[greenbean36191]

I don't have access to the Loehle paper since E&E isn't a very widely distributed "journal," but here's a fun critique of it. http://www.realclimate.org/index.php...constructions/
If the criticisms he points out are true then maybe I'm not missing out on much with E&E.

http://wattsupwiththat.wordpress.co...rature-part-34/

That really is a bad place to take the temperature for use in trend analysis and it's true that it's not unique. However it's not a surprise to climatologists. The heat island effect is already accounted for in the analysis. There are enough land based sites that aren't suffering from the heat island effect to say that the trend really does exist and provide us with reliable data. We also have a significant amount of corroborating data taken at sea along with sea surface temperatures. That same data is also what allows us to tell that the heat island effect isn't just theoretical.

On the proxy data, if I were to create a line representing the length of time the earth has been around, (or more relevant, the length of time the earth's environment is similar to today's), what percentage of that time would be covered by the proxy data? All of that time, only a fraction of that time?

The Earth is about 4.6 billion years old and in some sense we have proxies dating back to about 3-4 billion. As you go farther back in time though the number of proxies, the types of measurements you can get, and the resolution all decrease. For the modern era you could look at things such as direct observations, tree-rings, coral cores, sediment cores, and ice cores and the overlap and large number of data points provide a lot of clarity. If you want to look back a few thousand years though, you lose the tree rings. After about 100,000 years you start to lose the majority of the ice cores. After 800,000 years all of the ice cores end and beyond that eventually all you have is the geologic record, which doesn't provide a whole lot of precision.

If my low level understanding is at all close, then is there a generally accepted level of confidence in the proxy data and how is that confidence determined?

Where they overlap you can compare the proxies between each other and the observational data. The amount of variation between datasets allows you to statistically determine the margin of error with a defined confidence level. 95% is pretty much the the standard confidence interval, so that there's a 95% chance that the actual value falls within the margin of error. As the variability of the data decreases or the number of data points increases the uncertainty shrinks. For the most recent data we have lots of data points, so from a statistical point of view, low uncertainty and 95% confidence. As we go farther back we still have 95% confidence but uncertainty increases if that make sense. Basically after we calibrate the data using known values it comes down to statistics to tell us what we can make of it.

The book referred, and I haven't read it just yet, seems to talk about looking at 100,000 years, yet the total history of climate change is so very much longer. I know nothing about the other forms of proxy data gathering, but as a lay person, you may see my problem.

Alley was one of the scientists involved in drilling the Greenland ice cores, which only go back about 100,000 years, so that's what he talks about most. However other proxies including other ice cores go back much further and the book does talk a fair bit about those other data sources.

Towards the very old end of the spectrum you get more into areas where you can tell that it was hotter or warmer but you can't tell by how much. That's fairly immaterial for studying future trends though. We already have records for a long enough time period (including large changes) to verify the models.