Saturday, July 23, 2011

New paper shows some satellite measurements of temperature could be off by more than 5°C (9°F)

According to a paper published online today in the Journal of Atmospheric and Oceanic Technology, measurements of Earth's surface and lower atmospheric temperatures from long-lived satellites could be biased by "more than" 5°C or 9°F. Could be a problem considering the total global warming over the entire satellite era is only 0.46°C [less by a factor of 11 or more].

Climatological Diurnal Cycles in Clear-sky Brightness Temperatures from the High-resolution Infrared Radiation Sounder (HIRS)

Anders V. Lindfors*
The University of Edinburgh, United Kingdom and Finnish Meteorological Institute, Finland
Ian A. MacKenzie and Simon F. B. Tett
The University of Edinburgh, United Kingdom
Lei Shi
National Oceanic and Atmospheric Administration, USA

We developed a climatology of the diurnal cycles of HIRS clear-sky brightness temperatures based on measurements over the period 2002–2007. This was done by fitting a Fourier series to monthly gridded brightness temperatures of HIRS channel 1–12. The results show a strong land-sea contrast with stronger diurnal cycles over land, and extending from the surface up to HIRS channel 6 or 5, with regional maxima over the subtropics. Over seas, the diurnal cycles are generally small, and therefore challenging to detect. A Monte Carlo uncertainty analysis showed that more robust results are reached by aggregating the data zonally before applying the fit. The zonal fits indicate that small diurnal cycles do exist over sea. The results imply, that for a long-lived satellite such as NOAA-14, drift in the overpass time can cause a diurnal sampling bias of more than 5 K for channel 8 (surface and lower troposphere).


  1. Just to be sure, this means that the satellites could be overestimating temperature by 5 degrees, or that they are underestimating and HadCRUT is right?

  2. Could be either way based on the abstract alone, but point of the post is that satellite data is considered by many to be definitive, yet as this paper is just now pointing out, may have very large uncertainties that have not been adequately considered.

  3. The paper is behind a pay wall. Without access to the full paper I have a hard time accepting that any drift in data could be that wide. Is it wise to post such a claim without giving readers the means to verify it?

  4. I'm just repeating the concluding sentence of the abstract.