Ice Cores

Ice cores are used to demonstrate on charts what the global mean temperature was thousands of years ago.  I read this wikipedia article on Ice Core and it was talking about how the cores only show local temperature.  Put this great bit of knowledge together with today's explanation of cooler temperatures being a side-effect of global warming because an increase in global temperatures will cause a greater disparity of local temperatures, up or down.  So, how can we trust the temperatures in just a few locations to show us to any degree of accuracy of what the temperature really is?

Measuring Global Mean Temperature

Ever since ClimateGate hit the news, I have been wondering about the collection of temperature measurements and how that figures into Global Mean Temperature.

I imagine it's all done by ground based stations and at any moment in time, all the stations' temperature readings are averaged together giving a global mean temperature.  That's probably a very simplistic view of the scenario.  Here's a few problems that I see:

1)  Adding more measurement stations means you are adding more points of data, and depending on where they were added (say warmer climates), it could skew the results.  How do you calculate for this?

2)  Heat islands.  Ever stand in the middle of an asphalt parking lot and feel yourself baking?  Ever stand just to the side of one and still feel the baking?  What happens if buildings or parking lots or roads are built close to these stations?  Is the data corrected for this change?  How do you correct for this?