• ADiff

    “[T]hey make no reference or calibration to any surface temperature” except with respect to (pre-flight) regressions to estimate non-linearity. But since this is accuracy enhancement rather than basic calibration, the statement’s fundamentally valid just the same.

  • After reading the article I’m not all that comfortable with this method of calibration.
    They point their sensor at a local (on the satellite) known temperature target for a warm measurement and then they point their sensor at deep space to get a cold measurement and those numbers are used to “calibrate” their sensor.

    I’m sorry but that’s just weak.

    All that tells me is that they can very good at measuring the temperature on a nearby warm object in the vacuum of space and the temperature of deep space at a temp of 2.7 Kelvin. It may be all they can do but it does not give me any confidence that they can the point their sensor at the earth and its atmosphere and have any confidence. They calibrate with mesurements taken in the vacuum of space and assume the same sensor response to an atmosphere.

    They are calibrating a device with data points that are completely different in location and environment than the actual location and environment they are trying to measure. In my book that is not calibration for mesuring atmospheric temperatures but a calibration for measuring temperatures in a space vacuum.

    I realize that may be the best they can do.

    All I can assume from this is that the satellite temps can be useful for relative changes over time but have an unknown accuracy of absolute temperature.

  • I think Jeff’s objections are a bit overblown. What they are calibrating is a simple device, a multi-channel microwave radiometer, and they are calibrating it strictly for drift. Their methods look pretty good to me.

    If you want to criticize their temp measurements, look at the assumptions that go into the conversion of radiometer counts to temperature. However, for measuring temperature change, I think those work well also. This is a much better way of measuring temperatures than tree rings or surface based thermometers.

  • klee12


    The question for me is: does surface termperatures from weather
    stations necessary for the calibration of satellites. If so, then
    the all the GHCN databases and its derivatives are not entirely independent. I have read Spencer’s article and its not clear to me. Can someone here more knowledgeable than I answer the question definitively?

    Thanks in advance


  • pft

    The more you think of it, clouds obscure over 50% of the surface and the time of measurement is not constant, so any satellite coverage is incomplete and measures only surfaces not subject to cloud cover (hotter in the day, cooler at night, perhaps offsetting, perhaps not, unless somehow measurements are made through clouds somehow) at a given point in time.

    The calibration method is of course imperfect (two points from 2.7 to 300 deg K for measuring 290 deg with an admitted measurement error of 1 deg C), not to mention algorithms and intercalibrations not discussed that may be prone to parochialism. I have noticed a convergence in recent years with satellite and surface temperature measurements and this is cause for concern.

    It seems to me they should calibrate via a tropical sea surface and Antarctic ice where temperatures could be known to a fair degree of certainty (narrowing the calibration range from 220 to 290). This would not be subject to the biases of surface measurements on land, or rely on the very cold vacuum of space for one of the 2 points on the calibration curve, although it could be used as a 3rd point.

    It may be that the data is fine today, but given where the funding comes from, the adjusted data is potentially corruptable, so it can not be trusted absolutely.

  • TanGeng

    Nice repost. Where is the oxygen that is being measure? I don’t see the discussion of how they know it’s a surface temperature monitoring.

  • JP

    UAH and RSS do a pretty good job. Yes, there times when the calibrations are off; but, it is pretty obvious when it does and Dr Spencer does a pretty good job in correcting it.

    Unlike surface based temp observations, the satellites measure anomalies where all of the action is occuring (between 8000-15000 MSL). Thier coverage of the tropical oceans is very valuable (despite what NOAA says).