Monday, October 20, 2014

More "pause" trend datasets.

In two recent posts (here and here), I have shown with some major indices how trends, measures from some variable time over the last two decades and now, have been rising. This is partly due to recent warmth, and partly to the shifting effect (on trend) of past events, as time passes.

This has significance for talk of a pause in warming. People like to catalogue past periods of zero or negative trend. A senior British politician recently referred to "18 years without warming". That echoes Lord Monckton's persistent posts about MSU-RSS, which does have, of all indices, by far the lowest trends over the period.

Here I want to show results about other indices. Cowtan and Way showed that over this period, the trend in Hadcrut was biased low because of non-coverage of Arctic warming. I believe that TempLS with mesh weighting would also account properly for Arctic trend, and this would be a good way to compare the two, and see the effect of full interpolation. I expected GISS to behave similarly; it does to a limited extent.

So a new active plot is below the jump. You can rotate between datasets and months separately. There is also a swap facility so you can compare the images. And I have individual discussion of interpolation data vs grid data groups.

Sunday, October 19, 2014

Tails of the Pause.

I've been writing lately about matters which, I'm sorry to say, lack scientific gravity. One is the possible record warm 2014, and the other is the tailing of the Pause, as measured by periods of negative trend. My excuse is, people do talk about them, and there is interesting arithmetic which I can illustrate.

In my "pause" posts, I showed plots of trend of global temperature to present, plotted for periods shown on the x-axis, with trend shown at the starting point. A "pause" starts when, for some index, the axis is first crossed from pos to neg. The plots were active, and you could see the curves rising steadily over recent months. This meant the start of the pause moves forward, with eventual jumps where a previous excursion below the line no longer makes it.

Here is the recent active (buttons) plot to show that effect:


In this post, I'll quantify the rate of motion, and describe how much cooling would be required to reverse the trend. The effect of a new month's reading depends on its status as a residual relative to the regression line for the period - ie is it above or below the line, and by how much. But one reading is a different residual for each such period. I plot the present month as a residual, again referred to the start year, and also plot the rate of change of trend produced by the current (August) temperatures.

Friday, October 17, 2014

Record warmth in 2014?

Not according to the satellite measures; they are showing quite a cool year so far. But surface measures, apparently propelled by SST, have been consistently high since March, and a record for calendar 2014 looks possible.

In August 2010, I showed a plot of the progress of the cumulative monthly anomaly sums, which will reach the final sum that determines the year average. 2010 did turn out to be the hottest year in many indices. It was different in that the El Nino was late 2009/10, so late 2010 was cooling. At this stage 2014 seems to be warming.

So I started to repeat that 2010-style plot, which is below the jump. It didn't work as well; the variation doesn't much show. But it puts the thing calculated in context - a cumulative sum that, if it exceeds 2010 at year end, will set a record. I've shown the progress of 2005 (a previous record), 2010 and 2014, with a line showing the 2010 average rate. The plots are spaced with an arbitrary offset.

But, more effectively, there is then an active plot with the average 2010 trend subtracted. The variation is clearer. The key thing is not so much whether the current total is above the line, but how it is trending, which is a measure of current warmth.

Thursday, October 16, 2014

QC for TempLS

I plan to do more with TempLS (see last post) so I want a stable quality control (data) scheme. GHCN unadjusted is a document of record, and there is weird stuff in there which it seems they don't like to touch. I've noted current examples earlier in the year. So I did a survey of the data since 1850. Here is R's summary of the monthly averages:

Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
-87.0 6.6 15.9 14.5 24.1 154.41166444

That's out of 10349232 months (of years with some data). Yes, the max is 154.4°C. There were 28 months with a min/max (not max) average >50°C.

To be fair, they use flags, and these oddities seem to be mostly where a decimal point slipped in the originating data. But they are big enough to have effects, so I have been using my own QC. On first look, I found the GHCN flags numerous and unhelpful, so I used a scheme where I checked with the adjusted file. This seemed to weed out the problem points without replacement. However, it excluded a lot of other points, so I allowed those if within 3°C of the appropriate mean.

Monday, October 13, 2014

A catch-up on TempLS

I've been writing a lot about TempLS (my global temperature index) recently, and realizing that I don't have a unique reference that explains exactly what it is and what has recently been happening to it.

TempLS dates back to a period in early 2010 when there was a flurry of amateur efforts to replicate the monthly global surface temperature indices from the major producers (which some thought suspect). This post by Zeke (with links to earlier) gives an overview. Jeff Id and Romanm started it with a reconstruction that used a least squares method for aggregating a single cell, yielding offsets rather than requiring a fixed anomaly period. I thought that could be applied to the whole recon.

So I developed TempLS, which was basically a big OLS regression, based on GHCN unadjusted station monthly averages. It was quick to run, and I incorporated choice mechanisms which made it easy to calculate regional or special (eg rural, airport) averages. A rather complete summary of this stage of development is here. An important feature was the incorporation of SST data. This comes gridded, often 5°x5°, and so I simply entered these as stations.

I made a point of using unadjusted GHCN, because there were many claims that warming was an artefact of adjustment. I have myself no objections to adjustment, though I did show that it makes relatively little difference to the index.

TempLS combines weighted regression with spatial integration, much as BEST did later. It weighted initially by the inverse of grid density, estimated by stations/cell in a 5°x5° grid. I posted at one stage a very simple version for incorporation in Steven Mosher's RGHCNv3. You can regard this weighting as that which a spatial integration formula woud provide, with each grid estimated by its station average anomaly, or equivalently, each function value (observed average) assigned an area equal to its share of the cell.

Sunday, October 12, 2014

GISS September 0.77°C, up by 0.08°C

This my first report in the new style, where I record details of both mesh-weighted and traditional grid-weighted TempLS, along with the latest GISS. It is part-mechanical, but I annotate. TempLS numbers on the latest data page are constantly updated, but the monthly reports here won't change.

So the headline is, a very warm month. I think in GISS at least, 2014 may well be a record year. TempLS mesh showed a similar rise; TempLS grid dropped slightly. I expect the mesh version to more closely follow GISS, and the grid version to continue tracking NOAA. My next post will probably be an updated explanation of TempLS.

I've given both TempLS maps below. You can see again that TempLS mesh is closer to GISS than grid.

New ideas on TempLS reporting - mesh

For over three years now, I've been running my least squares based GMST index TempLS each month and reporting the results, with a second post comparing with GISS. See here and here for last August. The second has past links. I did a mini-review here recently, and there is a new summary of TempLS here.

I'm planning a change. For some time, I've believed that using mesh based weighting (see here and here, for example) is better than what I call grid-based weighting, where observations are weighted on a cell-based density estimate. I was deterred from changing because the mesh generation took a long time, but I've fixed that. I'll persist with the grid model because as I noted, it has uncanny agreement with NOAA, and also tracks HADCRUT well. But I think that it also has the faults of those, in dealing with empty cells. For TempLS it takes the form that stations in empty areas have a capped weight based on the size on one cell, which in the Arctic can be small.

The data cycle goes like this. ERSST posts a preliminary on about 3rd or 4th of month. It's actually complete, and the numbers are little different when updated in late month. GHCN starts the month with a rush of numbers from places with efficient electronic systems, and then stuff trickles in fairly slowly. Lately GHCN (unadjusted) has had early gyrations too. But notwithstanding, I think it is meaningful to to a mesh baed calc as soon as the SST comes in. Mesh is more robust to missing.

I've now posted an automatic report on the latest data page. It shows the current mesh-based report, with maps of temperatures and of reporting stations. I think I can maintain this, reporting with every new GHCN (most days). It would flip to the new month when the SST is posted. Obviously, early figures would be subject to change.

This would supersede my first monthly report on the grid results. I'd still publish the GISS comparison, and record the grid and mesh TempLS results there.

Update. GISS has produced a map for September, and it says the Sept temperature is 0.78°C, up from 0.70°C in August. This tracks the mesh TempLS rise from 0.628°C to 0.673°C (grid TempLS went down slightly). I expect that mesh TempLS will follow GISS more closely. I'll post the GISS comparison soon. The new number is not yet on their datafile. It's getting warm.