By Andy May
Christian Freuer has translated this post into German here.
In the latest IPCC major report, AR6, they report: “a best estimate of equilibrium climate sensitivity of 3°C, with a very likely range of 2°C to 5°C”. They also report that CO2 concentration might control climate change. This estimate includes the laboratory estimate that CO2 alone, if doubled in the atmosphere, would increase the average global surface temperature about 1.2°C. The remaining 1.8°C (60%) is supposedly achieved through feedbacks to the initial CO2-caused warming. The main proposed feedback mechanism is an increase in atmospheric water vapor caused by the CO2 warmed air. Water vapor is a more powerful greenhouse gas than CO2, due to its greater abundance, and can cause more warming.
The IPCC estimates the water vapor feedback using the Clausius-Clapeyron relation. The relation states that as temperature increases, more evaporation occurs and atmospheric water vapor increases, especially in the upper atmosphere. Models suggest that relative humidity should remain “roughly constant,” as climate changes. According to AR5 and AR6:
“… total column water vapour (TCWV) very likely increased since the 1970s, at a rate that was overall consistent with the Clausius-Clapeyron relationship (about 7% per °C) given the observed increase in atmospheric temperature.”(IPCC 2021, p 330).
Later in the report they write this:
“The Clausius–Clapeyron equation determines that low-altitude specific humidity increases by about 7% °C–1 of warming, assuming that relative humidity remains constant, which is approximately true at a global scale but not necessarily valid regionally.”(IPCC 2021, p 1065).
A corollary to the Clausius-Clapeyron relationship is that as the specific humidity (total water vapor in the air, the units used here are kg of water vapor per kg of dry air) goes up precipitation also increases by about the same amount, keeping relative humidity about the same. Model studies suggest that the increase in precipitation is less, around 3%. These assumptions make some sense logically, but they are not definitively supported by real world data. Further, it appears that the rate of evaporation is strongly influenced by wind speed, as well as temperature in the real world.
AR6 makes an oblique reference to the evaporation and total atmospheric precipitable water (TPW, shown here as kg/m2) dependence upon wind speed but refers to it as a “dynamical process.”
“According to theory, observations and models, the water vapour increase approximately follows the Clausius–Clapeyron relationship at the global scale with regional differences dominated by dynamical processes …. Greater atmospheric water vapour content, particularly in the upper troposphere, results in enhanced absorption of LW [longwave] and SW [shortwave] radiation and reduced outgoing radiation. This is a positive feedback.”(AR6, p 969).
The Clausius-Clapeyron relationship between specific humidity and temperature is sound in the laboratory, but observations show the relationship between humidity, temperature, and precipitation in the real world is more complex. Benestad (2016) reported that the European Centre for Medium-range Weather forecasts (ECMWF) interim reanalysis shows the total volume of water vapor in the atmosphere decreasing by -0.018 kg/m2 per decade from 1979-2011, a period of rapid global warming. This is a decrease of about .07%/decade or 0.007%/year (1979-2011). Miskolczi (2014) reports that the NOAA R1 dataset shows that global surface air temperature has increased 0.687K between 1948 and 2008, but the water content has decreased by 0.636% or -0.0106%/year, like what is seen in the ECWMF dataset.
The NOAA R1 dataset still exists and is a weather reanalysis dataset maintained by NCEP, it has been updated since Miskolczi used it in 2014. Figure 1 shows it and its successor NCEP R2, as well as the HadCRUT4 global surface temperature dataset.
In figure 1 we see that before 1978, NCEP R1 TPW declines more steeply than temperature, but both decline. After 2005, temperature and both NCEP reanalysis datasets increase, but again at different rates. Between 1978 and 2005 TPW declines in both datasets and temperature increases quickly. This is a 27-year period, why opposite trends? Obviously, global temperatures are not the only thing influencing TPW and the impact of temperature is not that significant. Yu and Weller emphasize that wind speed is a strong modulator of surface latent heat flux (evaporation). Wind moves saturated air out of the way, so that evaporation can continue. Evaporation stimulates circulation since water vapor is less dense than dry air and humid air usually rises as a result.
NCEP reanalysis 1 and 2 are separated when plotted in kg/m2, but they look much more alike when plotted as anomalies from the 1988-2022 mean as shown in figure 2. We have also added the HadCRUT4 global temperature anomaly from the 1988-2021 mean. The difference between the temperature anomaly and the TPW anomalies in the blue shaded area is more dramatic between 1978 and 2005 in this plot.
Carl Mears and colleagues have published a satellite microwave brightness record of TPW over the world’s ice-free oceans, from roughly 60°S to 60°N. In addition, to covering only the ice-free oceans, heavy rain affects the signal, and these areas must be excluded. Since it is not truly a global dataset, like NCEP R1 and R2, and it excludes the drier land and polar areas, it has a much higher absolute TPW than the NCEP datasets. But we can compare it to NCEP as an anomaly to the 1988-2022 mean, see Figure 3.
The more recent ECMWF-ERA5 specific humidity plotted in figure 3 has different units than the NCEP or RSS datasets. The ECMWF-ERA5 values plotted are the global area-weighted average specific humidity (kg water vapor per kg of dry air) from 1000 mbar to 1 mbar (roughly .1 km to 32 km altitude). Figure 3 is quite busy, so to make the relationship between the modern ECMWF-ERA5 specific humidity and HadCRUT4 clearer, I include Figure 4 below.
Figure 3 shows that the apparent correlation between global temperature and the RSS TPW dataset is just that—an apparent correlation. The RSS TPW rises much faster than temperatures, and the comparison ignores both polar regions. Figure 5 shows where the RSS data comes from in brighter colors. The darker blue and black areas are not used in the RSS “global” average.
It is pretty clear that while temperature must have some influence on total precipitable water, it isn’t the only influence. Many will argue that TPW estimates from NOAA, ECMWF, and the satellite measurements by RSS are all poor, and they would be correct. But the trends in all the datasets agree after 1960, except in the blue anomalous area. NCEP R1 and the AMO (Atlantic Multidecadal Oscillation) have similar trends as shown in Figure 6. Figure 6, is the straight AMO, and not the detrended AMO index you often see, thus the units are degrees C. It is the area-weighted average sea surface temperature from, roughly, 0 to 70°N latitude and 0-80°W longitude (see here).
The AMO and other ocean oscillations might influence TPW, but it is hard to tell since many have questioned the quality of the early weather balloon hygrometer data, and modern estimates of TPW, like ECMWF-ERA5 show less of a correlation.
Over the short term, say 3-4 years (ENSO spans of time), the correlation between TPW and temperature trends is good, as shown in Figure 3. Figure 3 shows that El Niños and ENSO in general, have a large influence on TPW, but since these oscillations affect the transfer of both heat and moisture from the ocean to the atmosphere, this is not surprising. Over the blue shaded 27-year period, using the ECMWF-ERA5, NCEP R1 and R2 data, the correlation is poor. In this period TPW trends downward as temperature increases, why? I don’t think anyone knows. As seen in figures 2 and 3, the correlation deteriorates in earlier time periods, probably due to poor data quality. The correlation is visually good after 2005. The time period and the data selected matters. We can see why AR5 and AR6 do not provide plots that compare global surface temperature to TPW.
TPW in the Upper Troposphere
As Garth Paltridge, et al. have noted climate models predict that specific humidity will increase in the upper troposphere as global warming continues. Yet, this is not what they see in the NCEP reanalysis 1 data as shown in Figure 7. Paltridge, et al. found that all levels above 850 mbar (~1.5 km) have a negative trend through 2007 in the tropics and southern midlatitudes in that dataset.
Remember the quote from AR6 from earlier in the post? I repeat part of it here:
“Greater atmospheric water vapour content, particularly in the upper troposphere, results in enhanced absorption of LW and SW radiation and reduced outgoing radiation. This is a positive feedback.”(AR6, p 969).
Specific humidity from the more modern ECMWF-ERA5 dataset correlates better with surface temperature at 500 mbar (~5.6 km). This point is made by Dessler and Davis in a rebuttal to Paltridge, et al. However, the correlation between 1985 and 2008 is still poor and neither Dessler and Davis, nor the IPCC address this problem. The area of poor correlation is highlighted in this post with blue shading. At 500 mbar, the poor correlation is moved forward about seven years, as shown in figure 8, but it is still there.
In many ways the opposite trends in figure 7 are counterintuitive since logically we would expect more evaporation with warming. More evaporation should cause a higher TPW, unless rain efficiency increases. From Paltridge, et al.:
“Negative trends in q [TPW] as found in the NCEP data would imply that long-term water vapor feedback is negative—that it would reduce rather than amplify the response of the climate system to external forcing such as that from increasing atmospheric CO2.”(Paltridge, Arking and Pook 2009).
This was also the conclusion reached by Ferenc Miskolczi (Miskolczi 2014). Others, such as Roy Spencer and Richard Lindzen, have suggested that warmer temperature will cause more clouds, which will increase the albedo of the Earth and lower temperatures or reduce the rate of warming (provide negative feedback) as a result. David Enfield, et al. show that rainfall patterns in the United States are closely related to the AMO, yet the climate models do not take the AMO into account. Obviously, rainfall affects TPW. The world is more complicated than the Clausius-Clapeyron relation suggests.
Dessler and Davis rebuttal
AR6 has very little discussion of the Clausius-Clapeyron relation and refers to AR5:
“According to AR5, radiosonde, Global Positioning System (GPS) and satellite observations of tropospheric water vapour indicate very likely increases at near global scales since the 1970s occurring at a rate that is generally consistent with the Clausius–Clapeyron relation (about 7% °C–1 at low altitudes) and the observed atmospheric warming”(IPCC 2021, p 1080).
AR6 provides no chapter number, section or page number in AR5, but we were able to find the following:
“Because global temperatures have been rising, the above arguments imply WVMR [water vapor mixing ratio, that is the specific humidity] should be rising accordingly, and multiple observing systems indeed show this … A study challenging the water vapour increase (Paltridge et al., 2009) used an old reanalysis product, whose trends are contradicted by newer ones (Dessler and Davis, 2010) and by actual observations.”AR5 p 586 Ch 7.
Dessler and Davis point out that the newer reanalysis datasets, like ECMWF, do not show a downward trend in specific humidity and that ENSO is reflected in the specific humidity. This is true. However, AR5, AR6, and Dessler and Davis, do not plot surface temperature versus specific humidity as we have done here. Thus, they do not explain why the trends do not correlate well over the 1978-2008 period. Dessler and Davis point out that:
“Our understanding of upper tropospheric water vapor suggests that it should be in relatively close thermodynamic equilibrium with the surface temperature on time scales of longer than about 1 month [e.g., Minschwaner and Dessler, 2004]. Thus, the water vapor response to a climate fluctuation with a time scale of a few years (e.g., ENSO) should be about the same as for long‐term warming.”(Dessler and Davis 2010)
We have no problem with the one-month equilibrium period, but concluding that ENSO-related warming should be the same as long-term warming is inconsistent with the data shown above for the 1978-2008 period. Data that the IPCC and Dessler and Davis ignore.
Conclusions and Summary
The various estimates of total atmosphere TPW and specific humidity available do not agree with one another very well. Even the two NCEP estimates, both global, vary by 3% over 1988-2022. The NCEP R1 data gathering and processing procedures were very complex and prone to error, which was why NCEP R2 was developed. NCEP R2 was of much higher quality than NCEP R1, but since NCEP R1 goes back to 1948, it has been cleaned up as much as possible, and is still used. It must be viewed with the understanding that the data prior to 1979 is of lower quality.
These global estimates are 16% lower than the RSS ocean-only ~60S to 60N TPW estimate. However, this is explainable. The atmospheric water vapor content over oceans and in the lower latitudes is much higher than over land and in the higher latitudes. The three estimates are compared in Figure 9.
Since about 2005 all the total atmospheric water vapor estimates trend upward, as the AMO begins to flatten at a high level. Prior to 2005, the story is more complex. The longer NCEP reanalysis 1 estimate trends down from 1948 to 1975 in sync with the AMO, but different from the HADCRUT4 and ECMWF trends. All datasets agree that short term ENSO changes (~5 years) are reflected in total atmosphere TPW, but it is not clear that longer-term changes (>30 years) in TPW are related solely to global surface temperatures, they seem to be impacted by other factors as well, perhaps including the AMO.
Global climate models predict that global warming will increase upper tropospheric specific humidity, but reanalysis, based mainly on weather balloon data, shows a decline in specific humidity and in TPW from 1978 to 2005 in the global atmosphere, and a flattening from 1985 to 2008 in the upper troposphere, both are periods of rapid surface warming. The humidity data declines in quality with altitude and lower temperatures, but even in the tropics where water vapor concentration is high at high altitudes, this trend persists. This also contradicts satellite data, but the ability of satellites to separate the signal of the upper troposphere water vapor from the lower is unclear. The accuracy of the specific humidity calculations in the upper troposphere can be questioned. However, both the NCEP reanalysis and the European reanalysis show a decline or flattening during the period of interest.
Consider this quote from Pierrehumbert (Pierrehumbert 2011):
“For present Earth conditions, CO2 accounts for about a third of the clear-sky greenhouse effect in the tropics and for a somewhat greater portion in the drier, colder extratropics; the remainder is mostly due to water vapor.”(Pierrehumbert 2011).
So, we see the crucial role assumed for water vapor in the entire man-made climate change catastrophe hypothesis. CO2 has only a minor role to play in warming the Earth. It is only the assumed, but poorly measured, feedback from water vapor that allows a possibly large impact on our climate to be calculated. Yet, as shown above, this assumed feedback cannot be measured with any accuracy with the data we have available. In fact, over some climate-relevant time scales (~30 years) we cannot even be sure the net feedback is positive. There is a strong correlation between surface temperature and total atmospheric water vapor concentration over short time periods, but it falls apart over, at least some, longer periods. I agree some of the data presented in this post is questionable, but it is data, and data trumps IPCC models. From Paltridge, et al.
“… it is important that the trends of water vapor shown by the NCEP data for the middle and upper troposphere should not be ‘written off’ simply on the basis that they are not supported by climate models—or indeed on the basis that they are not supported by the few relevant satellite measurements.”(Paltridge, Arking and Pook 2009).
Bottom line, water vapor feedback is a huge (66% according to Pierrehumbert) part of the dangerous greenhouse gas hypothesis. Total atmospheric water vapor content is very difficult to measure accurately, but the measurements and trends we have today do not support the hypothesis over all time periods. It seems likely that the Clausius-Clapeyron relation is not the only factor affecting TPW. This casts considerable doubt on the CMIP6 model results, which rely only on Clausius-Clapeyron, human activities, and sporadic volcanism.
AR5 and Dessler and Davis claimed in 2013 and 2010 respectively, that:
“In summary, radiosonde, GPS and satellite observations of tropospheric water vapour indicate very likely increases at near global scales since the 1970s occurring at a rate that is generally consistent with the Clausius-Clapeyron relation (about 7% per degree Celsius) and the observed increase in atmospheric temperature.”(IPCC 2013, p 208)
AR6 simply references these sources and assumes that specific humidity (TPW) responds to temperatures and is a positive feedback. However, the data shown in this post casts doubt on the quote above and the AR6 assumption. Thus, the data we have, poor as it is, does not support the idea that the Clausius-Clapeyron relation works at all time scales.
The R code and other information, including spreadsheets containing the data used to make the figures in the post can be downloaded here.
Download the bibliography here.
(IPCC 2021, p 93) ↑
(IPCC 2021, p 179), (Manabe and Wetherald 1967), and the National Research Council Charney Report (Charney, et al. 1979) ↑
(Lacis, et al. 2010), (Lacis, et al. 2013), (Dessler 2013), (Wijngaarden and Happer 2020) ↑
(Lacis, et al. 2010) ↑
(Wijngaarden and Happer 2020) ↑
(IPCC 2021, p 969) ↑
(Allen and Ingram 2002) ↑
(Allen and Ingram 2002) ↑
(Yu and Weller 2007) ↑
(Mears, et al. 2018) ↑
(Paltridge, Arking and Pook 2009) ↑
(Dessler and Davis 2010) ↑
(Enfield, Mestas-Nunez and Trimble 2001) ↑
(Kanamitsu, Ebisuzaki, et al., NCEP-DOE AMIP-II Reanalysis (R-2) 2002) ↑
(Dessler and Davis 2010) ↑