climate division dataset.

The 36 models are those catalogued at the KNMI Climate Explorer website, using Tas (surface air temperature), one member per model, for the ssp245 radiative forcing scenario. (The website says there are 40 models, but I found that four of the models have double entries). The surface temperature observations come from NOAA/NCEI.

The official NOAA observations produce a 50-year summer temperature trend of +0.26 C/decade for the U.S., while the model trends range from +0.28 to +0.71 C/decade.

As a check on the observations, I took the 18 UTC daily measurements from 497 ASOS and AWOS stations in the Global Hourly Integrated Surface Database (mostly independent from the official homogenized NOAA data) and computed similar trends for each station separately. I then took the median of all reported trends from within each of the 48 states, and did a 48-state area-weighted temperature trend from those 48 median values, after which I also got +0.26 C/decade. (Note that this could be an overestimate if increasing urban heat island effects have spuriously influenced trends over the last 50 years, and I have not made any adjustment for that).

The importance of this finding should be obvious: Given that U.S. energy policy depends upon the predictions from these models, their tendency to produce too much warming (and likely also warming-associated climate change) should be factored into energy policy planning. I doubt that it is, given the climate change exaggerations routinely promoted by environment groups, anti-oil advocates, the media, politicians, and most government agencies.

Yesterday I posted a critique of Lord Christopher Monckton’s latest explanation of why he believes climate sensitivity is low. At issue is his claim that researchers have somehow neglected that the feedback response to a climate perturbation (e.g. how much warming occurs from adding CO2 to the atmosphere) needs to include the feedback response to the total emission temperature of the system, which he claims then greatly reduces the system “gain factor” and thus calculated climate sensitivity. I maintain that this is not how climate sensitivity in climate models is determined — only actual physical processes are modeled — and I used clouds as an example of why the system response to small perturbations cannot be determined by including the response of a cold (e.g. 2.7 Kelvin) Earth to solar heating (this is what I claim his argument amounts to when he includes the total system temperature in his system gain calculation). While he and I agree sensitivity to increasing CO2 is likely to be low, I laid out my explanation of why his reasoning is faulty. I invited him to respond, and I present that response, below, without comment. At a minimum this exchange might help us better understand exactly what Christopher is saying from a physical process standpoint, rather than a “system gain” standpoint.

I am most grateful to my friend Dr. Roy Spencer, one of the world’s foremost and most expert meteorological researchers and commentators, for the attention he has kindly devoted to our conclusion that official climatology has an insufficient understanding of control theory and has, therefore, led itself into a persistent and grave error.

I am still more grateful to him for this opportunity to reply to his latest posting on this topic, so as to set the record straight. Roy talks of my “feedback arguments suggesting a very low climate sensitivity”. Let me begin my response to that posting by clearing up the misconceptions that are evident in that thought. First, the arguments we make are not my arguments alone. My team includes many experts more than usually competent in both theoretical and applied control theory.

Secondly, our arguments do not “suggest a very low climate sensitivity”. Consider the position at the temperature equilibrium in 1850. The reference temperature that year was the 267.1 K sum of the 259.6 K sunshine or emission temperature and the 7.5 K directly-forced warming by, or reference sensitivity to, preindustrial noncondensing greenhouse gases; and the observed HadCRUT equilibrium global mean surface temperature was the 287.5 K sum of 259.6 K and the 27.9 K total natural greenhouse effect, which itself comprises the 7.5 K reference greenhouse-gas sensitivity and 20.4 K total feedback response.

Early papers on equilibrium doubled-CO2 sensitivity (ECS) based on explicitly quantifying feedback response, from Hansen (1984) onwards, show that the original reason why climatology imagined ECS to be of order 4 K was that the system-gain factor (the ratio of equilibrium sensitivity after feedback response and reference sensitivity before accounting for feedback response) was 27.9 / 7.5, or 3.7 (or, using the round numbers in vogue at the time, 32 / 8, or 4). Since midrange reference doubled-CO2 sensitivity (RCS) is 1.05 K, it was thus imagined that midrange ECS was 3.7 times 1.05, or about 4 K.

Once Hansen and others after him had repeated that midrange estimate often enough, it became impossible for the climatological community to move away from it. They were stuck with it. The whole shoddy house of cards would collapse if they revised it significantly.

The correct system-gain factor for 1850 was not 27.9 / 7.5, or 3.7. It was (259.6 + 27.9) / (259.6 + 7.5), or 1.08. In effect, climatologists had forgotten the Sun was shining and had, therefore, forgotten that there is a feedback response to emission temperature. They had overlooked that large emission-temperature feedback response, and had added all of it to the actually small feedback response to preindustrial greenhouse-gas reference sensitivity. They had thus reached their high midrange ECS of about 4 K by imagining, incorrectly, that the feedback response to emission temperature was zero, which is nonsense.

In reality, such feedback processes as subsist in the climate system at any given moment (such as 1850) must, at that moment, necessarily respond equally to each Kelvin of the entire reference temperature. Feedbacks do not, repeat not, respond solely to perturbation signals, the reference sensitivities. They also respond to the base signal, the emission temperature that would prevail even if there were no greenhouse gases in the air, because the Sun is shining.

Roy says that the underlined words are not true. [“Feedbacks do not, repeat not, respond solely to perturbation signals, the reference sensitivities. They also respond to the base signal, the emission temperature that would prevail even if there were no greenhouse gases in the air, because the Sun is shining.”] When I first realized that climatologists — on both sides of the debate — simply did not understand enough control theory to appreciate the truth of the underlined words, I discovered that a control theorist who was a friend of one of my distinguished co-authors did not realize they were true either. But he had his own lab. So he built a feedback amplifier circuit and tested the matter for himself. That was not easy, because so small is the true unit feedback response that he had to run wires into the next room so that his body temperature did not affect the readings. To his surprise, he found that the underlined words are correct.

Another control theorist, also a co-author, suggested that we should consult a national laboratory of physical engineering to put the point beyond doubt. So we did, and the lab came to exactly the same conclusion, after months of delay because the operator’s body temperature again interfered with the readings, and he had not thought to run wires into an adjacent room. So the matter is not in doubt.

Next, Roy incorrectly assumes that we maintain that “the climate system’s response to a small perturbation from its current state might be discerned from its response to the presence of solar heating assuming an initial cold Earth”.

In reality, we start not with “an initial cold Earth” but with the climate of 1850. We do not need to know what might have happened at 2.73 K ambient temperature. In 1850, when the equilibrium temperature was measured to a respectable precision, the system-gain factor — the ratio of equilibrium to reference temperature — was 287.5 / 267.1, or somewhat below 1.08. All we say, therefore, in relation to 1850 (we go back no further than that) is that ECS based on climatology’s original method adjusted to take account of the fact that in 1850 the feedback processes then extant had to respond equally to each Kelvin of reference temperature regardless of its origin is 1.08 times the 1.05 K RCS, or about 1.1 K.

We then demonstrate via a detailed energy-budget calculation that using mainstream midrange initial conditions it is perfectly possible that the system-gain factor following a CO2 doubling compared with 1850 remains somewhat below 1.08 and that, therefore, ECS is about 1.1 K.

However, we also draw explicit attention to the fact that, precisely because feedbacks respond to the entire reference temperature, and precisely because the base signal, emission temperature, is 30 times larger than the perturbation signal, reference sensitivity to natural and anthropogenic greenhouse gases, even a very small change in the feedback regime compared with the equilibrium in 1850 would exert a disproportionately large influence on ECS. In fact, a mere 1% increase in the system-gain factor at a new moment of equilibrium compared with 1850 would push ECS up by 300% to the 4 K that is the CMIP6 models’ current midrange projection. Therefore, our method does not prove that ECS is low: instead, it shows that it may be low, but proves that ECS is not reliably constrainable.

We draw the conclusion, applying standard feedback analysis, that it is simply not possible to derive ECS as climatologists now do, by diagnosing feedback strengths from the outputs of the general-circulation models and then deriving ECS therefrom. Or, to put it another way, the interval of system-gain factors implicit in IPCC’s current 3 [2, 5] K ECS interval is only 1.10 [1.09, 1.13], an interval so tiny as to fall well within the published uncertainty envelope of feedback strengths, rendering any attempt to predict ECS no better than guesswork.

Albeit by an entirely different method, we reach the same conclusion as Pat Frank in his important paper of 2019, in which he demonstrated that the envelope of uncertainty in ECS arising from propagation of the published uncertainty in a single climatic variable — the low-cloud fraction — was so large that all projections of ECS that have ever made fall within that envelope and are, therefore, mere guesswork. They have no predictive validity at all.

Roy devotes much of his article to the question of clouds. However, in the entire posting by my to which his piece is a response, the word “clouds” occurs only once, and in a context peripheral to the central argument. We point out, in common with Professor Lindzen, that at emission temperature, when by definition there are no greenhouse gases in the air, there would be no clouds either, wherefore, by the Professor’s calculation, emission temperature would not be 259.6 K but more like 271 K, which would of course reduce ECS still further. However, we explicitly point out that we take no account of that fact at all. Our analysis does not depend on the value of the cloud or any individual feedback. Roy says our analysis implies that further warming will not be mitigated by an increase in cloud cover. But our method carries no such implication, for it takes no view on ECS, other than to point out that on the basis of mainstream, midrange data it is possible that ECS may be as little as 1.1 K.

Roy then says climate sensitivity does not depend upon feedback analysis. Indeed, models do not implement feedback formulism directly. Instead, feedback strengths are diagnosed from the models’ outputs (see e.g. Soden & Held 2006 or Vial et al. 2013 for the method). However, the climate is a feedback-moderated dynamical system. Therefore, feedback formulism in control theory is applicable to it and we may, as we have done, apply feedback formulism to the published ECS interval. We may, as we have done, show that in this as in any system where the base signal exceeds the perturbation signal by orders of magnitude it is not possible reliably to predict the output signal in response to a given small perturbation in the total input signal where, as in the climate, the envelope of uncertainty in feedback strength grossly exceeds the interval of uncertainty in the absolute system-gain factor.

It is for this reason that it matters that climatologists had, in effect, forgotten that the Sun is shining and that, therefore, at any time in the industrial era, in the presence of the greenhouse gases, some 29/30ths of total feedback response is feedback response to the emission temperature — i.e., to the surprising fact that the Sun is shining.

It is simple to deduce, again from mainstream, midrange data, that each $1 billion spent on attempting to reach global net-zero CO2 emissions by 2050 would abate between one five-millionth and one millionth of a Kelvin of future global warming, at a total cost potentially exceeding total global corporate profits over the next 30 years (and indefinitely thereafter). Even if there were a real “climate emergency”, the expenditure would not be justifiable, because it would purchase an abatement amounting to only 3/8 K (if you believe IPCC’s midrange ECS estimate) or 1/7 K (if instead we note that since 1990 the world has warmed at little more than a third of the originally-predicted rate). In short, there is nothing we can do to abate future global warming other than reverting to the Stone Age — the decision that the UK Government under the unlamented Boris Johnson had in effect taken.

But there is no rational or legitimate excuse for doing anything about global warming on the basis of any current predictions, because, as Pat Frank has already demonstrated in his way and as we have demonstrated in ours, all predictions of global warming are mere guesswork. Would you trash the Western economies, and continue the inexorable transfer of industries, jobs, profits, wealth and global economic and political hegemony from the democratic, Judaeo-Christian, freedom-loving West to the grim oligarchs of Communist-led China and Russia on the basis of forecasts that are proven guesswork and are not borne out by events? We wouldn’t. I do hope that this has cleared up some misconceptions about our result.

— Christopher Monckton (4 October 2022)

Lord Christopher Monckton is a talented mathematician, and there are many things on which we agree. But it is unhelpful to the skeptical response to claims of a supposed climate emergency to be chasing rabbits down holes when others have already gone on that chase. So, what follows is my latest attempt to explain why Monckton’s feedback arguments supporting a very low climate sensitivity cannot be supported. This doesn’t mean his conclusion is wrong, only the line of reasoning that led him to that conclusion.

Couched in the obscure language of feedback analysis, and the mathematical gymnastics deriving from initial assumptions regarding those feedbacks, Lord Monckton’s latest explanation of his climate feedback theory (Why It Matters That Climatologists Forgot the Sun Was Shining) tends to skirt around actual physical processes. For if he were to actually investigate what meteorologists and climate scientists already know of atmospheric processes, he would not still be pushing his current theory.

Here I will try to explain, based upon actual atmospheric processes, why his argument does not make physical sense.

Christopher’s latest installment explaining his logic begins (emphasis added),

It is now almost two years since we submitted our paper on the central error perpetrated by climatologists in their attempts to derive climate sensitivity to anthropogenic greenhouse-gas forcings — namely, their failure to appreciate that such feedback processes as subsist in the climate system at any given moment must, at that moment, necessarily respond equally to each Kelvin of the entire reference temperature. Feedbacks do not, repeat not, respond solely to perturbation signals, the reference sensitivities. They also respond to the base signal, the emission temperature that would prevail even if there were no greenhouse gases in the air, because the Sun is shining.

I cannot emphasize enough just how bold (and wrong) the underlined assertion is. The idea that the climate system’s response to a small perturbation from its current state might be discerned from its response to the presence of solar heating assuming a theoretical initial cold Earth is not new, but was rejected many years ago based upon the known behavior of clouds and the atmospheric circulations associated with them.

The issue is not unlike the Ramanathan and Collins (1991) “cloud thermostat” hypothesis, which imagines that just because the Pacific Warm Pool is limited in its warmth by local cloud formation, that global warming will be limited by even more cloud formation. Hartmann and Michelsen (1993) and Lau et al. (1994) quickly responded to that claim by pointing out that vertical circulations created in the cloudy air must also produce descending, clear air elsewhere. Thus, more clouds on one region can actually cause fewer clouds elsewhere. This shows than even an expert in atmospheric radiative transfer (Ramanathan) could be misled without an adequate understanding of atmospheric circulation systems.

I’m not claiming that further warming of the climate system won’t be mitigated by an increase in clouds, as Monckton’s analysis implies. Just that we cannot get to that conclusion from the evidence presented.

Yes, Clouds Cool the Climate System

It has long been known that clouds, on average, cool the climate system. Sunlight heats the surface of the Earth, which combined with the atmospheric destabilization from the greenhouse effect, leads to convective heat transport away from the surface. Due to the presence of water, clouds form, reflecting sunlight back to outer space. While those clouds also enhance the water vapor-dominated greenhouse effect, the solar reflection (albedo) effect dominates, leading to the observation that clouds, on average, cool the climate system.

So, it might seem logical to assume (at least as a starting point) that any additional source of heating (positive energy imbalance) would lead to even more clouds, and thus a negative cloud feedback. As far as I can tell, this is the physical underpinning of Monckton’s argument. Of course, clouds might not be the only element of his argument, but clouds are arguably the most prominent example.

The trouble is that when clouds form, most of them are embedded in ascending air currents. All of that ascending air must be exactly matched by an equal amount of descending air, which is almost always cloud-free.

Thus, one cannot create more clouds without creating more clear air. When you experience a cloud-free day, it’s because ascending cloudy air with precipitation, hundreds of miles away, is forcing the air over you to sink. This is why cloud feedbacks are so uncertain, and why we cannot use the average base-state response of the climate system to the presence of sunlight to estimate climate sensitivity.

Another way to express this is that the climate system’s response to solar heating is non-linear. Initial warming from a base state of a cold, dark Earth to a solar heated one is to create clouds (a cooling effect), but the resulting vertical air circulations means you cannot created an ever more cloud-covered Earth with ever more heating. Descending air currents in response to rising air currents will not allow it.

Even Climate Models Tell Us This is the Case

Like weather forecast models, modern 3D climate models deal with the equations of motion, conservation of mass, energy, moist processes, and the atmospheric equation of state. In other words, they depend upon physics. (This does not mean all of those physical processes — especially cloud microphysical processes — are sufficiently well known to allow useful predictions of future average climate states. I don’t believe they are. My point is that the models depend upon our knowledge of the physics of a wide variety of complex processes.)

If you start-up a computerized climate model from an initial cold state (pick any cold temperature you want, say 50 Kelvin), with no clouds, the modeled system will warm, clouds will form, and the system will eventually reach a state of quasi-equilibrium, with the global area-average rate of absorbed solar energy equaling the average rate of infrared cooling to outer space. These results are consistent with the statement that “clouds cool the climate system”.

But if a small energy perturbation is then added (e.g. from more CO2 in the atmosphere reducing the rate of IR cooling, or from increasing the intensity of sunlight), clouds in the model will often respond by being reduced, not increased, in response to the small CO2-induced warming. Years ago we did this experiment with a limited-domain version of the ARPS cloud-resolving model. Global climate models would do the same thing.

The cloud response to the perturbation is not prescribed by the modelers as a cloud feedback. It is the result of the physics (and cloud microphysics) in the model. Climate model feedbacks are not prescribed; they are diagnosed after the model is run from model output.

I’m not claiming cloud feedbacks are negative or positive. Only that you cannot use the observation that “clouds cool the climate system” as a basis for determining cloud feedbacks in response to adding more CO2 to the atmosphere. And, as far as I can tell, this is the physical assumption Monckton makes in his feedback-based arguments.

Climate Sensitivity Does Not Depend Upon Feedback Analysis

For better or worse, Jule Charney and his co-authors in 1979 decided to use the forcing-feedback paradigm to explain the response of the climate system to increasing CO2. As a result, some climate skeptics have seized upon the lack of a direct one-to-one correspondence between feedbacks in electrical circuit design and climate feedback analysis. But the use of the forcing-feedback paradigm was simply a way for climate researchers to explain, in conceptual terms, how the climate system responds to an imposed energy imbalance.

While this paradigm has been useful (even quantitatively), the sensitivity of modern 3D climate models does not depend upon feedback analysis, per se. One could talk about sensitivity kernels or other plain-language terms for the partial derivatives without using the f-word. The feedback concepts which Lord Monckton imagines the climate system depends upon are only used by climate modelers as a simple way to conceptually describe the behavior (output) of climate models: that for an imposed energy imbalance in the climate system, a certain amount of warming takes place after all temperature-dependent adjustments (e.g. cloud and water vapor changes in response to warming) in the system occur. These temperature-dependent responses (feedbacks) either amplify (positive feedback) or reduce (negative feedback) the direct warming effect from the imposed energy imbalance. (Remember, almost without exception, the temperature change in anything is the result of energy imbalance).

Now, it is true that feedbacks in the models are indeed quantitatively diagnosed based upon perturbations from the models average pre-industrial climate state. But that is the only way it makes sense, because the warming in response to a perturbation (say, a doubling of atmospheric CO2) involves changes in (say) clouds from their average pre-industrial state. The fact that sunlight shining on a theoretical cold, dark earth creates warming which creates clouds (“climatologists forgot the sun is shining“) is not relevant to climate sensitivity — and even the climate models themselves (run from a cold, dark Earth state) will produce the process which Monckton imagines controls climate sensitivity.

I consider Christopher Monckton a friend, and I implore him to stop chasing this rabbit. I am asked about his ideas from time to time, and as a result I must, once again, attempt to explain why I believe he is wrong.

The Version 6.0 global average lower tropospheric temperature (LT) anomaly for September, 2022 was +0.24 deg. C, down slightly from the August, 2022 value of +0.28 deg. C.

The linear warming trend since January, 1979 still stands at +0.13 C/decade (+0.11 C/decade over the global-averaged oceans, and +0.18 C/decade over global-averaged land).

Various regional LT departures from the 30-year (1991-2020) average for the last 21 months are:

YEAR MO GLOBE NHEM. SHEM. TROPIC USA48 ARCTIC AUST 
2021 01  0.12  0.34 -0.09 -0.08  0.36  0.50 -0.52
2021 02  0.20  0.32  0.08 -0.14 -0.66  0.07 -0.27
2021 03 -0.01  0.13 -0.14 -0.29  0.59 -0.78 -0.79
2021 04 -0.05  0.05 -0.15 -0.28 -0.02  0.02  0.29
2021 05  0.08  0.14  0.03  0.06 -0.41 -0.04  0.02
2021 06 -0.01  0.30 -0.32 -0.14  1.44  0.63 -0.76
2021 07  0.20  0.33  0.07  0.13  0.58  0.43  0.80
2021 08  0.17  0.26  0.08  0.07  0.32  0.83 -0.02
2021 09  0.25  0.18  0.33  0.09  0.67  0.02  0.37
2021 10  0.37  0.46  0.27  0.33  0.84  0.63  0.06
2021 11  0.08  0.11  0.06  0.14  0.50 -0.43 -0.29
2021 12  0.21  0.27  0.15  0.03  1.63  0.01 -0.06
2022 01  0.03  0.06  0.00 -0.24 -0.13  0.68  0.09
2022 02 -0.00  0.01 -0.02 -0.24 -0.05 -0.31 -0.50
2022 03  0.15  0.27  0.02 -0.08  0.22  0.74  0.02
2022 04  0.26  0.35  0.18 -0.04 -0.26  0.45  0.60
2022 05  0.17  0.24  0.10  0.01  0.59  0.23  0.19
2022 06  0.06  0.07  0.04 -0.36  0.46  0.33  0.11
2022 07  0.36  0.37  0.35  0.13  0.84  0.55  0.65
2022 08  0.28  0.31  0.24 -0.04  0.59  0.50 -0.01
2022 09  0.24  0.43  0.06  0.03  0.88  0.69 -0.29

Hurricane Ian approaches SW Florida on 28 September 2022.

With Hurricane Ian (now a tropical storm) exiting the east coast of Florida, there is no shortage of news reports tying this storm to climate change. Even if those claims actually include data to support their case, those data are usually for cherry-picked regions and time periods. If global warming is causing a change in tropical cyclone activity, it should show up in global statistics.

The latest peer-reviewed study (March 2022, here) of the accumulated wind energy in tropical cyclones since 1990 (when we started have sufficient global data) showed a decrease in hurricane activity. There was an increase in Atlantic activity, but this was matched by an even larger decrease in Pacific activity, due to a shift from El Nino to La Nina conditions during that time.

So, yes, there is climate change involved in the uptick in Atlantic activity in recent decades. But it’s natural.

Looking at just the numbers of global hurricanes since 1980, we see no obvious trends.

Global hurricane activity counts by year during 1980-2021.

Even if we did see an increase, the improvements in global satellite monitoring would be responsible for some of that. It is impossible to talk about meaningful global statistics (especially trends) before the 1980s due to a lack of satellite data. Ships of opportunity are insufficient for trend calculations, especially since ships try to avoid storms, not sample them.

A document-based study of hurricanes impacting the Lesser Antilles since the late 1600s found a downward trend (not statistically significant) in hurricane activity during 1690-2007.

In my 2017 Kindle book Inevitable Disaster: Why Hurricanes Can’t Be Blamed on Global Warming, I looked at major hurricane landfalls in Florida, which showed no trends. With Hurricane Ian and Michael (2018) added to the dataset, there is still no statistically significant trends in either intensity or frequency of landfalling major hurricanes in Florida.

Major hurricane landfalls in Florida over the last 120 years.

Of course hurricane damages have increased dramatically during the same period, but this is due to the explosive growth in coastal infrastructure there. Miami had only 444 residents in 1896, and now the metro area has over 6,000,000 population. As seen in the following plot, Florida population has increased by a factor of over 40 since 1900.

Yearly population of Florida, 1900 through 2021.

Given that hurricanes will always be with us, what is the best defense against them? Wealth. Hurricane Ian came ashore with 150 mph sustained winds, but warnings from modern instrumentation and forecast tools led to mass evacuations. At this writing, only 5 deaths have been reported (I’m sure that will rise). Modern building codes help reduce wind damage. I watched storm chaser Reed Timmer live reporting from the eyewall of Hurricane Ian as it made landfall, and I didn’t see any roofs coming off the houses (but I’m sure there were some that did). Damage from storm surge flooding, however, will be extensive and costly.

The Version 6.0 global average lower tropospheric temperature (LT) anomaly for August, 2022 was +0.28 deg. C, down from the July, 2022 value of +0.36 deg. C.

The linear warming trend since January, 1979 still stands at +0.13 C/decade (+0.11 C/decade over the global-averaged oceans, and +0.18 C/decade over global-averaged land).

Various regional LT departures from the 30-year (1991-2020) average for the last 20 months are:

YEAR MO GLOBE NHEM. SHEM. TROPIC USA48 ARCTIC AUST 
2021 01  0.12  0.34 -0.09 -0.08  0.36  0.50 -0.52
2021 02  0.20  0.32  0.08 -0.14 -0.66  0.07 -0.27
2021 03 -0.01  0.13 -0.14 -0.29  0.59 -0.78 -0.79
2021 04 -0.05  0.05 -0.15 -0.28 -0.02  0.02  0.29
2021 05  0.08  0.14  0.03  0.06 -0.41 -0.04  0.02
2021 06 -0.01  0.30 -0.32 -0.14  1.44  0.63 -0.76
2021 07  0.20  0.33  0.07  0.13  0.58  0.43  0.80
2021 08  0.17  0.26  0.08  0.07  0.32  0.83 -0.02
2021 09  0.25  0.18  0.33  0.09  0.67  0.02  0.37
2021 10  0.37  0.46  0.27  0.33  0.84  0.63  0.06
2021 11  0.08  0.11  0.06  0.14  0.50 -0.43 -0.29
2021 12  0.21  0.27  0.15  0.03  1.63  0.01 -0.06
2022 01  0.03  0.06  0.00 -0.24 -0.13  0.68  0.09
2022 02 -0.00  0.01 -0.02 -0.24 -0.05 -0.31 -0.50
2022 03  0.15  0.27  0.02 -0.08  0.22  0.74  0.02
2022 04  0.26  0.35  0.18 -0.04 -0.26  0.45  0.60
2022 05  0.17  0.24  0.10  0.01  0.59  0.23  0.19
2022 06  0.06  0.07  0.04 -0.36  0.46  0.33  0.11
2022 07  0.36  0.37  0.35  0.13  0.84  0.55  0.65
2022 08  0.28  0.31  0.24 -0.04  0.59  0.50 -0.01

Key Points

• Contrary to claims that drought is causing Lake Mead water levels to fall, the Colorado River natural flows into Lake Mead show no long-term trend since 1930.
• Decadal time scale variations in river flow do occur, though, related to the Pacific Decadal Oscillation (PDO).
• Since about 2000, use of Lake Mead water has exceeded river inflow, causing water levels to drop. The negative phase of the PDO since that time has exacerbated the problem.

Natural Water Flows into Lake Mead: No Long-Term Trend

Record low water levels in Lake Mead are widely blamed on drought, although what “drought” means is seldom specified. The public perception is that lower precipitation amounts have reduced water supply to Lake Mead (which comes from the Colorado River), usually attributed to human-caused climate change, and that this is why water levels are falling.

But data from the U.S. Bureau of Reclamation (USBR) show that there has been no long-term trend in natural Colorado River flow into Lake Mead:

Fig. 1. Yearly “natural” water flows into Lake Mead, corrected for local human-induced changes in water flow upstream. Details of those corrections are described here. Data source here.

The flows in Fig. 1 have been slightly adjusted for local human-caused changes to the flows upstream, and provide our best answer to the question of whether long-term global climate change is responsible for a decrease of river water flow into Lake Mead.

The answer is “no”.

Does Climate Change Theory Even Predict Reduced Precipitation? No

The next question is, does climate change even predict future reductions of precipitation over the Colorado River watershed? The following plot shows an average of 183 climate model simulations of average yearly precipitation in an area approximating the Colorado River watershed. The models suggest a slight increase in total precipitation with warming.

Fig. 2. CMIP6 model average yearly precipitation 1930-2050 over an area approximating the upper Colorado River watershed. Data source here.

Most of the water entering Lake Mead is from snowmelt in the mountains; little of the water falling on lower elevations tends to be used by local vegetation with little runoff reaching the Colorado River. Fig. 3 shows there has been no long-term trend in the snowpack measurements in the upper Colorado River watershed.

Fig. 3. April snowpack measurements in the upper Colorado River watershed, 1938-2022.

So, not only has there been no observed long-term reduction in water flow into Lake Mead, or reduction in the watershed snowpack, climate change theory doesn’t even support such a change up to the current time (or even to 2050).

So, Why are Lake Mead Water Levels Falling?

What has changed since Hoover Dam was constructed in the 1930s is the amount of water being removed from Lake Mead. Since about 2000, that water use has exceeded the water input into the lake. This is the most recent available demonstration of that fact, published in 2012:

Fig. 4. The Colorado River basin water supply exceeded demand up until the year 2000 or so, and since then Lake Mead water levels have fallen due to overuse.

As long as water use exceeds supply, Lake Mead water levels will continue to fall. (This is somewhat dependent upon the regulated releases from Lake Powell, upstream. There is a “Fill Mead First” initiative that would draw down Lake Powell in an attempt to raise Lake Mead, based upon calculations that net natural water losses from combined evaporation and bank seepage from Mead and Powell would be reduced.)

The Role of the Pacific Decadal Oscillation (PDO) in the Current Problem

While the major problem with Lake Mead is overuse, there are multi-decadal fluctuations in Colorado River flows which have made matters worse since approximately 2000. If we take the river flow data in Fig. 1 and compute the accumulated departures from the long-term average flow (because this is how a reservoir like Lake Mead responds), we find that there have been periods of lesser and greater flows.

Fig. 5. As in Fig. 1, except time-accumulated departures-from-average Colorado River flows into Lake Mead.

Before the 1980s, there was somewhat reduced river flow into Lake Mead, but it made little difference because water use (Fig. 4) was still low.

Then from the 1982-83 super El Nino year to approximately 2000 there were above average flows, so Lake Mead could handle the increasing water usage. In fact, the lake reached near full-pool status.

But as usage peaked around 2000, river input to the lake was reduced once again. This put Lake Mead into an unsustainable state where more water was being extracted than the Colorado River could replenish it.

It has been long known (e.g. here) that precipitation in this region is affected by El Nino (more precip) and La Nina (less precip). Also, the Pacific Decadal Oscillation (PDO), which is basically a low-frequency manifestation of El Nino and La Nina activity is related to precipitation in this area.

I computed the cumulative average departures from the long-term mean of both the PDO index and the MEI (Multivariate ENSO Index). The PDO is somewhat higher correlated (r=0.52) with the cumulative river flow data in Fig. 5. As Fig. 6 shows, positive PDO periods are generally associated with higher stream flows, and negative PDO with lower stream flows. Most notably, the period since 2000 has seen more negative PDO activity, which is worsening the problem with Lake Mead not getting enough water. Of course, this will eventually reverse when the PDO flips back into its positive phase.

Fig. 6. Cumulative departures of the Pacific Decadal Oscillation index from its long term mean, which is r=0.52 correlated to cumulative streamflow into Lake Mead from the Colorado River (Fig. 5.)

Conclusions

The popular narrative that drought due to climate change is causing Lake Mead to have less water available to it is incorrect. Since 1930, there has been no long-term change in the Colorado River flow upstream of what is now Lake Mead.

The latest climate models do not even predict a reduction in precipitation in the upper Colorado River watershed.

Multi-decadal changes in river flow do occur, though, and are related to the Pacific Decadal Oscillation, a natural fluctuation in weather patterns over the northeast Pacific. Recent record-low water levels in Lake Mead are primarily due to record high water demand from the lake, since approximately 2000. The problem is being made somewhat worse by the negative phase of the PDO, also since approximately 2000.

UPDATED: Fixed Bureau of Reclamation study link, added Colorado River basin snowpack graph and discussion.

In today’s news is yet another article claiming the record-low water levels in Lake Mead (a manmade water reservoir) are due to human-caused climate change. In fact, to make the problem even more sinister, the Mafia is also part of the story:

Climate change is uncovering gruesome mafia secrets in this Las Vegas lake

While it is true that recent years have seen somewhat less water available from the Colorado River basin watershed (which supplies 97% of Lake Mead’s water), this is after years of above-average water inflow from mountain snowpack. Those decadal time-scale changes are mostly the result of stronger El Nino years (more mountain snows) giving way to stronger La Nina years (less snow).

The result is record-low water levels:

Lake Mead water levels since the construction of Hoover Dam (source: NBC News)

But the real problem isn’t natural water availability. It’s water use.

The following graph shows the fundamental problem (click for full resolution). Since approximately 2000, water use by 25 million people (who like to live in a semi-desert area where the sun shines almost every day) has increased to the point that more water is now being taken out of the Lake Mead reservoir than nature can re-supply it.

This figure is from a detailed study by the U.S. Bureau of Reclamation. As long as that blue line (water supply) stayed above the red line (water use), there was more than enough water to please everyone.

But now, excessive demand for water means Lake Mead water levels will probably continue to decline unless water use is restricted in some way. The study’s projection for the future in the above figure, which includes climate model projections, shows little future change in water supply compared to natural variability over the last century.

The real problem is that too much water is being taken out of the reservoir.

As long as the red line stays above the blue line, Lake Mead water levels will continue to fall.

But to blame this on climate change, whether natural or anthropogenic, ignores the thirsty elephant in the room.

UPDATE: Since it was pointed out in comments (below) that the latest Bureau of Reclamation study is rather dated (2012), and supposedly the drought has worsened since then, here’s a plot of the Colorado River basin April (peak month) snowpack, which provides about 50% of the water to Lake Mead. The rest is provided in the non-mountainous areas of the river basin, which should be highly correlated with the mountainous regions. I see no evidence for reduced snowpack due to “climate change”… maybe the recent drought conditions are where the demand by 25 million water consumers originates from, causing higher demand?

April snowpack in the Colorado River basin, the greatest source of water input to Lake Mead (data from https://www.nrcs.usda.gov/Internet/WCIS/AWS_PLOTS/basinCharts/POR/WTEQ/assocHUCco_8/colorado_headwaters.html)

SUMMARY: A simple time-dependent CO2 budget model shows that yearly anthropogenic emissions compared to Mauna Loa CO2 measurements gives a declining CO2 sink rate, which if continued would increase atmospheric CO2 concentrations and presumably anthropogenic climate change. But accounting for ENSO (El Nino/La Nina) activity during 1959-2021 removes the decline. This is contrary to multiple previous studies that claimed to account for ENSO. A preprint of my paper (not yet peer reviewed) describing the details is at ENSO Impact on the Declining CO2 Sink Rate | Earth and Space Science Open Archive (essoar.org).

UPDATE: The CO2 model, with inputs and outputs, is in an Excel spreadsheet here: CO2-budget-model-with-EIA-growth-cases.

I decided that the CO2 model I developed a few years ago, and recently reported on here, was worthy of publication, so I started going through the published literature on the subject. This is a necessary first step if you want to publish a paper and not be embarrassed by reinventing the wheel or claiming something others have already “disproved”.

The first thing I found was that my idea that Nature each year removes a set fraction of the difference between the observed CO2 concentration and some baseline value is not new. That idea was first published in 2013 (see my preprint link above for details), and it’s called the “CO2 sink rate”.

The second thing I found was that the sink rate has (reportedly) been declining, by as much as 0.54% (relative) per year, even after accounting for ENSO activity. But I only get -0.33% per year (1959-2021) before accounting for ENSO activity, and — importantly — 0.0% per year after accounting for ENSO.

This last finding will surely be controversial, because it could mean CO2 in the atmosphere will not rise as much as global carbon cycle modelers say it will. So, I am posting the model and the datasets used along with the paper preprint at ENSO Impact on the Declining CO2 Sink Rate | Earth and Space Science Open Archive (essoar.org). The analysis is quite simple and I believe defensible. The 2019 paper that got -0.54% per year decline in the sink rate uses complex statistical gymnastics, with a professional statistician as a primary author. My analysis is much simpler, easier to understand, and (I believe) at least as defensible.

The paper will be submitted to Geophysical Research Letters for peer review in the next couple days. In the meantime, I will be inviting the researchers who live and breathe this stuff to poke holes in my analysis.

The Version 6.0 global average lower tropospheric temperature (LT) anomaly for July, 2022 was +0.36 deg. C, up from the June, 2022 value of +0.06 deg. C.

The linear warming trend since January, 1979 still stands at +0.13 C/decade (+0.11 C/decade over the global-averaged oceans, and +0.18 C/decade over global-averaged land).

Various regional LT departures from the 30-year (1991-2020) average for the last 19 months are:

YEAR MO GLOBE NHEM. SHEM. TROPIC USA48 ARCTIC AUST 
2021 01  0.12  0.34 -0.09 -0.08  0.36  0.50 -0.52
2021 02  0.20  0.32  0.08 -0.14 -0.66  0.07 -0.27
2021 03 -0.01  0.13 -0.14 -0.29  0.59 -0.78 -0.79
2021 04 -0.05  0.05 -0.15 -0.28 -0.02  0.02  0.29
2021 05  0.08  0.14  0.03  0.06 -0.41 -0.04  0.02
2021 06 -0.01  0.30 -0.32 -0.14  1.44  0.63 -0.76
2021 07  0.20  0.33  0.07  0.13  0.58  0.43  0.80
2021 08  0.17  0.26  0.08  0.07  0.32  0.83 -0.02
2021 09  0.25  0.18  0.33  0.09  0.67  0.02  0.37
2021 10  0.37  0.46  0.27  0.33  0.84  0.63  0.06
2021 11  0.08  0.11  0.06  0.14  0.50 -0.43 -0.29
2021 12  0.21  0.27  0.15  0.03  1.63  0.01 -0.06
2022 01  0.03  0.06  0.00 -0.24 -0.13  0.68  0.09
2022 02 -0.00  0.01 -0.02 -0.24 -0.05 -0.31 -0.50
2022 03  0.15  0.27  0.02 -0.08  0.22  0.74  0.02
2022 04  0.26  0.35  0.18 -0.04 -0.26  0.45  0.60
2022 05  0.17  0.24  0.10  0.01  0.59  0.23  0.19
2022 06  0.06  0.07  0.04 -0.36  0.46  0.33  0.11
2022 07  0.36  0.37  0.35  0.13  0.70  0.55  0.65