Open Mind

AMSU mystery

November 2, 2008 · 11 Comments

In a previous post, someone linked to daily data from UAH (Univ. of Alabama at Huntsville) for AMSU (Advanced Microwave Sounding Unit) channels (if you click the “draw graph” button, you’ll find a “show data as text” link). Being a data junkie, of course I’ve taken a look at these numbers. I’m surprised by what I see.


There are a number of channels reported, including:

  • chLT: 900 mb
  • ch04: ??? (not given)
  • ch05: 600mb
  • ch06: 400mb

    All these datasets are global averages. I looked first at the chLT data. I’d guess that “LT” stands for “lower troposphere,” although the phrase “lower troposphere” more usually indicates from about 1000mb (very near the surface) to about 300mb. Maybe “LT” stands for “lowest troposphere.” Nonetheless, I was interested in what’s going on at 900mb, only about a km. above the surface. Here’s the data (temperature is in Kelvins, which is deg.C + 273.16):

    Here I’ve added a smoothed version:

    One thing is clear: these data indicate that the “lowest troposphere” (at the 900mb level) is getting warmer. Fast. The mystery is that it’s warming so fast. In spite of starting at Aug. 1998 and catching the tail end of the 1998 el Nino, these data show warming at 0.12 deg.C/yr. That’s 1.2 deg.C/decade, or 12 deg.C/century. And that’s a lot faster than I’d expect, in fact it’s about 7 times as fast as the estimated surface warming since 1975.

    I looked at other channels too. The next higher level (I presume) is ch04, and although a pressure level isn’t given I’d say that 750mb is a pretty good guess. Here’s the smoothed version:

    and this too indicates rapid warming, at 0.04 deg.C/yr. That’s more than double the estimated rate of surface warming. Looking at ch05 (at the 600 mb level) we see:

    This too is warming, at 0.024 deg.C/yr, considerably faster than surface warming. Only when we get to ch06 (at 400mb) do we see cooling in the atmosphere:

    The ch06 data indicate cooling at -0.048 deg.C/yr.

    Further channels are above the 300mb level, so I made a crude estimate of temperature for the atmosphere from 1000mb to 300mb by computing 0.25 T(900) + 0.21 T(750) + 0.25 T(600) + 0.29 T(400). That gives this:

    We still see warming, at a rate of 0.31 deg.C/yr. However, the monthly UAH version of TLT (lower-troposphere) data since August 1998 indicates warming at only 0.08 deg.C/yr.

    Of course there’s more to these data than meets the eye, and a lot more than I’m aware of. I’d guess that they don’t represent temperature in thin layers of the atmosphere, but in much broader layers; the ch06 data (400mb) may even show nontrivial impact from the stratosphere (above the 100mb level). But it’s still quite a surprise that the lower atmosphere, and especially the channel indicated as the 900mb level, is warming so much faster than the surface. The rates indicated by these data seem implausible to me.

    There might be a perfectly good, even simple, explanation. But I don’t know what it is.

  • Categories: Global Warming

    11 responses so far ↓

    • Phil. // November 2, 2008 at 6:33 pm

      See here or some explanation:
      http://www.remss.com/msu/msu_data_description.html

    • Gavin's Pussycat // November 2, 2008 at 7:05 pm

      Feeling cruel today… orbit drift?

    • Deepclimate // November 2, 2008 at 7:16 pm

      Tamino,
      As Phil implies, we probably all need to read a “primer” on the AMSU microwave units. The data archived appear to be brightness “temperatures” measured at different heights and frequencies (i.e. the various channels of the AMSU units on NOAA-15 satellite). It appears that they are combined with various “weightings” to produce the TLT and TMT data (at least in RSS). It also looks like further processing of these data to correct for various effects like orbital decay and diurnal drift. Certainly, they can not used directly to obtain the temperature anomalies (you can see this by comparing the latter part of 1998 with 2007 or 2008 in both the “LT” AMSU channel, and the UAH TLT monthly anomalies).

      Here ia another description of the AMSU-channels:
      http://pm-esip.msfc.nasa.gov/amsu/index.phtml?0

      I haven’t found a description of the UAH processing (corresponding to that one posted by Phil for RSS), but I imagine it’s out there somewhere.

    • Deepclimate // November 2, 2008 at 7:35 pm

      Here is an explanation of AMSU channels and the “brightness” temperatures measured.
      http://pm-esip.msfc.nasa.gov/amsu/index.phtml?0

      Presumably they are combined and weighted in different ways to achieve the various data products (TLT and TMT). Also, I believe that corrections for orbital decay are applied along the way. So the “raw” brightness temperatures can’t be used directly. (To see this, compare UAH TLT monthly anomalies and LT brightness “temperatures” for 1998 and 2007/2008).

      Here is an explanation of spurious warming, caused by diurnal drift, from UAH:
      http://vortex.nsstc.uah.edu/data/msu/t2lt/readme.03Jan2008

      Since presumably UAH and RSS have been using the same raw data you cite above, the key to the UAH annual cycle mystery presumably lies in the the details of the correction calculations . The differing channel weighting functions may also play a role, but my hunch is that this does not account for most of the UAH/RSS differences, including the UAH annual cycle.

    • gjp // November 2, 2008 at 9:07 pm

      I don’t know what chLT means but AMSU channel 1,2 and 3 are strongly influenced by skin temperature,while higher channel are influenced by lower stratosphere:
      http://www.ecmwf.int/newsevents/training/rcourse_notes/DATA_ASSIMILATION/INVERSION_METHODS/Inversion_methods75.gif
      Here is an article about land skin temperature retrieval from AMSU:
      http://ams.confex.com/ams/pdfpapers/99340.pdf

    • Deepclimate // November 2, 2008 at 10:39 pm

      gip, regarding channel LT:

      From IPCC AR4 WG1, section 3.4.1.2.2:

      “By differencing T2 measurements made at different slant angles, the UAH group produced an updated data record weighted for the lower and mid troposphere [dubbed T2LT by the IPCC].”

      In other wotds, chLT is not an actual AMSU channel, but a “synthetic” one. FWIW, the RSS LT method makes more sense to me and the resulting data do seem to fit better fit with surface data and models, especially in the tropical troposphere.

    • TCO // November 3, 2008 at 1:10 am

      It’s actually kind of annoying that the sattelite temps (both groups) have so much modeling in them and need to be compared to surface for accuracy rather than standing on own. I wish there were a more accurate sattelite (maybe one with power to stay on station?) Or maybe balloons would be a more accurate direct measurement?

    • Hank Roberts // November 3, 2008 at 3:00 am

      > a more accurate satellite (… to stay on station)

      Why there IS one. And we know exactly where it is. In a warehouse, delayed by the Administration who have refused to let the Europeans take it and launch it and use it. It would have resolved much of the uncertainty eight years ago.

      Cui bono?

      http://books.nap.edu/openbook.php?record_id=11187&page=116

      Lagrange Point Sensing: Triana, currently in storage at GSFC and renamed Deep Space Climatic Observatory (DISCOVR), will be the world’s first Earth-observing mission to L1, the neutral gravity point between Earth and the Sun. This unique observing position will provide a continuous view of the sunlit portion of Earth as the planet rotates “below” the satellite. Continuously, every 15 minutes, data will be collected on Earth’s radiation budget, cloud cover, aerosols, ozone, vegetation canopy, and the interplanetary medium. This low-cost mission will rely on heritage retrieval algorithms. …

      Er, that is, unless the Administration takes its last opportunity to destroy the satellite rather than leave it in the warehouse where it could finally be revived.

      Vicious, stupid bastards.

      OF course you won’t read this in the US press. The Canadians are covering it:

      http://www.desmogblog.com/dscovr-mission-to-be-gutted

      “DSCOVR Mission May Be Gutted
      29 Oct 08
      Here is the latest twist in the bizarre story about the Deep Space Climate Observatory (DSCOVR). Apparently, the US Air Force is in discussions with NASA to take over the mission, with one important catch: that all the Earth observing instruments be removed….”

      Greek tragedy has nothing on US politics.

    • dhogaza // November 3, 2008 at 4:31 am

      It’s actually kind of annoying that the sattelite temps (both groups) have so much modeling in them and need to be compared to surface for accuracy rather than standing on own. I wish there were a more accurate sattelite (maybe one with power to stay on station?) Or maybe balloons would be a more accurate direct measurement?

      Given that the satellites were never designed to give direct measurements, anything (including balloons) designed to give direct measurements might be more accurate.

      But, of course, direct measurements can be jiggled by various things, which Watts et al try to exploit by their totally bogus surface station photography project. Just because their premise is silly doesn’t mean that instruments doing direct measurements are, a priori, going to be more accurate than synthetic temp data fro satellites.

    • dhogaza // November 3, 2008 at 4:33 am

      It’s actually kind of annoying that the sattelite temps (both groups) have so much modeling in them

      Hmmm, to continue, you do realize that the satellites don’t actually measure temps?

      Temps are derived from modeling because the satellites don’t measure temperature, and weren’t designed to do so.

      Your “too bad (both groups)” comment seems to be a bit derogatory towards both groups, but given that temps aren’t actually being measured, there’s no alternative.

      It’s not the fault of the groups doing the analysis, neither of which had anything to do with the design of the satellites …

    • dko // November 3, 2008 at 5:28 am

      New to this site, so pardon if this has been done before.

      I took GISS, HadCRUT, and NCDC data from 1979 to present and adjusted each series so that Jan 1979 to Dec 1998 equals zero. I averaged these adjusted values for each month to get a “best guess” for surface temps.

      Next, I subtracted these values from the monthly RSS and UAH data and called the results (quite cleverly) dRSS and dUAH, which I plotted.

      A linear regression fit through dRSS and dUAH proved enlightening. The slope for dRSS (the departure between RSS and the surface values) was a mere +0.0005 C per decade, while dUAH was -0.0398 C per decade.

      This doesn’t prove that the UAH data are flawed, necessarily, but it does demonstrate that RSS more closely approximates actual surface temp readings.

    Leave a Comment