A recent set of posts at Anthony Watt’s blog, particularly this one has sparked some interest over the internet as of late. From a quick glance, it looks like negative trends in specific humidity over the last half a century. Readers were quick to pick up on the connection to water vapor feedback, which is expected to at least double the sensitivity of climate to external perturbations (e.g., by human released CO2).
The accepted theory of water vapor feedback in climate change is that in a warming world the global relative humidity will not change much, and an increase in temperature with little change in relative humidity means an increase in specific humidity. So any increase in temperature caused by something (maybe CO2, increased solar irradiance, etc) increases the saturation vapor pressure which allows water vapor concentration to go up, which further amplifies the greenhouse effect since H2O is also a strong greenhouse gas being the most important gaseous source of infrared opacity in the atmosphere. Most of the water vapor feedback relevant for global warming occurs in the upper layers of the atmosphere, which are considerably drier than the boundary layer. Only about 10% of the water vapor feedback is from below 800 mb.
So what do the plots in Anthony Watts blog show? Well, they show specific humidity as a function of time, with multiple graphs corresponding to different altitudes in the atmosphere. For example, at 700 mb
and so on…
The figures come from plotting annual specific humidity at global coverage from this site. Now where is this data coming from and how exactly does it relate to reality? The constructed timeseries is from the NCEP Reanalysis Dataset. The goal of the NCEP/NCAR reanalysis project is to “produce new atmospheric analyses using historical data (1948 onwards) and as well to produce analyses of the current atmospheric state.” Application has been used to process multi-decadal sequences of past observations using modern data assimilation techniques, which brings its share of observational and model problems (coverages and bias varying over time, biases in background forecasts which are combined with observations over a short time period to get an analysis of the state of the atmosphere, etc). Radiosondes provide water vapor information in the atmosphere since the 1940′s, but earlier products had a lot of biases, and since then changes in instrumentation have taken place which may lead to discontinuities in the data, and problems arose especially for upper atmospheric data. Reanalysis products contain discontinuities from changes in observational data available, changes in error characteristics of data, changes in sampling, better assimilation techniques, and other things. These make them virtually unusable for examining trends and low-frequency variability. Reanalysis products are not reliable sources for trends on water vapor, precipitation, clouds, etc though estimates other climatic variables like temperature do well after the satellite era. Here is a quote from Soden et al (2005),
“Although an international network of weather
balloons has carried water vapor sensors for
more than half a century, changes in instrumentation
and poor calibration make such
sensors unsuitable for detecting trends in
upper tropospheric water vapor (27). Similarly,
global reanalysis products also suffer
from spurious variability and trends related to
changes in data quality and data coverage (24).”
Satellites are a nice tool though. By measuring the upwelling radiance in different spectral bands which water vapor absorbs, you can obtain measurements of water vapor concentrations in various parts of the atmosphere. The Soden paper, for example, uses satellites to detect upper tropospheric moistening from 1982 to 2004 as evidenced by changing emission levels in T12 (the High Resolution Infrared radiation sounder channel 12) because of increased opacity to water vapor. As the IPCC AR4 report discusses in Chapter 3, interannual variability is not often captured well by reanalysis techniques, even after the satellite era. Probably the best paper discussing upper atmospheric trends is the above paper, and IPCC AR4 (Chapter 3) also summarizes the literature and supports the conclusion of no detectable changes in relative humidity, but trends in increased specific humidity.
Putting aside the data and model conclusions, from a purely theoretical framework, a lack of water vapor response to global warming (especially in the upper troposphere) would mean a very insensitive climate system, as the discussion here (see final figure as well) goes over. The water vapor effect also works in sync with other feedback effects. For example, more water vapor means more warming which means more of an ice-albedo effect, and that in turn means more warming. If climate sensitivity were very low (they would be on the low end of the IPCC scale, or lower if the WV feedback effect was not substantial) then we would not be able to explain the variability over the paleoclimatic record, or the 20th century rise. But as a lot of research has shown (see empirical evidence by James Annan for instance) there is unlikely to be anything wrong with the estimates of 2 to 4.5 C per doubling of CO2. Overall, I’ve not been convinced that there is anything significantly wrong with our understanding of water vapor feedback, or that (as some of Watts’ readers proclaim) that this is some nail in the coffin for AGW.