Making Data Dance

The Microwave Sounding Unit (MSU) data appears to be pretty reliable. While the fits are far from perfect, some indication of the CO2 forcing change, Solar forcing change and even changes in mean sea level can be teased out of the data. That should mean that even more can be coaxed out of all the noise.

That is a chart of the UAH southern extent, tropics and northern extent lower stratosphere data from 1994. I chose 1994 because there appears to be a general shift from cooling to neutral in the global stratosphere temperatures. The stratosphere should cool as the lower troposphere warms. In the lower troposphere, the 1998 super El Nino was a huge temperature event. It is not so huge in the troposphere. In the chart above, 2003 appears to be the largest stratospheric temperature event. That gave me an idea, possibly not the smartest idea I have ever had, but one that may be worth pursuing.

In this chart I plotted the mid-troposphere temperature divided by the lower stratosphere temperature for both the northern extent and the southern extent, again using the UAH data. By dividing by the lower stratosphere temperature, spikes should show when the monthly temperature approaches the mean anomaly. Depending on the sign of each, there would be positive ore negative spikes. It both data sets were fluctuation around their mean, the spikes would be evenly distributed, there would be more noise. In the chart above, there is much less noise than I expected. There was also more interesting noise than I expected. Since the mean for the data sets are based on the 1981 to 2010 period, most of the record, I would have expect no spikes at all prior to the 1994 to 95 leveling off of the stratospheric cooling with lot more noisy spikes after 1994 similar to the Sothern extent Orange curve. Since the Northern extent has more surface warming, I would have expected it to have more noise early in the series and less later, not fairly well distributed as it appears to be. That may mean that is could be possible to adjust the base line periods for the two data sets to establish similar noise patterns. That would give some indication of the relationship of "average" for each set with common events. With similar patterns, it could be possible to "slide" the relative timing of each series to isolate lag times.

Here, by adding 0.25C to the Northern extent lower stratosphere average, I was able to get the 1998 spike, the tropics 1998 spike is based on the 1978 to 1998 average of the tropical stratosphere and the Southern extent get a large negative spike with -0.02 used to adjust the average. I need a more logical method to adjust for optimum correlation, but there seems to be some potential here. I also am trying the differential of the lower troposphere and the lower stratosphere to compare changes in the emissivity of the atmosphere. Also a fairly crude method, but interesting.

This is the differential temperature for the entire UAH series.

And this is the temperature differential from 1995. The slopes of the regressions are much closer since 1995. With the short data series this will be a major challenge, but the average slope that best compares to Greenhouse gas forcing may be found by removing solar forcing change. It possible, then I may be able to estimate the land use impact. A lot of ifs and mays here, but the nearly double slope of the northern extent is fairly close to what I would expect for land use impact amplified by CO2 forcing. This is all probably a glorious waste of time. Still, there appears to be an outside chance of teasing out some useful information. This is more of a personal note than a real post.