But first I should mention an error in my earlier processing of V3b, noted in comments here. ERSST3b came in decade files; that helped, because decades before the present did not change and did not require downloading. So I collected them into a structure, and with each update to the current decade (every month) merged that with the earlier decades.
However, I misaligned them. Most climate data uses the strict decade numbering (1961-90 etc), but ERSST uses 2000-2009. I got this wrong in the earlier data, creating a displacement by 1 year. Fixing this naturally gave better alignment with other datasets.
I'll describe the new TempLS version in detail in coming days. V 2.2 of TempLS had become unwieldy due to accumulation of options and special provisions for circumstances that arose along the way. I have made a new version with simplified controls, and using a simple iterative process described here. It should in principle give exactly the same answers; however I have also slightly modified the way in which I express ERSST as stations.
The new version is an R program of just over 200 lines. Of course, when simplifying there are some things that I miss, and one is the ability to make spherical harmonics plots (coming). So for a while the current reports will lag behind the results tabled above. That table, and the various graphs on the latest temperature page, will now be using ERSST v4.
Below the fold, I'll show just a few comparison results.
I've used running means to smooth out monthly noise, and focussed on the period since 1990. At some stage I might say more on the V4 changes, but that has been covered extensively elsewhere. I show the old results before fixing the missing year problem, then the V3b results after fixing, still using V2.2. Then there are results using new ERSS v4 and new TempLS V3. Ideally I would separate the effects of the new code and new ERSST, but getting V3b working with ERSST v4 would take a while, and the combined effect makes very little change.
So here are the grid weighted results. The effect of the year error is clear, but there is little change going from ERSST 3b and TempLS2.2 to ERSST 4 and TempLS 3.
And here are the grid weighted results, which are similar.
I haven't been paying close attention recently but I don't remember such a large misalignment when comparing TempLS to other the analyses. How long has the error been around?
ReplyDeleteIn any case, cool stuff.
Have you ever considered adapting it to accept other data sources like radiosondes or marine air temperature? It's always seemed to me that the least squared method combined with a Berkeley earth style "scalpel" method would work well for those types of data. For Radiosondes you could compare against satellite data (either processed or "raw") to detect discontinuities. For MAT you could compare against the standard TempLS output.
Obviously, a lot of work, but you seem to have infinite energy!
CCE,
DeleteYes, I was surprised that it had escaped notice. It shows up more here with the annual smoothing. The file with the error is from August 2011. It mostly affects pre-2010 data, which are not closely watched. SST is less volatile, so often the change over a year isn't great.
The new version will show just what is involved in using more exotic data. I don't think the scalpel is a magic bullet; splitting series comes at a cost. I think both radiosonde and MAT would have coverage problems.
FWIW, there is a new merged radiosonde dataset going back to 1905.
Deletehttp://www.earth-syst-sci-data.net/6/185/2014/essd-6-185-2014.pdf
http://www.earth-syst-sci-data.net/6/297/2014/essd-6-297-2014.pdf
http://onlinelibrary.wiley.com/doi/10.1002/2014JD022664/full
It looks much better now. Feb 1998 is the warmest month of the 20th century and Jan 2007 is the warmest ever, like the other global data sets.
ReplyDeleteCCE, KNMI climate explorer has ICOADS v2.5 MAT with global average, updating every month, in contrast to the UK MAT series that were discontinued 2010. However the ICOADS MAT are raw, no adjustments for ship heights etc, but that has probably not changed much the last decades. Global MAT for May is up 0.06 C from April for instance, which agrees with the major SST datasets.
Nick, have You ever produced a fully global land station only TempLS mesh, a kind of Gistemp dTs analogue? The Gistemp is limited by the max 1200 km infill, C&W produce a CRUTEM4 kriging but unfortunately put a land mask on it. BEST Land with their large station data base and ability to use short segments of data would be ideal for this purpose, but they also use a land mask.
Olof,
Delete"Nick, have You ever produced a fully global land station only TempLS mesh"
All it needs is to leave out the SST stations. I have experimented over the years with reduced stations numbers, as here and here. There actually isn't, IMO, much justification for retaining the full land set with the oceans so poorly represented.
The TempLS60 is quite impressing, it follows Gistemp dTs very well, and is not much noisier. Yet, there is seemingly room for improved ocean coverage, Hawaii and the Azores for instance.
DeleteThis kind of experiment with reduced number of station gives confidence to e.g. RATPAC with 85 sites with a good global distribution.
My interest in pure 2 m air temp datasets is for CMIP5 comparisons, that by standard are 2 m temps. The CMIPs themselves tell that SST and 2m air over ocean don't follow each other in a warming world. The SST trend for 2000-2030 is about 25% lower than the air over ocean trend...
Interesting topic certainly. There is a group particularly looking at the question you're discussing in particular.
DeleteBtw, the recent century trend (2000- now) of TempLSmesh has increased slightly with the correction. It is now 1.44 C/ century, beating C&W, new NOAA, Karl et al kriging. Wonder what the inclusion of ERSST v4 will do with this trend, 1.6 C... ?
ReplyDeleteOlof,
DeleteI'm now using v4. That may have made part of the difference. In fact, I now have the new reporting system, with new maps etc, all using TempLS V3 and ERSST V4.