The data is out for both RSS and UAH, and I’m presenting them both here. Click for full sized graphs.
RSS from Remote Sensing Systems of Santa Rosa, CA. RSS data here (RSS Data Version 3.2)
UAH from Dr. Roy Spencer, University of Alabama, Huntsville. Reference: UAH lower troposphere data
Since Dr. Spencer released the April UAH data first on his own blog, I’ll give him the honor of explaining the data and possible reason for divergence of the two data sets.UAH Data
YR MON GLOBE NH SH TROPICS
2009 1 0.304 0.443 0.165 -0.036
2009 2 0.347 0.678 0.016 0.051
2009 3 0.206 0.310 0.103 -0.149
2009 4 0.091 0.126 0.055 -0.010
Once again there is a rather large discrepancy between our monthly anomaly (+0.09 deg. C.) and that produced by Remote Sensing Systems (RSS, +0.20 deg. C). We (John Christy and I) believe the difference is due to some combination of three factors:
1) we calculate the anomalies from a wider latitude band, 84S to 84N whereas RSS stops at 70S, and Antarctica was cooler than average in April (so UAH picks it up).
2) The monthly anomaly is relative to the 1979-1998 base period, which for RSS had a colder mean period relative to April 2009 (i.e. their early Aprils in the 1979-1998 period were colder than ours.)
3) RSS is still using a NOAA satellite whose orbit continues to decay, leading to a sizeable diurnal drift adjustment. We are using AMSU data from only NASA’s Aqua satellite, whose orbit is maintained, and so no diurnal drift adjustment is needed. The largest diurnal effects occur during Northern Hemisphere spring, and I personally believe this is the largest contributor to the discrepancy between UAH and RSS.
UPDATE: Basil Copeland writes in comments.
And for those who are unhappy with either linear or 4 order polynomial trends, may I suggest Hodrick-Prescott smoothing?
http://i40.tinypic.com/30ngom0.jpg
I like to also keep track of the USA48 UAH anomalies:
The USA48 series appears flatter than the global series. That’s an illusion created by the differences in scale. The global series is not as volatile as the USA48, because it averages out all kinds of regional variation in climate around the globe. The scope of this averaging can be seen by plotting the two together, on the same scale:
http://i41.tinypic.com/2rw8bhw.jpg
The “Average Decadal Change Rate” shown on the chart is calculated as 120 times the average 1st difference of the smoothed trend lines, a number that should be fairly immune to any claims of cherry picking.
Frankly, I was surprised. E.g., on its own, the USA48 chart looks flatter. But it isn’t, really. In fact, it is steeper. Before anyone concludes that the high rate of growth for USA48 somehow demonstrates AGW, do keep in mind that during most of this time frame, the PDO was in a warm phase, and that the PDO warm phase has a strong influence on continental US temperatures.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Don’t forget the sudden stratospheric warming event from January which influenced temperatures. It is clearly visible in the satellite data.
bill (01:26:43) :
“Thank you for removing the very misleading, unscientific, and totally in-appropriate 4th order poly fit to the curve Dr. Spencer choses to add to his plot.
http://www.drroyspencer.com/latest-global-temperatures/”
Anthony – could you please add back Dr. Spencer’s curve fit, as a linear curve fit is entirely inappropriate for modeling a highly non-linear process…Thanks
bill, interesting comment on poly fit to data.
I think single straight lines are misleading because the phenomenon is clearly non linear. Personally, I would choose 4th order as a conservative fit to a clearly cyclic phenomenon — but, then, discount extrapolations at the endpoints.
Nonetheless, a sequence of straight lines might make sense for the data.
That is, a horizontal line (at roughly 0.0) from month 0 to roughly 260, then a higher horizontal line (0.25) from 261 to roughly 350. And, now a third horizontal line from 351 on into the future — almost coincident to the horizontal line from 0 to 260 (looking like roughly 0.0 to 0.1 today).
In other words little net change over the last 360 plus months.
NOAA release US temp data for April 2009
“The average temperature in April 2009 was 51.2 F. This was -0.8 F cooler than the 1901-2000 (20th century) average, the 36th coolest April in 115 years.”
http://www.ncdc.noaa.gov/oa/climate/research/cag3/na.html
I wonder what shade of orange or red the GISS anomoly map will end up showing for the US for April 09?
I think everyone is worrying too much about minutiae of a few tenths of a degree. On both sides of the debate, this is much ado about nothing.
Global warming theory says these numbers should have increased by about 0.6C over this period.
John Finn (02:12:27)
anna v (02:23:38)
Allan M R MacRae (02:38:23)
Yes, John, I’ve read that on Dr. Roy Spencer’s web site, but the difference is just too big´! If the difference coming from different satellites for AMSU and UAH is really so big, it makes one of them totally irrelevant. AMSU would not even be “a rough guide” and there are not just “some data adjustments”… And then Dr. Spencer could not say: “(Want to see how the current month’s temperatures are shaping up?” as it does not really even indicate how current month’s UAH temperatures are shaping up?
anna v., Allan – no, I’m referring to channel 5. You have to choose either channel 5 from the drop down menu at the bottom of the page or 5 km on the left. Also choose temepratures in Celsius. Then you’ll get the absolute temperatures plus 1979-1998 averages and will thus see the anomalies.
http://discover.itsc.uah.edu/amsutemps/
rbateman (02:20:55) :
He’s right. The RSS data is always drifting higher. If your sensing equipment is depending on a time-length return for signal, a decaying orbit will make your data drift with it.
Who is right? If you mean Roy Spencer – of course he’s right, but diurnal drift is a well understood effect. RSS (and UAH previously) apply ‘corrections’ to the data to compenate for it.
DJ (03:35:18) :
You might also care to report on the linear trends…
UAH trend: +0.13°C/decade
RSS trend: +0.155°C/decade
I reckon the difference in trends is due to the lower RSS temps in the early years. In point 2, Roy Spencer says
“2) The monthly anomaly is relative to the 1979-1998 base period, which for RSS had a colder mean period relative to April 2009 (i.e. their early Aprils in the 1979-1998 period were colder than ours.)”
There is almost a ‘step’ change in ~1992 when RSS temps moved higher than UAH (Tamino covered this in a post). Since 1992, though, the trends have been almost identical (i.e. 0.22 deg per decade).
I’ve commented on points (2) and (3) so I might as well have my say on (1) which is
“1) we calculate the anomalies from a wider latitude band, 84S to 84N whereas RSS stops at 70S, and Antarctica was cooler than average in April (so UAH picks it up).”
The area between 70S and 84S only represents ~3% of the earth’s surface (can someone check that) so it would take a pretty big shift in temperatures in that latitude band to significantly affect the global figure. In summary, then
(1) Insignificant.
(2) Possible – but only if Aprils between 1979-1992 were relatively colder than other months in the 1979-92 period, otherwise the “discrepancy” would always be present.
(3) Most likely explanation – and, I’m glad to say, Roy appears to agree with me.
Conclusion: I’ve go far too much time on my hands.
E.M.Smith (03:02:59) :
“You know, it looks almost exactly like what is called a “blowoff top” in stock chart terms. ”
After some googlefu I found a page with some charted examples of a “blowoff top” to illustrate your point. Eerie:
http://www.clivemaund.com/article.php?art_id=1619
Thank Dr. Spencer for the explaination. If indeed all that info is correct, UAH is certainly more accurate and reliable than RSS.
We’ll probably have to wait for the UK Met Office and GISS and then we can say whether it’s UAH that is anomalous or RSS. If it’s RSS, then it’s probably the NOAA satellite drift that is to be blamed (for both RSS and AMSU), otherwise UAH has a problem to solve…
DJ (03:35:18) :
You might also care to report on the linear trends…
UAH trend: +0.13°C/decade
RSS trend: +0.155°C/decade
================================================
Along with the fact that any trend calculated over the span of the data needs to be adjusted for the major cooling impacts of El Chichon and Pinatubo.
Also need to report on the fact of downward trend for 12 years despite monotonically increasing CO2 levels.
C’mon DJ let’s party!
Bob Tisdale (01:36:36) :
Do you have any idea how they managed to measure the SST in e. g. Hudson Bay which was entirely ice-covered throughout April? Are they by any chance using the defective SSM/I for SST measurements as well?
For those looking elsewhere, or wanting the zonal data of UAH, look here:
http://vortex.nsstc.uah.edu/public/msu/t2lt/uahncdc.lt
After staring at these charts for the past year or so, it seems to me that the temperature exhibits, on average, a step function. From 1979 up to 1997 the temp osciallates around zero. Then we have what looks like an anomalous event of the 1998 el Nino, followed by a “relaxation” of the system. Then finally, the temps exhibit a step function up to a value of around 0.2 degrees for the last decade. I would graph it (using the old ASCII trick) to look something like:
As a physicist, and if I didn’t know that the data I was looking at was world temps, I would say the spike induced a “change of state” transition in the system. Definitely not a linear system, and certainly does not show a continuing upward non-linear trend as predicted by the GCM’s.
Wait: RSS is not global? How do the datasets compare if only the areas covered by both UAH and RSS are used?
Oops.. it looks like my ASCII graph got botched… moderator maybe you can snip that section?
Sven (05:27:23) :
anna v., Allan – no, I’m referring to channel 5. You have to choose either channel 5 from the drop down menu at the bottom of the page or 5 km on the left. Also choose temepratures in Celsius. Then you’ll get the absolute temperatures plus 1979-1998 averages and will thus see the anomalies.
http://discover.itsc.uah.edu/amsutemps/
And by what logic is the 5 kilometer height temperature to be compared with surface temperatures? 5 kilometers is the top of the Himalaya.
CH04 is called the “near surface temperature” and I would suppose that would be the one to compare with the other three measures of surface temperature though it does not say what height. -17C makes no sense unless it is higher than the one kilometer plot, which is around the -2 C level? In the 1 kilometer plot the difference with 2008 is still consistent and in the 5 kilometer plot but I do not understand the absolute temperatures and do not know what the anomaly is measured against.
I think we would all appreciate if if there were an explanation at that link of what the temperatures really mean with respect to the surface the we live on.
bill (01:26:43) :
Regarding your statement that a “4th order poly fit [is] . . . very misleading, unscientific, and totally in-appropriate.” No more so than a linear fit. In fact, give volcanic perturbations and well-known oscillations, a polynomial fit might be a better choice for a GMT time series. Consider this example: If in mid-1944 you did a linear fit of German vs. Russian military success, you would give the future to Germany. However, if you did a polynomial fit, you would get a better handle on the future. Of course, Dr. Spencer points out that his polynomial fit is not purported to have predictive abilities, and we should bear that in mind also when we see linear fits.
One thing I will have to say for CO2 theories is that it is understood that greenhouse affects are of a long-term nature and will not reflect in month to month data. Where I diverge on this statement alluding to the slow response of Earth’s atmosphere to CO2 is that I think it also takes a while for the Earth to respond to all sources of heating and cooling, with the possible exception of sudden intrusions of massive amounts of aerosols that stay suspended. Due to the seasonal nature of our atmosphere and the slow turning of our ocean conveyor belts, it takes a few seasonal/oscillating turns to buildup or release heat. The beating nature of the noise seems to follow this kind of natural oscillating method of cooling off and heating up.
The lack of data right at the poles is due to the nature of the orbits used to collect Earth Observation data at a consistent observation time across the globe.
The NASA satellites from which the data are generated are not in polar orbits ( even though they are called POES.) They are in specialized orbits called Sun Synchronous, so that their passage over the equator takes place at the same time every orbit all year long. This way the temperature is measured at the same time all of the time, along with the other measurements being made.
In order to accomplish this orbit, it is necessary for the orbit to drift “east” about a degree a day, as the earth moves in its orbit around the sun. This keeps a constant sun angle (beta angle) WRT the orbit plane. The orbits use a perturbation due to the earth’s oblateness to accomplish this. To get the 0.986 degree drift per day necessary, the orbits must necessarily be inclined past the 90 degree polar orbits to inclinations in the 94 to 105 degree range depending on altitude. So satellites in sun sync orbit cannot fly over the poles and cannot collect nadir data over those areas.
You can see this effect in the lack of UAH data above 84 degrees north and south. You can also see this in the Cryosphere today plots of icecaps. There is a hole in the data right at the poles. If you look at the polar ice plots, you can get a sense for the size of the missing area.
http://arctic.atmos.uiuc.edu/cryosphere/NEWIMAGES/arctic.seaice.color.001.png
anna v (06:15:15):
“And by what logic is the 5 kilometer height temperature to be compared with surface temperatures? 5 kilometers is the top of the Himalaya.
CH04 is called the “near surface temperature” and I would suppose that would be the one to compare with the other three measures of surface temperature though it does not say what height.”
No, anna v., we are not talking about absolute temperatures but anomalies. RSS and UAH are measuring mid troposphere and their only measure for comaprison is their own average (1979-1998, Met Office is 1961-1990 and GISS is 1951-1980). As MetOffice-GISS and RSS-UAH are measuring different different things (heights) and reference periods are different, they can not be directly compared as figures. But we can look at their anomalies compared to their own average and look at their how they “behave”. So – RSS stayed pretty much the same while UAH went sharply down. Now, if MetOffice and GISS would behave either like RSS ior UAH, we can see which one is having trouble.
And for those who are unhappy with either linear or 4 order polynomial trends, may I suggest Hodrick-Prescott smoothing?
http://i40.tinypic.com/30ngom0.jpg
I like to also keep track of the USA48 UAH anomalies:
http://i44.tinypic.com/2vxn3u9.jpg
The UAH series appears flatter than the global series. That’s an illusion created by the differences in scale. The global series is not as volatile as the UAH48, because it averages out all kinds of regional variation in climate around the globe. The scope of this averaging can be seen by plotting the two together, on the same scale:
http://i41.tinypic.com/2rw8bhw.jpg
The “Average Decadal Change Rate” shown on the chart is calculated as 120 times the average 1st difference of the smoothed trend lines, a number that should be fairly immune to any claims of cherry picking.
Frankly, I was surprised. E.g., on its own, the USA48 chart looks flatter. But it isn’t, really. In fact, it is steeper. Before anyone concludes that the high rate of growth for USA48 somehow demonstrates AGW, do keep in mind that during most of this time frame, the PDO was in a warm phase, and that the PDO warm phase has a strong influence on continental US temperatures.
Anybody know what’s going on with the Icecap website?
REPLY: server failure
Frank K. (04:44:13) :
I’ve come to hate seeing polynomial curve fits in climatological graphs. They’re okay when the important data is in the middle, but they diverge wildly outside of the data, and you can see the beginnings of that divergence at the edges. Since people are always looking for where the climate is going, they’re a bad choice.
I looked for, and failed to find, a web page showing the problem.
If someone has some graphing software handy, please create something showing a 4th order poly fit of the middle part of the UAH data but showing the curve over the entire history. The result would likely silence any fans of Spencer’s curve.
Also, a 3rd or 5th order fit would be interesting – I think the slope at the endpoints would have a different sign than even order fits.
REPLY: See the update to the post for the curves fitted with Hodrick-Prescott smoothing, I think it is most representative. Linea nd 4thord polynomial fits both have their problems. We’ve covered HP smoothing last year on WUWT. – Anthony