“X” Candidates and Questions

What would this time series look like if the very short term variability was also largely removed? To explore that, here are the 15-day moving averages of ADA (“adjusted daily averages”):

This smoothed anomaly data has the appearance of oscillatory behavior. Typically, about every 25 to 40 days, the troposphere (as represented by the channel 5 data) warms and then cools by 0.1 to 0.2 K.

Is this oscillatory appearance confined to 2005-2006? No. A similar plot covering eight years (2003-2010) is here . The 2005-2006 data is representative of the full period.

Regarding the oscillation period, a plot of the period (days between peaks) for 2003-2010 is here . Most of the days-between-peaks are between 20 and 45 days, a bit shorter than the 30 to 60 day oscillation period generally associated with the Madden-Julian Oscillation (MJO).

If the oscillation is associated with a real tropospheric oscillation or pattern, which one?

Among the candidates (thanks to WUWT for a list ) are the Madden Julian Oscillation (MJO), Arctic Oscillation, Antarctic Oscillation, North Atlanic Oscillation, Southern Oscillation, Pacific North American Index, East Pacific North Pacific Pattern, West Pacific Pattern, Pacific Transition Pattern, Tropical Northern Hemisphere Pattern, Polar Eurasian Pattern, Scandanavian Pattern, 27-Day UV Oscillation and Stratospheric Antarctic Intraseasonal Oscillation.

The physical nature of such an apparent temperature oscillation, whatever its name or association and assuming that it’s real, is intriguing.

*  Does the oscillation represent changes in heat input, such as changes in tropical thunderstorms moving energy from the surface layer into the mid troposphere?

*  Does it involve changes in the rate of energy loss to outer space, perhaps from changes in water vapor/cloud coverage?

*  Is it associated with changes in atmospheric angular momentum? Does it involve a negative feedback mechanism?

*  Is it a stratospheric phenomenom, not tropospheric, the result of contamination of channel 5 data by its stratospheric component?

*  If it is tropospheric, does the oscillation affect tropospheric water vapor content?

*  Is the oscillation a local intense event or is it diffused across the globe?

*  Does its amplitude have any relationship to the timing or severity of extreme events (an interest of mine)?

The conventional candidate is MJO but I’ve found little relationship between ADA and the traditional MJO indicators like tropical OLR and zonal wind variability.

Posted in Uncategorized | 2 Comments

AQUA Annual Cycle and ENSO

Below is a plot of AQUA channel 5 daily data for 2005 and 2006, as provided at the UAH Website. There is nothing special about 2005 and 2006 – they are selected simply for illustration. This time series shows two annual cycles, cycles which are generally attributed to the effects of land mass differences between the northern and southern hemispheres. See here for a further discussion. The data also shows jagged higher-frequency variability. It is this higher-frequency, intraseasonal (20 to 90 day) variability which is of interest here.

Removal of the annual cycle should allow a better visualization of the higher-frequency variability. The removal can be accomplished, approximately, by subtracting the “climatologically normal temperature” ( plot here ) for a date from each daily reading on that date. The “climatological normal temperature” for a date, as defined by me, is the average channel 5 value for that date over 2003-2010. That indeed is a short period for such an application but AQUA data is available only since late 2002. The result of the annual cycle removal is here . The derived values (actual minus “climatological”) will be called “daily anomalies” in this post.

The annual cycle is only part of the tropospheric temperature variation. There is also intermediate-term (interseasonal) variability in the channel 5 data, mainly attributable to ENSO effects. Subtracting a moving average (say, a 45-day, 60-day or 90-day average) from the data should remove much of the interseasonal trend while leaving shorter-term (intraseasonal) variability.

The result of subtracting the 45-day moving average is shown below. I term these derived values (daily anomaly minus 45-day moving average) as “adjusted daily anomalies” or “ADA” for short. The “adjustment” is simply the (approximate) removal of ENSO variability.

Posted in Uncategorized | 2 Comments

Image Storage 1

Extra Image 4


Family Tree 2 (1)

Mitchell Branch

Pot pie


Thweatt Line

Tallula Gallaway Littleton


Extra Image 3

Extra Image 2

Extra Image 1

Climatological Normal





Posted in Uncategorized | Leave a comment

Aqua Channel 5 and Short-Term Temperature Variability

Channel 5 data provides a tropospheric temperature estimate every 24 hours. A look at seasonal and interannual variation (see posts) found that the AQUA data, on longer time scales, behaves as-expected at least with respect to seasonality and ENSO. In other words, there is some mid-tropospheric physical realism to the data, so far.

Now it’s time to see how the data behaves over shorter terms (intraseasonal or shorter). There may be interesting things to see which are not visible in the usual monthly averages of global temperature data.

First, here is a time series from the early years of AQUA – 2003. This is a plot of the daily anomalies. (The daily anomalies are calculated by subtracting the “normal” temperature for a date from the 2003 value for that date. “Normal” is defined as the 2003-2010 average of temperatures for a date.)

Perhaps the most apparent visual impression is the “seesaw” or “sawtooth” appearance of the data. Rather than random movement around a central value there appear to be swings in temperature.

The plot contains small red letters “A” thru “G”. The letters designate the peaks of swings over the first six months of 2003. The time between peaks are spaced 22, 23, 26, 24, 23 and 19 days, respectively. The amplitudes vary but are generally in the 0.2 to 0.3 K range, which are rather substantial changes in global tropospheric temperature.

I find the similar spacing and amplitude of these swings to be interesting. What are the underlying physical mechanisms? For example,

*  Specifically, what places the additional heat into the troposphere (or slows the removal of heat) and what triggers that?

*  What is the nature of the decline – is a cessation of whatever caused the rise or is it some negative response whatever caused the rise?

*  Is the warming over a relatively small region, such as a portion of the tropics?

* Does the swing affect the water vapor content of the troposphere and thus its radiative properties? In particular, might the amplitude change with tropospheric temperature?

*  What controls the amplitude of the swings?

*  What controls the timing of the swings – why, in the first half of 2003, did the swings space themselves about 23 days apart?

*  Why are there relatively orderly swings in temperature in the first half of 2003 and not-so-much in the second half?

There is a second interesting aspect in this time series, with examples designated as “H” and “I”. Those are short periods of near-zero tropospheric temperature change.

*  In this 2003 time series with large daily and weekly swings, why are there also periods where the temperature change approaches zero for days? Are those periods simply statistical artifacts or is there some underlying physical cause?

OK. Here is another time series – 2005:

The year shows the saw-tooth appearance of the earlier example (2003) but the 2005 “teeth” aren’t nearly as orderly as was seen in the first half of 2003. The amplitudes of the swings are around 0.2 K to 0.4 K, about the same as in 2003. Of interest are several periods (designated by red letters) in which the global temperature change essentially stops.

*  Why do the 2005 changes appear less orderly than those of the first half of 2003?

Finally for this post, here is a two year period from mid-2009 to mid 2011:

Again, a sawtooth appearance is apparent. Also apparent is the temperature rise and decline associated with changes in ENSO. (The first part of the period was associated with an El Nino event which then shifts into a La Nina state.)

*  The second half of the red line shows a period when Pacific ENSO temperatures were declining rapidly, yet the global temperature remained more or less constant. Why didn’t the troposphere immediately respond to the decaying El Nino and what finally triggered the sharp temperature decline?

*  The green lines suggest a decline in the amplitude of the swings. If that appearance is real and not an artifact, what is the physical explanation of the decline in amplitude?

Summary: there are many interesting things (at least to me) to observe when the daily data is plotted. In future posts I will try to explore some of the questions raised. I am particularly interested in what underlies the rapid temperature changes, including whether these short-duration events are somehow important to the removal of heat from Earth.

Posted in Uncategorized | Leave a comment

AQUA Channel 5 Data and ENSO

Is there correlation between the AQUA channel 5 data and ENSO?

To answer that we must first do something about the annual cycle which is present in the channel 5 data. Global surface temperature cycles over a year, with the global peak during the boreal summer, apparently due to variation in the amount of landfalling insolation. Here’s a plot of the annual cycle:

This is created by averaging the 2003-2010 temperatures for each date of the calendar. For example, the average for January 22 is created by averaging the January 22 readings recorded for 2003, 2004, 2005, 2006, 2007, 2008, 2009 and 2010.

The difference between an actual daily value (say, for January 22, 2010) and the 2003-2010 average for January 22 is what I define as the “daily temperature anomaly”. By tataking these differences one can effectively remove the annual cycle.

Here is the time series of the daily temperature anomalies from January, 2003 through mid-2011:

Daily Temperature Anomaly

It’s an interesting pattern, with little net change from 2003 to late 2007 then several sizeable changes, up and down, thereafter.

How does that compare with ENSO, the best-known driver of global temperature anomalies on interannual and intraannual timescales? Here’s the same anomaly plot along with the “ONI” (Oceanic Nino Index). In this plot ONI leads the channel 5 data by three months:

ONI and the daily anomalies have an r-squared value of 0.37 over the period January 2003 thru June 2011. The relationship is significant. As expected, channel 5 data shows that the tropospheric temperature varies with ENSO.

Posted in Uncategorized | Leave a comment

AQUA Channel 5 and the Annual Temperature Cycle

UAH maintains an interesting website which provides atmospheric data from the AQUA satellite. Dr Roy Spencer of UAH introduces the nature of the data here.

Of interest is AQUA channel 5. Channel 5 measures a portion of the atmosphere which is largely the troposphere. The weighting function graph for channel 5 (red circle), and the other AQUA channels, is shown below.

The plot indicates that channel 5 largely measures the free troposphere, with some contribution from the lower stratosphere and from the near-surface layer.

AQUA has been operational since mid-2002. The UAH website provides daily AQUA data dating to mid-2002, including channel 5.

What does the AQUA daily data look like? First, let’s look at the time series of daily channel 5 values since January 2003. Then let’s look at the same data by day of the year,:

The plot resembles colorful spaghetti, with year-to-year variation. Now suppose that those daily values are averaged over 2003-2010 so as to present the “normal” tropospheric temperature for any day of the year. I realize that eight years of data hardly constitutes a robust climatology, but hey it’s what we have. Here’s a plot of that daily average:

The image shows a clear annual cycle. Global temperature on channel 5  peaks in July and August and reaches a minimum around January. That seems a bit odd, since global TOA insolation peaks in the Northern Hemisphere winter, when Earth is closest to the sun.

One potential explanation of this channel 5 annual pattern is based on the facts that (1) Earth’s land fraction varies by latitude (with a greater land fraction in the Northern Hemisphere than in the Southern Hemisphere) and (2) land has a lower heat capacity than the oceans.

Here’s the thought: if the fraction of TOA (“top of atmosphere”) sunlight falling on land varies through the year and if the land, due to its low heat capacity, readily transfers that energy to the adjacent troposphere, then the hemisphere with the greater land surface would show greater tropospheric warming in local summer and greater cooling in local winter than the other hemisphere. This would induce an annual cycle into the global data.

Let’s explore that. It is possible to create an approximation of insolation falling on land for each day of the year. This is done by using two sources of information. One is the NOAA insolation calculator, which provides TOA data by latitude, and the other is Figure 4 from this Milankovitch article.

I choose the middle latitude of each latitude band to represent the entire band’s insolation. To this I applied the fraction of land in the latitude band.

Here is a plot of the scaled “landfalling insolation” (actually, it’s the TOA insolation above land masses) and the annual cycle provided by AQUA channel 5:

The visual impression is that the two cycles are similar, with global tropospheric temperature lagging landfalling insolation by a month or two. The r-squared correlation value for the two is 0.72 with no lag, which is not shabby.

Here’s a scatterplot of insolation above land vs. temperature for each day of the year. This type of plot, in my opinion, communicates well the presence of a lagged relationship between insolation and temperature.

Suppose that we lag temperature 30 days behind insolation above land. What do we get? We get the plot below, with an r-squared value of 0.95:

So, it looks like insolation above land provides a pretty good explanation for the annual variation in global tropospheric temperature. as indicated on channel 5. And, global temperature appears to lag insolation above land by around 30 days, more or less.

One final observation of this data – there is a wiggly area in the Northern hemisphere winter, where the correlation seems to break down a bit. Why is that? Engineers are always curious about odd data wiggles – we can’t help ourselves.

A thought – possibly the wiggliness is related to Northern Hemisphere snow cover and its impact on absorbtion of sunlight. Here is an annual plot of NH snowcover by day of the year:

The plot indicates that NH snowcover remains more or less constant until about March. This happens even though high-latitude NH landfalling insolation is increasing. The albedo of the snow may effectively slow any warming of the land and associated troposphere until late winter, when the snow and ice melt. I have not explored this in detail to see if the hypothesis holds water, but it is plausible.

All in all, a basic seasonal pattern in AQUA channel 5 data can be reasonably explained. It’s nice when data behaves as-expected.

Posted in Uncategorized | 6 Comments

The Blog

This blog is, in a sense, my notebook, with a few parts in public view. It is not an attempt to persuade anyone of anything, especially in the Thunderdome that is the global warming debate. I don’t mind making mistakes if the mistakes are made in my pursuit of improved understanding and if, in the end, I learn something.

Most of the posts will involve the atmosphere in one way or another, particularly climate data.

Years ago I was an active participant at Steve McIntyre’s Climate Audit website. The “hurricane wars” were my special interest and I consider it an honor that Steve published several short articles of mine. However, I’m a lukewarmer, a moderate, not a skeptic. My views include:

* Earth is warmer than it would otherwise be, due to human influences, including our carbon dioxide emissions,

* reducing our use of and dependence on fossil fuels would have considerable benefits, but a large, rapid reduction would be be socially harmful, especially to poorer people. Even if the reductions are in the developed world, their impoverishment would hurt poor people in developing countries.

* climate science today has an unfortunate ideological bent which, unlike healthy science, wants to send dissenters to the gulag. That’s bad

* the eventual impact of the warming, in terms of benefits vs. negatives, is probably a wash, as I see no reason why the net impact should be predominately positive or negative,

* climate change could be harmful in places despite a largely benign overall impact. Local negative effects deserve attention and possibly impact mitigation (including resources from the broader world)

There – that should be enough to aggravate both sides!

Posted in Uncategorized | Leave a comment