The only thing the satellites are not detecting is global warming.

Temperature Trend by Latitude

In the global warming debate there is the idea that for the first time ever the entire Earth is showing warming.  In the past there were aberrations where a single part of the Earth would warm up.  An hotly debated example of that is the Medieval Warming Period (MWP) where the idea is presented that the warming 1,000 years ago was only around Greenland and Norway while the rest of the Earth was cool.  This leads to the idea that what is happening now is unprecedented.
I decided to take a closer look at the modern temperature data for the Earth by different region.  I chose the UAH data for this because it has the most data by specific latitude.  Using three specific zones it is possible to break the Earth down into three specific regions that are nearly identical in coverage.

Northern (NoExt):  85N to 20N,  166.8 million km² or 32.7% of the Earth’s surface.

Tropics (Trpcs):  20S to 20N, 174.5 million km² or 34.2% of the Earth’s surface.

Southern (SoExt):  85S to 20S,  166.8 million km² or 32.7% of the Earth’s surface.

So between these three zones 99.6% of the Earth’s lower atmosphere is measured.  This spatial coverage of the Earth is the best available.  The resolution for the satellite data is 4 km².  The end result is that the 508.1 million km² are equivalently measured by 127 million thermometers evenly spaced 2km apart on a grid that evenly covers 99.6% of the Earth.

Tropics:

The tropics are the region that show the least historical climate variability.  During the last glacial period the tropics were still tropical.  They were warm and the animals living in the rainforests were capable of surviving.  It is also the part of the Earth that is the most influenced by the ENSO cycle.  El Nino and La Nina are easily detectable in the temperature data from this region.

Temperature Anomaly for 20S to 20N from Jan 1979 to May 2011.

Linear regressions can be done of this which would create a trendline that has no meaning whatsoever simply because this is not linear data.  This is clearly cyclical data and the cause of that cycle is known to be the ENSO.  To be thorough I will show why.
The trendline for the tropics is 0.07 +/- 0.03 °C/decade.  The R² for the regression is 0.05.  This means that the regression is a very poor model when used to represent the temperature data for the tropics.  In fact, that is true for all regions.  Smoothing the data (using yearly data instead of monthly data) makes little difference.  The data is not linear so treating it that way will ALWAYS be misleading.
A reasonable assessment is there is a lot of variation in the tropical region, but there is no significant trend.  There is certainly no statistically significant temperature trend in the Tropics.

Southern:

The Southern region covers the Southern Hemisphere from 20S to nearly the South Pole.  It covers an identical amount of area that the Northern will.  The 32.7% of the Earth’s surface that it covers is only slightly smaller than the Tropics region.

Temperature Anomaly for 20S to 85S from Jan 1979 to May 2011.

Once again a trend could be shown, but the R² for the regression is 0.10 meaning that only 10% of the data really fits the linear regression model.  This is why scientists prefer the R² value to by much closer to 1 before claiming significance.

The trendline is interestingly enough almost identical to the Tropics though at 0.07 +/- 0.02 °C/decade.  Note that this trendline by region is less than half the values that are being used by warmists.  They typically claim values that are as high as 0.22 °C/decade.

When the Earth is viewed in thirds, with 2 of the 3 looked at, the results are very different.  For 66.9% of the Earth the rate of warming is nothing like the value claimed by the warmists.

Northern:

This leaves us with the region from above 20N to 85N which is the last third of the Earth.

Temperature Anomaly for 20N to 85N from Jan 1979 to May 2011.

Now here is one that shows a difference over the past 30 years.  The linear regression for this looks much better as well.  The trend works out to 0.27 +/- 0.03 °C/decade and the R² is 0.49.  Still not great, but not too bad either.

The problem is that the data still isn’t linear.  From 1998 to 2010 the trend is no different than the Southern or Tropical regions.  What really appears to have happened is a step function change that took place in 1998 where there are two distinct behaviors before and after 1998.  The trend over the past 12 years is 0.06 °C/decade with an R² of 0.01.  So global warming fully appears to have been a one time event that took place in 1998 and only happened in the Northern region of the Earth.

These results are very bad for the global warming crowd.  One of the key claims is that the whole Earth is warming up and that in the past there was only regional variations in climate.  That is a key attribute to dismissing things like the Little Ice Age and the Medieval Warming Period.

The modern data is showing that the warming is regional and not steady and certainly not increasing.  1998 really seems like a significant transition year as well.  The enormously powerful El Nino that year drove the Tropical temperature anomaly up over 1 °C for 4 months.  That is by far the largest anomaly for large regions of the Earth in the data.  That corresponds to the change in temperature in the Northern region.

Keeping that 1998 transition in mind I did the trend for the period up to 1998 and the trend after 1998 by each region.

Trend by region treating 1998 as a dividing line.

The Southern region was showing a cooling trend that matches the magnitude of all the warming trends with the exception of the Northern region in the period from 1979-1998.  That is the only warming that appears to be larger than the baseline variation in warming and cooling trends.

If global warming is real and is happening on the global scale.  Why does the satellite data fail to detect it?  The spatial coverage of the satellite data is unmatched and measures far more of the overall atmosphere than any other method.  It isn’t seeing anything of significance that is taking place.  The satellite data certainly detects each El Nino  and La Nina event.  It also detected the transition of the Northern region to a warmer temperature.

The only thing the satellites are not detecting is global warming.