Another failing prediction of doom

For some time I have been working on my own reconstruction of the planet's temperature trend, which will be based on "raw" data, meaning data that has not been adjusted for various reasons. If you read the explanatory file, you will see that these stages are:

STAGE 0: Digital image and hard copy
STAGE 1: Keyed in native format
STAGE 2: Converted into common format
STAGE 3: Consolodated master database
STAGE 4: Quality controlled derived products
STAGE 5: Homogenized products

From the larger description, stage 3 combines separate files describing the same station (they might have been found in different archives, for example), and uses a computer algorithm to resolve any discrepancies. Since I want "untouched by human hands" data, that is as far as I can go. It is the disputes about (esp.) the homogenised products that are the entire point of contention in the dispute about the amount of historical global warming. If I take that data, I have effectively used their method in disguise. Hence I chose stage 3.

I have indeed looked into the various adjustments, time of observation error (TOBs) in particular, and I am preparing a paper describing why these adjustments are much larger than the real value. I am a qualified physicist, with an applied mathematics masters. So no, I am not some naive ignorant blundering where only the "experts" should dare go.

Re both latest comments:

"Fix the data": where the likely bad data is, we simply do not know what it should be. Just cover it up prior to Sept 2013. We only have a small region of believable data.

"...the graph you use to challenge their predictions shows a drop in temperature of a few degrees in about 6 years, and you agree that this is likely rubbish." No I don't agree at all. I am discounting everything before Sep 2013 (which is actually a minimum point). After that the yearly average falls about 1C. Weather moves faster than climate, a year can be significantly colder or warmer than the one before. The continuity of this region suggests that it is credible.

"Well, lead by example. Lets examine the data, look for errors and ways to allow for them, and then try and figure out what it suggests." I think you are expecting the impossible. Short of going there and discussing the measurements, there is no simple way to do better than we already have.

Just think of what we are trying to achieve here. We are not trying to find the "real" trend, or to get highly accurate records - because, for one, we cannot. We are simply asking if the reported scientific prediction of climate heat catastrophe by 2020 is at all reasonable. It would need a fast (and permanent) upward shift of something over 4C in the next four years to make that prediction come true. Nothing we do to the data, no improved analysis, will answer that question. Even if we could get the wonderful data you seem to think can be somehow got, it would not prove anything. The point if the original post was, simply "this doesn't look likely". It wasn't supposed to be, or claimed to be, a deep technical analysis, just a "We're nearly half way through, how's it going?" post. It looks a flat-out forgone fail to me, but not, so it seems, to you. So let's come back in four years and we'll know the answer.

Re the five year error bars: Errors grow the further out you go. Try shooting at a target 100 m away 50 times and measure your average spread. Now double the distance and try again. If the places five times further away in time only have a 5 year error, then there's no way the error on this place can be anything like 5 years. Here's the funny (to me) thing: every wacky claim from climate alarmists ("it is terrifying") is believed and published as true, but from a skeptic, a simple and reasonable assessment like a close-by error being smaller than a far off error, something that is absolutely reasonable and expected in experimental science - yet it get challenged as if it needs some kind of exhaustive validation. Frankly, it does not. The converse is what would need to be argued, because that is what would be surprising (very surprising).

I don't think I can add any more to this. Let's look at it in four years.

"4C ?? Are you using 35.4C as the reference point? It's a shift of around 2C compared with the more likely to be real peak in the 1960's, and that is well within the realms of possibility."

The prediction is that EVERY year will be hotter than ANY year previously. 4C is just an eyeball guess from looking at the graph, including some seemingly good data much earlier. Since years can easily be 2C different from those around, you need 2C on top of 2C to NEVER fall below the hottest ever year. Look, this whole article was a "doesn't look likely" surmise. It was not intended to be a scientific analysis. You expect me to waste huge amounts of time justifying a "look at this" article, but even if I did, it wouldn't prove anything - which you would undoubtedly then point out. You then add:

"Oh, and perhaps you can show me where the people who wrote this paper used the terms "terrifying" or "end of the world disaster"."

But I don't recall the authors of this study objecting when it was portrayed like this on mainstream media around the world. Frankly, it seems an exercise in plausible deniability: the hairy scary version is reported in big bold headlines, the actual report is more modest, but the scientists never correct the journalists. Sorry if I sound jaded, but I have seen this little trick run over and over. I have no time for it.

"Care to explain the logic behind that? Why exactly is the max more important than the min, or the average?"

I have been doing analysis of US temperature records, and I found that, in particular, the TOBs error, can be reliably identified and quantified in the max but not in the min. Also a recent paper argues, correctly in my mind, that max is a more reliable measure of global warming than min. See 

Christy, John R., and McNider, 2016. Richard T. Time Series Construction of Summer Surface Temperatures for Alabama, 1883-2014, and Comparisons with Tropospheric Temperature and Climate Model Simulations. Journal of Applied Meteorology and Climatology 55(3), 811–826, 2016. DOI: 10.1175/JAMC-D-15-0287.1

Average is just a useless measure because you learn nothing about extremes of either heat or cold.

"Exactly what did they say? Not the newspaper version."

To arrive at their projections, the researchers used weather observations, computer models and other data to calculate the point at which every year from then on will be warmer than the hottest year ever recorded over the past 150 years.

"it doesn't look to my untrained eye that a huge jump in temperature would be required for whole years to be on average hotter than before 2013"

Not some whole years, every whole year. The mean has to be so hot that a very cold year is still hotter than before.