Monday, 5 April 2010

The warmest year?

Since the whole brouhaha over stolen climate scientists’ emails broke last year, a clear emphasis of the science obfuscation lobby has been trying to con the public into believing the data is all flawed, and there’s nothing to worry about. One of the fall-backs in the attack on the mainstream has been the claim that satellite data hasn’t been showing as strong an increase as the data sets based on surface measurements. While NASA does their measurement independently from the UK Hadley Climactic Research Centre (CRU), the conspiracy-theoretic view is that they are all in it together.

Well, how is the satellite record looking lately? I moseyed over to the AMSU-A temperatures site maintained by Roy Spencer of the University of Alabama at Huntsville (a favourite of the denialosphere for his consistently lower warming trend) and Danny Braswell of NASA and did the following, which you can repeat to update the results:
  1. click Draw graph
  2. click all the checkboxes and click Redraw
This is my result:
 The orange line that stops in April is the daily temperatures for this year. Observe that since 10 January, every day has had the highest temperature for that date in the satellite record going back to 1998. This particular data only starts partway through 1998 so we don’t have the complete picture back to the last record or near-record year (depending whose data you look at; some put 2005 slightly higher than 1998) but if this trend continues, 2010 will easily set a new record for satellite temperatures. Is this because we’re in a super-monster El Niño? Not if you look at the SOI (most recent values at time of writing from the Australian Bureau of Meteorology):
The thing to note here is that a negative SOI value puts us in El Niño (higher than average temperatures) while a positive value indicates La Niña (lower than average temperatures). Also, there is a lag of about a year between SOI peaks and global temperature peaks. The El Niño in 2004-2005 with a peak early in 2005 resulted in 2005 being an unusually warm year. Do we see a similar pattern for the last 12 months? No. Although we have been in an El Niño phase since the second half of 2009, that effect has been nowhere near strong enough to account for a new record in temperatures, certainly not as early as January 2010.

Could the SOI be amplified by a rising solar cycle? Sunspot numbers are trending up but we are still pretty close to the solar minimum:
If you’d prefer to look at total solar irradiance (TSI) as measured by satellites, rather than sunspots, go to PMOD, where you’ll see the energy inputs from the sun are not on a steep upwards trend and should be significantly lower than in 2004-2005:
You have to line this data up with SOI to get a good idea of what is causing short term blips in global temperatures: El Niño does not in general coincide with the peak of the solar cycle. At some point this will happen, resulting a really hot year. But leave this aside, since it won’t happen soon. What we have now, early 2010, is a temperature trend tracking above all recorded temperatures since 1998 (or more properly 1999, since we don’t have 1998 temperatures in the first half of the year in the AMSU-A data set) and we can’t pin that effect on El Niño or on the sun.

What next? Is someone already planning to steal Spencer’s emails?


I’ve plotted the mean daily temperature anomaly versus 2005 (each day minus the same day in 2005) for the year to 25 April. The trend (linear regression) over this period is 6°C per century – making up for lost time from previous slower warming by satellite measurements? Oh, and the correlation coefficient of 0.79 is highly significant even with this small sample size. Enjoy (or not, if you don’t want warming to be real). For missing data (a small fraction of the total), I took the average of the two nearest days that did have data.
And here's the AMSU-A big picture for 25 April 2010:

and an update on the SOI picture:

SOI is not the whole picture for predicting the effect of ENSO on temperature; a more comprehensive model clearly shows 2010 should be a cooling phase, not a warming phase.

Still lots of time in the year …

Yet Another Update

If you go to the site now, it says Channel 4 failed in 2008. That’s what happens in science: a weird result is more likely to be a consequence of instrument failure than anything else. The major data sets do show 2010 as one of the warmer years though not by a significant margin (despite the other data indicating it should have been a cool year, including the solar cycle starting to exit an unusually deep low and a strong La Niña).

What I find particularly odd about all this is this note (dates in US format, so this is in March)
03/06/2008/1200UTC:NOAA-16 AMSU-A channel 4 has gone bad.  As a result, NOAA-16 ATOVS sounding files are no longer being produced by NESDIS, thus they are no longer in the "atovs" dumps in the GDAS and CDAS networks and they (temperature retrievals) are no longer available for assimilation by the CDAS.  The "atovs" dumps now contain only NOAA-15 ATOVS soundings and only these temperature retrievals (cloudy only) are assimilated by the CDAS.  In addition, the failed channel 4 data has resulted in no NOAA-16 AMSU-A data being assimilated by the NAM/NDAS or GFS/GDAS GSI (even though these files are still being produced by NESDIS and NOAA-16 AMSU-A data continue to be dumped in the "1bamua" files in the CDAS, GDAS, GFS, NDAS and NAM networks).
It seems someone knew nearly 2 years before the weird 2010 data went live that there was something wrong with the data. Curiously, we didn’t see wild accusations flung around the blogosphere about this one. I leave it to the reader to explain.


Hank Roberts said...

Funny how no matter whether you chose to see the temperature in Centigrade or Kelvin, the chart always tells you the difference in Fahrenheit.

Steve Bloom said...

FYI, Philip, the 20-year record choice on the plot adds the record for the prior MSU instruments going back to 1979. It doesn't give you individual years, but adding it in will show the true maximum (per UAH, anyway) for the entire period. The first three months of this year look a little less impressive when one does that, but still quite striking. Likely most of the difference for those months is due to 1998.

I suspect the prior 20 years is de-emphasized because it's much less reliable than the AMSU period, due to uncertainty arising from splicing multiple MSU instruments and constant loss of altitude by the satellites. Unlike its predecessors, the satellite carrying the AMSU has station-keeping capability.

Philip Machanick said...

Hank: I noticed the Fahrenheit thing too. Perhaps a plot to make the difference look bigger? A "trick" maybe?

Steve: I hope no one will read this article as an attempt at making a definitive statement about temperature trends over a few months but still, it's interesting that even the most conservative data set is now looking bad for denial. I'd like to see how it looks by the end of the year. And of course there's always the possibility of a correction.

Steve Bloom said...

RSS doesn't seem to care much about this short-term trend stuff, but you can get the daily data from them. The comparison might be interesting.

Philip Machanick said...

Steve, RSS has a much stronger trend that Christy likes to admit is possible (1.57K/century lower troposphere).

My point here is that even attempts at finding the lowest possible trend in the data are breaking down. The data is what it is. If you try to beat a lump out of rug, it comes up somewhere else.

If you check out Christy's blog you'll see he is switching focus to claiming it's all urban heat island (how this is heating the Arctic more than anything else is something I'd be keen to hear him explain).

pough said...

Can anyone tell how Christy's UHI issue affects his own satellite record? Doesn't it match pretty well with the supposedly erroneous thermometer records? So are the satellite measurements affected by UHI as well, or do they magically have some other, still mysterious but exactly the same as UHI issue?

And those various studies that show UHI to be a non-issue, what's to be made of those? Can we safely ignore them because Christy has a suspicion?

Let's ask the stolen emails:
"Christy and Spencer have made a scientific career out of being wrong."

Tim Curtin said...

A good article, Phil, but as you say early days, with part of 1998 missing. But in any case, so what? Have you been frying in Brisbane this year? How have your rains been? Here in Canberra we had our first frost on Monday am, relatively early, and the best Jan-April rainfall for 8 years, despite (because?) of the heating you report. In general I think we will all survive +3oC from doubling of CO2, especially all those who live in Northern Europe and Canada, especially Norwich and Edinburgh, and an extra 3oC from Khrtoum to Dubai will not even be noticed by the residents, inured and adapted to heat as they already are. Brisbane is already nasty in January and can hardly get worse!

Philip Machanick said...

Tim, good to hear from you again.

Unfortunately we can’t pick and choose where the extra heat goes. Last December, the Arctic was way warmer than average, but parts of Europe and the US had extremely cold weather. But anyway, let’s say for argument’s sake we end up with Siberia much more habitable, and the Sahara desert extends its reach into African farmlands. Do you think Russia would welcome millions of African immigrants?

I’m inclined to the view that the clean energy experiment is lower in risk.

Anonymous said...

Actually, I think I recall from coral data that the last time the planet was 2 degrees warmer (about 110 ka BP), sea level was about 6 m higher. Canberra would probably be OK but not sure about Brisbane.

Philip Machanick said...

Anon (please give a name so I can tell the anonymice apart): the effect of a 2°C rise depends on where you started. Much of the existing long-term land ice would not be there if we started from an ice-free world at today's temperatures. Your dates are a bit out; 110k years BP would have put you at the start of the ice age (lasted to about 10kybp). Nonetheless there is evidence of very fast ice loss in interglacials, and some of the latest research is predicting 1-2m sea level rise by the turn of the century. More here.

Also, I took another look at ANSU-A, now extended up to 28 April. Still clearly above the earlier years. James Hansen has predicted 2010 will be very likely a record year. Maybe he knew something the rest of us don't. Keep watching the data.

Anonymous said...

Actually, 10 ka BP is closer to the Holocene Climatic Optimum than Last Glacial Maximum. I found one reference, it is the Last Interglacial (ie last ice age + one Milankovitch cycle). Details here, they also say about 6 meters.

Anonymous said...

What's the trend for January 1999 to December 2009 ?

Philip Machanick said...

Last anonymous:

It makes no sense to look at a trend across a single year. You get strong correlation between individual daily data points and even consecutive months and this kind of internal correlation (autocorrelation) makes for very low statistical significance of any analysis over that sort of time period. This is why most climate studies work with average of at least 12 months, and generally longer.

It's pushing it a bit even to look at the slightly longer periods I'm looking at here. My essential point is that natural variability should have us at relatively low temperatures over this period, yet AMSU-A has 2010 increasingly looking like an anomalously warm year. If it carries on at the current rate, it could be as much as 0.3 above the previous maximum. Still early enough in the year for this to turn around of course.

Not enough yet to write a scientific paper: I'm just watching because the data looks anomalous. Given Spencer's history, you can't say he's faking the data to make it look alarming ...

J C Brookes said...

This size of the difference between this year and any others is staggering. It can't be true? I doubt it will continue like this for the whole year. If it does, wow....

Philip Machanick said...

Digger, the differences are bigger than you'd expect from NASA GISS and HadCRUT (both currently only updated as far as March). It's possible that the UAH people are to some extent correcting for an earlier bias towards minimising the heating trend. Even so it's hard to see what's happening now slowing up, because we are moving out of a solar low, and moving into El Niño warming. James Hansen is on record as predicting 2010 will set a new record, unless something exceptional happens.

Hank Roberts said...

Stoat's on this now too.

I tried email to the NASA guy whose name is at the bottom of the Spencer temperature charting page, suggesting it'd be worth a remark to the public if there's anything weird going on with the data or the satellite.

And as I mused somewhere else, what's more likely, that the satellite is broken, or the climate is broken?