In 2000, after the release of the NRC report global temperature change, the U.S. Climate Change Science Program set out to investigate the discrepancies between surface and atmospheric temperatures as observed by satellites and radiosondes (weather balloons). The Program’s preliminary investigation and plans for further investigation were set forth in the draft white paper in the second link. This report, released in May of 2005 presents the USCCSP’s final results in which they demonstrate the agreement between surface and atmospheric temperature records and suggest agendas for further research. The first link is to a page that has links to the full contents of each section, as well as one to the complete report (9.2 mb download). The second link is to an Executive Summary of the report.

The paper from Remote Sensing Systems (RSS) presenting Version 1.0 of their MSU derived upper-air analysis and resulting troposphere and stratosphere temperature trends. RSS analysis products use different methods for processing these records than the University of Alabama, Huntsville (UAH) products that were in conflict with climate model predictions at the time this paper was published. Specifically, RSS relied on the unprocessed MSU datasets which view bulk brightness temperatures for mainly in the middle and upper troposphere (or the free troposphere) while UAH relied on “synthetic channel” data that emphasized the lower troposphere by combining various side-looking views of the MSU detectors. RSS also use different methods for removal of noise from the data and combining the records of different satellites over time to produce a continuous trend. There are strengths and weaknesses to each approach and neither approach is clearly better than the other in all respects. The mainstream climate science community relies on both, but tends to favor this record over the UAH products because it agrees well with two more analyses that used analysis methods independent of its and UAH’s (Prabhakara et al, 2000; 1998 Geophys Res Lett), and because there is general agreement that RSS methods are less subject to noise of various kinds than their UAH counterparts. Climate skeptics have relied exclusively on the UAH products for their case because it yields the smallest warming trends of all available datasets.

These papers present the two most recent (as of Jan. 2007) UAH MSU analysis products. The first presents Version D (2000) and the second, Version 5.0 (2003). Version 5.2 (May 2005) was the first dataset in which the erroneous diurnal drift for the NOAA-11 satellite had been corrected (see Mears & Wentz, 2005, linked below). The final links are to the UAH public data access server where their datasets can be downloaded and used. The datasets labeled “5.2” at the last link are those with the correct diurnal drift adjustments.

The second of two MSU analysis products from a team lead by Prabhakara of NASA. This analysis is similar to the RSS products, except that their merge analysis assumed a constant temperature calibration (based on an onboard “hot target” at a known temperature). The UAH and RSS teams both derived corrections for each satellite in the series as part of their merge calculations. The trends in this paper agree well with the comparable RSS trends, but not with those of UAH Version D to which they are most comparable.

The paper by Vinnikov & Grody presenting their MSU derived upper-air retrieval and analysis, which was based on a unique statistical method to merge the different satellite records. Vinnikov and Grody’s method is very different than that of the other teams. It is based on certain assumptions about daily, annual, and interannual cycles in atmospheric temperature that, if valid, will yield the most accurate and noise-free results of any of the extant analysis products. However, if these assumptions differ significantly from the actual temperature cycles they simulate, considerable error might result. This product yields the highest warming trends of the currently published MSU analyses, and has received less attention than the UAH and RSS products.

This is the August 2005 paper first published online at Science Express (subscription required for full text) in which Carl Mears and Frank Wentz of RSS present their discovery that UAH MSU products had been applying corrections to the MSU data from the NOAA-11 satellite in the wrong direction, thereby introducing a spurious cooling into UAH analyses. When the error was corrected, beginning with UAH Version 5.1 (Christy et al, 2005) this record’s disagreement with extant climate model predictions of global warming all but vanished and with it the only potentially credible argument global warming skeptics ever had. The NOAA TIROS Series of satellites that carry the MSU packages are in polar orbits that are sun-synchronous (i.e.—in orbits that preserve their orbital plane with respect to the sun throughout the year so that they rise and set in the sky at the same time on any given day and location during their service lives). Imperfections in this sun/satellite synchronicity cause a drift in the satellite’s daily rising and setting times during its life that will appear to the sensor as a spurious warming or cooling depending on the direction (it’s usually much cooler at 6 AM than it is at noon!). This is referred to as diurnal drift. NOAA-11 had a comparatively large one—enough so that reversing the correction for it was enough to spuriously remove most of the warming that should have been present in UAH products.

This 2002 paper from the RSS team describes the issues involved in correcting remotely sensed temperature data from satellites for diurnal drift (variations in the sun-synchronicity of polar orbits) and the methods used to correct their MSU datasets. The UAH team relies on a different method based on comparisons of the MSU sensors’ side-looking views—that is, at angles to the left and right of nadir (straight down) along its flight path.

MSU sensors measure temperature by detecting the total amount of microwave radiation emitted by a layer of the atmosphere, the altitude and thickness of which depend on the frequency (wavelength) being measured. These measurements are taken on multiple channels, each of which looks at a different frequency and therefore a different “slice” of the atmosphere. The channel that is of most interest to global warming studies is Channel 2 which looks at a layer that covers most of the troposphere—where we expect the biggest global warming fingerprint. The stratosphere above it, which is measured by Channel 4, has long been known to be cooling (for other reasons). One problem that has beset these studies is that the atmospheric layer seen by Channel 2 dips a little ways into the lower stratosphere, making it a good, but not exact look a