Archive for July, 2008

Humidity Data

July 10, 2008

Wouldn’t it be nice if we could falsify the entire Enhanced Greenhouse Effect hypothesis with just one chart?  It would describe specific humidity trends for the entire troposphere on Earth. 

Anthony Watts thought that he had exactly that (or rather close to that) with this post.  After he made that post, an interesting, but not unusual, phenomenon ocurred in the blogosphere.  There are the big blogs: Climate Audit and Watts Up With That.  And then there are dozens of smaller blogs, feeding off of those blogs.  The possibility of something so simple that could completely overturn the warmers’ hypothesis seemed surprising, and of course, the smaller blogs audited the big blogs, finding that once again things are not as simple as they seem.  Watts’ chart read “up to 300 mb” (mb being a measure of atmospheric presure).  But this wasn’t what the graph represented; what the graph actually represented was specific humidity at 300 mb.  The phrase “up to 300 mb” meant that the data was only available up to a pressure of 300 mb.

Now for some background about why humidity matters.  This post’s purpose is to discuss climate data, so I’m not going to source any of the background information; it’s common knowledge anyway.  The big question concerning how sensetive surface temperatures are to changes in CO2 is often expressed as how much temperature increase we would expect from a double of CO2 levels from pre-industrial times (280 ppm).  Currently, we’re at 380 ppm, and quite far away from 560 ppm.  Assuming that the greenhouse effect hasn’t already been saturated, we would expect 1.2 degrees C of warming from a doubling of CO2 without any feedbacks.  The catastrophic climate models base their prediction off of a large water vapor feedback, which would increase climate sensitivity by providing positive feedback to a forcing.  The extent and sign of the water vapor feedback are not agreed upon. The basic premise is that specific humidity (ratio of water vapor to air in the atmosphere) must remain the same, thus forcing increased water vapor content, or relative humidity.  This increase in relative humidity would then cause a stronger greenhouse effect.

As I did in my last post, I’m skipping over the theoretical and going to right to the observations.  By comparing observations with theory, I can decide if the theory is failing without necessarily understanding the underlying physical problems that are causing it to fail.

A few days later, after everyone had enough time to fool around with the website that Watts extracted the data from, he made this post, which quotes an article from JunkScience.  I’m not satisfied though.

Here are some of the first thoughts that came to mind:

1) Is this reliable?  Is specific humidity data reliable back to 1950?  Are there other sources of data describing specific humidity?

2) Where is the water vapor?  At what altitude and pressure? 

3) Within the troposphere, are there some altitudes that are responsible for most of the greenhouse effect?

To answer #1, I did some research and found a great summary of water vapor in the atmosphere and current measuring techniques.  It’s old, but it still is very helpful. 

This paper, entitled “Trends and variability in column-integrated atmospheric water vapor, doesn’t like the data Watts used so much. 

From the Abstract: “Only the special sensor microwave imager (SSM/I) dataset from remote sensing systems (RSS) has credible means, variability and trends for the oceans, but it is available only for the post-1988 period. Major problems are found in the means, variability and trends from 1988 to 2001 for both reanalyses from National Centers for Environmental Prediction (NCEP) and the ERA-40 reanalysis over the oceans, and for the NASA water vapor project (NVAP) dataset more generally.”



So where do we go from here?  Look for more data sources of course!  The paper I quoted seemed to look favorably upon Remote Sensing Systems (RSS)’s data set: SSM/I.  That’s for another day.

To answer #2; greenhouse-relevant watervapor is located in the troposphere.  The troposphere, though, varies is height according to lattitude, as shown in the graph below.

Therefore, if we truly want to find specific humidity trends, we need to look at 0 to 10 km from -90 to -50 degrees, 0 to13 km from -50 to -30 degrees, 0 to 16 km from -30 to 30 degrees, and 0 to 10 km from 30 to 90 degrees.  When this gets translated into atmospheric pressure, it becomes clear that we need data that encorporates even the 100 mb region. 

The NCEP data that we are using only works to the 300 mb region.

And to answer #3, at this point I am not very sure.  This is an issue for the future, once we can find a reliable dataset that provides reliable results, regional data, and data extending 16 km into the atmosphere.

And just to throw in one more conflicting graphic, heres water vapor content from Roger Pielke, Sr.

The picture caption reads: Figure caption: (a) The North American Regional Reanalysis domain-averaged de-seasoned precipitable water vapor – PWAV (blue), total precipitable water content -PWAT (brown), and lower-tropospheric temperature (red), monthly anomaly time series; (b) The North American domain-averaged PWAV, PWAT, and Tcol yearly time series by season. The dashed lines represent a linear fit, and the magnitudes of the trends are also shown. The black dashed lines indicate PWAV trends for a fixed relative humidity scenario. Note the much lower (and statistically insignificant trend in PWAV and PWAT despite a significant trend in Tcol [ from Wang, J.-W., K. Wang, R.A. Pielke, J.C. Lin, and T. Matsui, 2007: Does an atmospheric warming trend lead to a moistening trend over North America? Geophys. Res. Letts., submitted].

There are still more studies out there that support the IPCC prediction of a constant specific humidity, but that’s for another post.

In conclusion, water vapor content and humidity data are obviously very messy.  For now, that’s about all that I can say.

Comparing NOAA SST Data to UAH and RSS

July 9, 2008

As Bob Tisdale pointed out . . . FLAWED ANALYSIS

Continuing with my data collection project, I decided that I should analyze what satellite measurements are most reliable.  In 1979, we entered what is called the satellite era, and from that year to the present, we have had two Microwave Sounding Unit (MSU) temperature datasets.  One is run at the University of Alabama at Huntsville by John Christy and Roy Spencer (hereafter referred to as UAH).  The other is run by Remote Sensing Systems (hereafter refered to as RSS) , a private corporation based in Santa Rosa, California.

Both measure temperatures in the lower troposphere, middle troposphere, and lower stratosphere.  They are in relative agreement except concerning the temperature changes in the tropics.  The difference in temperature trends in the tropics is due to different mathematical techniques for analyzing the data received by the MSUs.  Because I have no ability to assess the validity of the two techniques, I am turning instead to sea surface data as a method for confirming UAH or RSS trends.

To do this, I used data extracted from NOAA because it offers a unique and incredibly useful way to extract sea surface temperature trends within your chosen latitudes and longitudes (I attribute this find to Climate Observations (  There are instructions for extracting the data, though it is somewhat outdated, so I will write the instructions here.

1)     Go to

2)     Click “Access.”

3)     Search and click “Reynolds SST.”

4)     In the row that reads, “Smith-Reynolds Extended Reconstructed SST’s,” click “plot.”

5)     Select your desired data range and de-select “Generate control files….”  Click “Build order.”

6)     Click “Plot data.”

7)     The third row down should read “All” under the “cycle” column.  Select that row.  Under “Select Output Mode,” select “Time Series.”

8 )     Chose an “Extra Variable Operation;” I recommend “12 point.”

9)     The rest is self explanatory.  If you’re looking for the data itself, you’re given the option under the graph.

So what exactly are considered “tropics?”  Technically, the region is in between about 23 degrees N and 23 degrees S.  RSS uses 20 N to 20 S for tropics measurements.  It is unclear where UAH defines tropics, but for this analysis, I will assume that it is the same as RSS.  UAH data is easily accessible and user-friendly.  It can be found here:  T2 represents middle troposphere; t2lt represents lower troposphere; t4 represents lower stratosphere.  Under each folder (t2, t2lt, & t4), scroll to the bottom to find a file named “uahncdc.”  RSS data is less helpful.  Here’s lower troposphere data:  There is however, this page, describing their findings:

My logic is as follows.

1)      If the difference between tropic trends of UAH and RSS is caused by a difference in data handling (and not data collection), then we should see the same proportional difference between land and ocean temperatures according to UAH and RSS.  For example, if tropical ocean temperatures rose 2 times more according to RSS as compared to UAH, then we could expect tropical land temperatures to also increase 2 times more according to RSS as compared to UAH.  Even if this assumption is wrong, my conclusion should not be significantly altered.

2)     It would be helpful to compare RSS and UAH tropics data to actual readings by surface stations and sea surface data.  Most of the tropics is ocean, and the only major pieces of land in the tropics are South America and Africa.  Unfortunately, we have seen a very large decline in African and South American surface stations to the point that I would no longer trust any attempt by GISS or Hadley to estimate changes in African temperature.  Therefore, I am looking to NOAA Reynolds Sea Surface Temperature (SST) over the tropics to compare to UAH ocean tropics data.  RSS unfortunately does not divide its tropics data into ocean and land, so I will not be able to include RSS in the comparison.

3)     According to my linear regression of RSS and UAH tropics data, RSS shows 2.43 times the warming UAH shows.  Therefore, based on 1) and based on the fact that most of the tropics is oceanic, we can expect that RSS shows 2.43 times the tropical oceanic warming that UAH shows.

4)     Using data from 1979 to 2007.  UAH tropical oceanic data shows a warming of .07 degrees C per decade.  Extrapolating based on 3), RSS tropical oceanic data would show a warming of .17 degrees C per decade.  NOAA Reynolds Sea Surface tropical oceanic data shows a warming of .08 degrees C per decade.

In conclusion, UAH data better fits actual surface measurements better than RSS does.  While this does not prove that UAH is more reliable, this analysis makes me confident enough to use UAH rather than RSS when determining terrestrial temperature trends.

Below is a graph I created from this GISS resource:  From 1979-Nov 1981, GISS uses Hadley data for SST; from Dec 1981 to the present, GISS uses the same data that I used.  Therefore, this graph should be a rather accurate map of the SST I used.  Maps always help; this one shows us cooling in the East tropical Pacific (due to the trend for increased La Ninas over the past three decades), warming in the West tropical Pacific, and warming in the Indian Ocean.


For comparison, I requested a received a rectangular map of global tempeature trends  (1979-2008 ) from RSS.  Although RSS’s website shows trend maps, they are not rectangular and thus are difficult to read in some areas.  Below is the RSS trend graph – note the tropical warming that does not appear in NASA GISS’s Reynolds-based graph.

Below is a graph of UAH and Reynolds oceanic tropical temperature trends (1979-2007).