May 2008 Update on Global Temperature, Part 1
Posted by The Diatribe Guy on May 14, 2008
GISS has released its data, found here. This post represents my initial data review. I will be following up with updated charts, and updated predicted anomalies.
The April temperature anomaly is 41. This is within the bounds of a reasonable value when compared to the RSS and UAH data, though a little on the higher side. Interestingly, the anomaly for March of 67 was reduced to 60, which is a pretty large adjustment. It would appear that late-reporting data came in significantly more on the cooler side than what was originally estimated. Even with the adjustment, it is a historically high deviation from the satellite data (based on a cursory analysis on my part. I hope to address this more fully at some point in the future).
Let’s take a look at the predicted anomaly based on the model I had put together. You can find that post for a complete discussion of the predicted values. To take a closer look at the methodology, go here.
To summarize: I predicted anomalies based on changes in trend lines and past weighting factors, etc. I tested 60, 120, 180, 240, 300, and 360 month trend line shifts and determined the predicted values based on anticipated future shifts in the trend line. I also presented the average, average excluding high and low, and the median.
The results are pretty good (for this month, anyway). The predicted values were, respectively, 39, 41, 51, 45, 44, 47, 45, 44, and 45. The actual anomaly, as I said, was 41.
Given the reduction in the March anomaly from 67 to 60, this would have lowered the projected number a bit, though I won’t go through the exercise of testing just how much. So, in this light, the anomaly of 41 is pretty darn close to the average measures, as well as a number of the individual estimates.
In the link I provided, I projected anomalies going out 12 months, but since I just started with this, I can’t get too excited about a single month’s results. I’d love to test it retrospectively, but based on the complexity of the model and the necessity to change the weights after each new data point, I’d need to work on this full time to properly do that, so I’ll just keep tabs on a go-forward basis. I will update my predictions each month, but I plan to track the historical predictions as well so I can assess the accuracy of the model for not just the closest month, but also subsequent months.
As for looking at the data and trying to pull out some of the interesting information, here is what I have so far. No fancy charts or graphs yet. I’ll get those up when I can:
After the March adjustment to an anomaly of 60, the 12-month average anomaly as of the end of March would be restated at 48.1, which is the same value as the year ending February (March 2007 anomaly was also 60). The latest 12-month average anomaly, as of year-end April is 46.2. This is the lowest 12-month anomaly average since the year ending November 2001.
The 2008 April anomaly is cooler than 2007 April by 0.23 degrees Celsius. It is the coolest April since 2001, which had an anomaly of 39. However, it is still the 11th warmest on record, according to GISS. So, there’s that continuing argument about whether or not this represents continued warming because of its relative historical magnitude, or whether or not this represents that we peaked a few years ago, and we’re on a downturn, but similar to walking down from the top of the hill, the first steps are still close to the top. I guess only time will tell on that whole question. Some may have heard that the recent data showed that April was one of the cooler Aprils on record. That data was U.S. only, and for the U.S., 2008 was the 29th coldest April in the 114 year record, and it was a full degree Fahrenhiet below the average of the entire 114 year period.
With the adjusted March anomaly, April was the 9th consecutive month where the anomaly is the same or lower than the anomaly 12 months ago. I’m not sure if that’s particularly meaningful or not, but since seasonality may play some part in the magnitudes, it’s an interesting point to note. The last time there was a run of 9 consecutive months like that it was the period ending July 2003. So, it’s not necessarily an unusual turn of events, absent other broader trends. The longest run of continually lower anomalies in the record was 22 consecutive months, from May 1945 – February 1947.
The current cooling trend can be “cherry-picked” to begin now as far back as March 2001. Thus, the trend now extends back 86 months. When I first started looking at this a few months ago, it was, I think, a 73 month period for which you could fit a negative trend line. So this continues to be extended under current conditions. Two months ago, a negative trend line could only be fit back to October 2001, so moving it back 7 months and adding two more data points has since extended the trend line by 9 months. The last time there was an 86-month period where the trend line was negative was the period ending April 1997. So, such a period is not unheard-of even in the midst of a warming trend. I am not arguing that a cooling trend isn’t real, just keeping the analysis real as it compares to historical data.
The most recent 60-month trend line is at -0.0596. This steepened from -0.027 last month. (Had the March anomaly been 60, it would have been around -0.0291). The current trend line has the steepest negative fit since the 60-month period ending December 2001. Historically, the steepest 60-month slope dates back to the period ending September 1904, with a value of -0.6564.
The 120-point slope is 0.1357, which is slightly below the value from last month, but slightly higher than the previous two months. Before that, the last time the slope was this low was the period ending April 1998. Historically, the steepest negative 120-month slope was -0.2488, which was the period ending March 1951.
The 180-point slope is currently 0.1970. This is the lowest trend line fit since the period ending April 2005.
The 240-point slope is 0.1621, which is not a trough. The slope values on this trend measure had trended down from a high value of 0.1756 for the period ending March 2004 and trended down to a value of 0.1544 for the period ending December 2006. It then bounced back up over then next few months, and has since fluctuated between 0.1594 and 0.1629.
The 300-point trend is at 0.1582. This trend line had increased to a level of 0.1634 with the period ending February 2007, and has since trended down slowly. These longer trend lines don’t have huge adjustments over short periods of time. Even if there were an extended cooling trend, it would take some time for the numbers to significantly reflect it.
The 360-point trend line has a value of 0.1411. While a trend down from recent values, it’s similar to the 300-month trend line in that there just isn’t a whole lot of movement.
The overall trend, 1880-current is 0.0468, indicating a per-Century warming trend of 0.562 degrees Celsius. This is basically unchanged since last month, which is to be expected. Any movements in this trend figure will be very incremental.
The actual last Century trend line is 0.05678, indicating an increase in average temperatures over the last 100 years of 0.681 degrees Celsius.