UPDATE: I’ve added an animation to the end of the post. It illustrates the changes in specific humidity anomalies in response to the 1997/98 El Niño and the 1998-01 La Niña.
The UKMO Hadley Centre created a global surface (ocean and land) humidity dataset that runs from 1973 to 2003. It’s known as HadCRUH. This humidity dataset was introduced in Katharine Willett’s PhD thesis, Creation and Analysis of HadCRUH: a New Global Surface Humidity Dataset, and further documented in the 2008 Willett et al paper Recent Changes in Surface Humidity: Development of the HadCRUH Dataset. It was also the observations-based dataset used in Willett et al (2007) Attribution of Observed Surface Humidity Changes to Human Influence. We’ll call the last paper the “Willett et al attribution paper” to differentiate it from the others.
The HadCRUH data are available through the KNMI Climate Explorer in specific humidity (g/kg), Figure 1, and relative humidity (%) forms. I have not presented the relative humidity data in this post.
The abstract for the Willett et al attribution paper reads:
Water vapour is the most important contributor to the natural greenhouse effect, and the amount of water vapour in the atmosphere is expected to increase under conditions of greenhouse-gas induced warming, leading to a significant feedback on anthropogenic climate change. Theoretical and modelling studies predict that relative humidity will remain approximately constant at the global scale as the climate warms, leading to an increase in specific humidity. Although significant increases in surface specific humidity have been identified in several regions, and on the global scale in non-homogenized data, it has not been shown whether these changes are due to natural or human influences on climate. Here we use a new quality-controlled and homogenized gridded observational data set of surface humidity, with output from a coupled climate model, to identify and explore the causes of changes in surface specific humidity over the late twentieth century. We identify a significant global-scale increase in surface specific humidity that is attributable mainly to human influence. Specific humidity is found to have increased in response to rising temperatures, with relative humidity remaining approximately constant. These changes may have important implications, because atmospheric humidity is a key variable in determining the geographical distribution and maximum intensity of precipitation, the potential maximum intensity of tropical cyclones, and human heat stress, and has important effects on the biosphere and surface hydrology.
We know that climate models can’t properly simulate the natural processes that cause sea surface temperatures to warm, so their attribution of the rise in specific humidity to human influence is questionable.
What I found remarkable about the Willett et al attribution paper was that sea surface temperatures were never mentioned, though papers about sea surface temperature were cited.
Figure 1 presents the global specific humidity anomalies based on the HadCRUH dataset for the full term of the data. As shown, the dataset unfortunately ends in December 2003. The effects of the 1986/87/88 and 1997/98 El Niños stand out like sore thumbs, and the response to the 1973-76 La Niña is tough to miss, as is the apparent impact of the 1976 Pacific Climate Shift.
In many respects, the specific humidity data looks like noisy sea surface temperature data. Let’s see how closely they compare.
The last sentence of the abstract for the Willett et al (2008), the description paper, reads:
A strong positive bias is apparent in marine humidity data prior to 1982, likely owing to a known change in reporting practice for dewpoint temperature at this time. Consequently, trends in both specific and relative humidity are likely underestimated over the oceans.
So it’s best to exclude the early data in the comparisons that follow. And the year 1982 makes it convenient to compare the specific humidity data to Reynolds OI.v2 sea surface temperature data , which starts in November 1981.
COMPARISON TO GLOBAL SEA SURFACE TEMPERATURES
Figure 2 compares the global HadCRUH specific humidity anomalies to Reynolds OI.v2 sea surface temperature anomalies. It should really come as no surprise that specific humidity anomalies mimic the warming trend and yearly variations in global sea surface temperature anomalies—the oceans and seas cover about 70% of the surface of our planet and most of the moisture in the atmosphere comes from the oceans. Note that I did not have to scale the sea surface temperature data, and that the trends are remarkably similar.
We know that after 2003 global sea surface temperatures have cooled. Has global specific humidity also dropped?
Based on the graph of global specific humidity over the oceans, Figure 4, from the June 2012 NOAA Climate Watch Magazine article State of the Climate: 2011 Humidity, it appears global specific humidity has declined over the last decade.
EAST PACIFIC REGIONAL SPECIFIC HUMIDITY VERSUS THE REST OF THE WORLD
For the last couple of years, I’ve broken the global oceans down into subsets to illustrate the impacts of El Niño and La Niña events on satellite-era sea surface temperatures. First, the sea surface temperature anomalies of the East Pacific have not warmed in 31 years. Second, we’ve also illustrated and discussed how the sea surface temperatures of the Atlantic, Indian and West Pacific Oceans warm in steps, how the steps are caused by the release of naturally created warm water from beneath the surface of the tropical Pacific during strong El Niño events, and how the warm water that’s left over from those El Niños prevents the sea surface temperatures in that region from cooling proportionally during the trailing La Niñas. And we’ve discussed how the ocean heat content of the tropical Pacific confirms that El Niño events were fueled naturally. If this discussion is new to you, refer to illustrated essay “The Manmade Global Warming Challenge” [42MB].
So it probably won’t come as a surprise to you to learn that the regional land+ocean specific humidity for the coordinates of the East Pacific (90S-90N, 180-80W) mimics the sea surface temperature anomalies of the East Pacific Ocean from pole to pole—same coordinates. See Figure 5.
And it also won’t be surprising that the regional land+ocean specific humidity anomalies for the Atlantic, Indian and West Pacific (90s-90N, 80W-180) also rise in ENSO-induced steps in responses to the naturally created warm water that was released from below the surface of the tropical Pacific during the 1986/87/88 and 1997/98 El Niño events.
The upward shifts in the regional land+ocean specific humidity anomalies for the Atlantic, Indian and West Pacific (90s-90N, 80W-180) stand out quite plainly on their own, Figure 7.
Hopefully in the future I’ll take the time to create an animation of global specific humidity maps, so that we can watch the response of the specific humidity anomalies to ENSO. Figure 8 is a global map of specific humidity anomalies for the period of July 1997 to June 1998. It captures the peak of the 1997/98 El Niño. And it also gives us an idea of the areas without data.
With the decrease in specific humidity over the past decade, it’s kind of odd that alarmists keep telling us that precipitation (rainfall and snow) is increasing and hurricanes are getting stronger because of global warming. Do they bother to check data? Obviously not.
Attribution papers, like Willett et al (2007) Attribution of Observed Surface Humidity Changes to Human Influence, rely on a climate models that cannot simulate the natural processes (ENSO) that have caused the oceans to warm and, in turn, have caused the variations and long-term rise in specific humidity.
How poorly the models simulate sea surface temperature was presented in the following model-data comparison posts:
And for the older CMIP3 version models:
The following is an animation of global specific humidity anomaly maps during the 1997/98 El Niño and 1998-01 La Niña. Each map presents the average anomalies over a 12-month period. It starts with the period of January 1996 to December 1996, and is followed by the map for the period of February 1996 to January 1997, and so on, progressing in 1-month increments, through the map for the period of January to December 2001. The 12-month averages are used to reduce any seasonal component and weather noise. The animation is 4MB. You may need to click it to start it.