How Underground Temperatures Control Climate
The above plot (Trenberth and Fasullo) shows a curved trend line for sea surface temperatures which is now starting to decline.  More recent NASA data shows that the decline is continuing and the last 12 months have in fact been, on average, cooler than 2003.  In this paper Knox and Douglass also note the flattening of the trend around 2001 to 2002.  There has now been quite significant cooling towards the end of 2011.
It is appropriate to consider sea surface temperatures because the oceans contain close to 90% of all heat above the crust and it is the surface temperatures which have the most influence, not only above the oceans (which make up about 70% of the Earth's surface area) but also on islands and coastal areas where large portions of the population live.
But, even compared with the oceans, the amount of heat in the Earth itself is several orders of magnitude greater because the mass is far larger and the temperatures get up to about 5,400 degrees C in the liquid core.  This heat below the surface has a huge stabilising effect on climate as I shall explain below.
There is a temperature drop (or gradient) as heat flows from the liquid core out through the mantle and crust to the surface and thence through the atmosphere and out to space.  Above the surface there is of course far more heat contributed by the sun on a temporary basis each day, but the temperature in the first millimetre of so of the air is usually very close to that of the solid surface.  So there is a continuous temperature plot from the hot core to the top of the atmosphere (which is at about -270 deg.C) with a change in gradient at the surface.  (Strictly speaking there is a temporary small rise in the stratosphere, but that is not important in this discussion.)   So the stable base temperature at the surface is locked in by the temperatures at each end in the core and in space at the top of the atmosphere. 
Under the surface the process of heat transfer is mostly by conduction which involves collisions between electrons.  Then, between the surface and the air, some heat is also transferred by collisions when air molecules hit the surface.  This is a form of conduction which can also be called diffusion as explained in the second paragraph here.  Once that heat is in the air it then starts to rise by convection which involves physical movement of the air molecules.  In addition, some molecules (mostly so-called greenhouse gases) are also able to emit heat in the form of infra-red (IR) radiation, and the surface itself also reflects some high-energy radiation direct to space and emits some low-energy IR radiation.  Some of the IR radiation is captured by greenhouse gas molecules and re-emitted in all directions.  So some heat energy returns to the surface and exits again by conduction, convection and radiation.  This process probably extends the warmth of the day by a few minutes but there is no evidence that there must be a long-term warming effect - as explained below.
Over the life of the Earth we thus have a natural cooling process with associated temperature gradients from the hot liquid core to the surface and then from the surface to the top of the atmosphere (TOA) where the temperature is almost down to absolute zero.  Obviously humans have no control over the heat generated inside the Earth, nor the temperature in space.  But it is important to note that any natural variation in the heat generated under the surface will in time have an effect on the equilibrium temperature at the surface.  We should get early warning from changes in borehole data.
Now, the Sun will of course also warm the surface during the day, and it will cool off again at night.  We need to think of the underlying temperature gradient as providing a "stable base temperature" which would be about what we would expect perhaps an hour before sunrise on a calm mid-winter morning.  The Sun's heat flows some short distance into the surface and then back out again.  It is possible for some to be retained in local summer, especially in the oceans, but this tends to escape by winter.  There is a more detailed explanation on the third page.
Radiation from a cool body (including the atmosphere) does not add thermal energy (additional "warmth") to a body which is hotter than itself.  This includes the Earth's surface.  Hence the equilibrium which we observe in similar temperatures between the surface and the lowest few millimetres of the air must have been achieved by the diffusion process mentioned above.  This is facilitated by the flow of heat from the Sun into and out of the surface each day.  Any measurement of upwelling radiation will not distinguish between that from the surface itself and that from the air molecules.  The proportion of diffused heat which is then carried up by convection thus appears to be significantly understated in the models used by the Intergovernmental Panel on Climate Change (IPCC) and others.  The dominance of diffusion followed by convection (compared with radiation) was demonstrated in this experiment and can also be seen in a simple lamp cover experiment on my other site.
This in itself casts serious doubts upon the accuracy of such models which focus on radiation.  But the models are completely wrong because they are based on a false assumption, namely that radiation from a cooler atmosphere is capable of warming a surface which is already warmer.  This assumption has been proven wrong in this peer-reviewed paper published late 2011.  Even if a correction were made, measurements show that the upwelling radiation usually varies only between 99.5% and 100.5% of the downwelling radiation at TOA.  But the uncertainties  in the models when they calculate the above two totals are of the order of 1% to 2%, and so they cannot say for certain if the net figure obtained by subtraction (being of the order of 0.5%) is actually positive or negative.  This makes all the difference between warming and cooling.
Note also that the models do not consider the possibility of the underlying temperature trends varying by natural means.  Instead, because they focus only on flows of heat energy, they dismiss the net heat flow from underground as being insignificant.  Indeed the net flow is very small and only about 0.2% of that of total incoming solar radiation, but this is not the point.  Firstly, it is greatly supplemented by inward and outward temporary heat from the Sun each day, this heat increasing the amount of heat carried by convection in the atmosphere, and secondly, it ignores the "supporting" effect of the temperature under the surface which has built up over the life of the Earth, albeit from this very small heat flow.  The third page here provides more detail.
The underground situation is in fact somewhat more complex.  Probably only about 10% to 30% of that small heat flow originates in the liquid core, the rest in the mantle and crust.  A very small amount (about 1%) is generated in the very outer crust by "tidal" friction caused by the moon's gravity.  The effect of the moon's gravity is altered to a small extent by that from Jupiter, Saturn and a bit by Venus.  Over the course of many revolutions of the Earth this could have a noticeable effect on that component of the surface temperature which has resulted in a build up of heat from tidal friction.  This could explain long term cooling and warming, such as that seen since the 17th century.  The warming effect of Jupiter has been increasing in that period, but will turn to cooling in another 50 years or so.  Saturn is already starting to cause a smaller cooling effect which could last until about 2028.  It would be difficult to quantify the possible effect of planetary orbits, but just as difficult to disprove the possibility of any effect.  This plot calculated from planetary orbits has an uncanny correlation with past and present temperature trends.
Some other considerations are listed here ...
(a) Because nitrogen and oxygen do not radiate much heat themselves at atmospheric temperatures, then the heat they do acquire by conduction from the surface must be transferred also by molecular collision to greenhouse gas molecules. These GHG molecules can emit IR radiation and so, indirectly, they cool 98% of all atmospheric molecules.
(b) Including data up to October 2011, there is now statistically significant data confirming the existence of 60 year climate cycles the nodes of which correlate with Jupiter / Saturn resonance. In order to prove this I have had to take into account the regularity of intra-cyclic patterns which are apparent and significant
(c)  Analysis of borehole data from around the world demonstrates a strong correlation between surface temperatures and the projection of the underground temperature plot from depths at which the temperatures are higher than maximum surface temperatures.  It is common to see a trough in the temperature plot about 30 to 80 metres under the surface.  This tends to suggest that surface heat is only flowing in to this depth at which point it meets the downward trend from deeper depths.  Theoretical heat flow calculations are complex because of timing and the heat being generated en route, but this could confirm that sub-surface temperatures are forcing climate.
Third page:  More detailed explanation
Further information is on my site
The world is heading for cooling ....
Copyright: Douglas J Cotton, 10 September 2011