- Joined
- Jul 19, 2012
- Messages
- 14,185
- Reaction score
- 8,768
- Location
- Houston
- Gender
- Male
- Political Leaning
- Libertarian
Previously I've talked about the impossibility of distinguishing the current trend in the global temperature data set from zero. In other words, all the variance we've seen since 1850 could be due to random variation.
If it's just random variation then we would expect the temperature to drift in different directions from time to time; i.e., to go up for a while, then down for a while, to stay the same, etc. Looking at the HADcrut global temperature record from 1850 to the present it appears to be doing just that.
There's a related argument having to do with the uncertainty of the data. People in the Hadley center did estimates of uncertainty from all sources of error in 2005 and concluded that depending on what part of the earth is considered the uncertainty (95% confidence limits) is between 1 and 5 degrees C. What this means is that, even using a simple regression rather than a time series analysis, it's not possible to detect a significant signal from global warming by greenhouse gasses until the temperature is at least 1 degree above the value in 1968, and it still is not that high. Adding in the effects of autocorrelation, we won't be able to detect a definite upward trend in the temperature record until it exceeds 2 or 3 degrees. Again, if there's no upward trend then there's no point in speculating about what would cause an upward trend.
Much the same problem bedeviled the first working group of the IPCC and in the meeting of 1990 they were again prepared to announce that they were not able to detect a signal due to greenhouse gas warming in the climate record and didn't know when that would be possible. At that point the frustrated IPCC bureaucrats took the report away from the scientists and re-wrote it to say that a greenhouse gas signal had been detected.
If it's just random variation then we would expect the temperature to drift in different directions from time to time; i.e., to go up for a while, then down for a while, to stay the same, etc. Looking at the HADcrut global temperature record from 1850 to the present it appears to be doing just that.
There's a related argument having to do with the uncertainty of the data. People in the Hadley center did estimates of uncertainty from all sources of error in 2005 and concluded that depending on what part of the earth is considered the uncertainty (95% confidence limits) is between 1 and 5 degrees C. What this means is that, even using a simple regression rather than a time series analysis, it's not possible to detect a significant signal from global warming by greenhouse gasses until the temperature is at least 1 degree above the value in 1968, and it still is not that high. Adding in the effects of autocorrelation, we won't be able to detect a definite upward trend in the temperature record until it exceeds 2 or 3 degrees. Again, if there's no upward trend then there's no point in speculating about what would cause an upward trend.
Much the same problem bedeviled the first working group of the IPCC and in the meeting of 1990 they were again prepared to announce that they were not able to detect a signal due to greenhouse gas warming in the climate record and didn't know when that would be possible. At that point the frustrated IPCC bureaucrats took the report away from the scientists and re-wrote it to say that a greenhouse gas signal had been detected.