1) The following is a graph showing relationship between energy released (millions of tons of TNT equivalent) and Richter Magnitude:
http://www.exploratorium.edu/faultline/earthquakescience/images/magnitude_lrg.gif
I was told that each unit increase of the Richter Magnitude is a 31-fold in energy released, but why in the above graph, it is not a 31-fold in energy released?

2) To determine the Richter Magnitude of an earthquake using a nomogram, you need to know the epicentral distance (from S-P lag time) and amplitude of the largest wave recorded by the seismograph. This looks sloppy to me, does it mean the amplitude of S wave or the amplitude of Surface Wave? The surface waves actually procduce the largest amplitude...

Does anyone have any idea and would like to share? :)

I was told that each unit increase of the Richter Magnitude is a 31-fold in energy released, but why in the above graph, it is not a 31-fold in energy released?
Isn't it? It doesn't seem precise enough to tell, but I wouldn't be concerned anyway - it's more illustration than data source, so it could easily be a bit off the mark in its figures.

Here's another image for comparison. I wouldn't rely on it any more than the other (it contains at least one mistake), but at least the numbers are clear.
http://volcano.und.nodak.edu/vwdocs/vwlessons/lessons/Rolling_earth/Picture10.jpg (VolcanoWorld (http://volcano.und.nodak.edu/vwdocs/vwlessons/lessons/Rolling_earth/Rolling_earth10.html))
To determine the Richter Magnitude of an earthquake using a nomogram, you need to know the epicentral distance (from S-P lag time) and amplitude of the largest wave recorded by the seismograph. This looks sloppy to me, does it mean the amplitude of S wave or the amplitude of Surface Wave? The surface waves actually procduce the largest amplitude...
Everything I know I learned from Wikipedia (http://en.wikipedia.org/wiki/Richter_scale) :)

Have a look in the "History" and "Problems" sections.

Isn't it? It doesn't seem precise enough to tell, but I wouldn't be concerned anyway - it's more illustration than data source, so it could easily be a bit off the mark in its figures.

Here's another image for comparison. I wouldn't rely on it any more than the other (it contains at least one mistake), but at least the numbers are clear.
http://volcano.und.nodak.edu/vwdocs/vwlessons/lessons/Rolling_earth/Picture10.jpg (VolcanoWorld (http://volcano.und.nodak.edu/vwdocs/vwlessons/lessons/Rolling_earth/Rolling_earth10.html))

Everything I know I learned from Wikipedia (http://en.wikipedia.org/wiki/Richter_scale) :)

Have a look in the "History" and "Problems" sections.
1) If my graph on the top is showing is 31-fold of energy released for each increase in the Richter magnitude, the points plotted on the graph would have to have below 1 million tons of TNT for all < magnitude 7 earthquakes! I think you are right, it looks better now for me to visualize the 31-fold relationship...
Roughly,
9.0 Magnitude => 199 million tons of TNT
8.0 Magnitude => 199/31=6.42 million tons of TNT
7.0 Magnitude => 6.42/31=0.20 million tons of TNT
6.0 Magnitude => 0.20/31=0.0067 million tons of TNT
etc
Now it matches more with my graph, however the points plotted for < magnitude 7 earthquakes are obviously too high, probably making it easier to see...

However, I have seen the same graph on a text book, showing that a 8.9 magnitude earthquake equals 125,000 tons of TNT, instead of 125,000,000 tons of TNT, everything regarding energy released is like divided by 1000 from this graph, why? Is there any difference when using the word "ton" in different countries?

2) I checked out your web site, but there isn't anything on whether the maximum S wave amplitude or the maximum surface wave amplitude should be used to determine the Richter magnitude...

If memory serves me correct, I think the maximum amplitude of the S-wave is in fact used to determine the magnitude of the earthquake. Or at least that's the way they used to do it (is there a newer method?).

a 8.9 magnitude earthquake equals 125,000 tons of TNT

This has definitely got to be wrong. There's absolutely no way that an 8.9 could be as weaksauce as 125 kilotons.

If memory serves me correct, I think the maximum amplitude of the S-wave is in fact used to determine the magnitude of the earthquake. Or at least that's the way they used to do it (is there a newer method?).
So even the Surface Wave part of the semiogram is shown, it is useless in finding the maximum amplitude and Richter Magnitude since only information from P and S waves are required?

I have found that all sources is showing that the maximum S wave amplitude is measured from equilibrium position to the top of the wave. This diagram, for example shows that. Also, my teacher when measuring the maximum S wave amplitude, only magnitudes the top ones but not the bottom ones. This lead me to the assumption that the bottom part of the seismogram is not used. However, I really doubt this idea because if there happens to be a wave that has an maximum amplitude of 61mm from equilibrium position to the bottom of the wave, but only 52mm from equilibrium position to the top, shouldn't we take the bottom (61mm) one as the maximum S wave amplitude?
http://www.sciencecourseware.com/VirtualEarthquake/Images/VQuakeMiscImages/SampleSeismogram.gif