Friends, Opponents Sound the Defeat of President Trump's Climate Agenda

Thanks for the response, Nick. If one is calculating a global anomaly, why would one calculate a local anomaly first? I supposed the GSN-flagged stations were ones with better than average data and were an unchanging set, so that using them consistently would avoid the problem of constantly-changing stations. You’re not saying that one should use the station data to produce a local 1981-2010 baseline, then calculate an annual local anomaly from that, and then average all those anomalies across all 968 GSN statons to get the global anomaly, are you?
Also, can you explain the justification for the made-up data from interpolation being used in the calculations? Whenever I read the NOAA FAQs regarding such matters, they merely state their position as an assertion with no support whatsoever. For example, the Sahara Desert piece up above. When I read something like “Absolute estimates of global average surface temperature are difficult to compile for several reasons. Some regions have few temperature measurement stations (e.g., the Sahara Desert) and interpolation must be made over large, data-sparse regions,” I see that as hand-waving.
If one has four stations around the edge of the Sahara, with hundreds of kilometers between them and hundreds of thousands of square kilometers unreported, how can an interpolation between stations be considered data that can be used? That’s like measuring the time it takes for a ball to drop five feet, then using that number to interpolate the time it took for the ball to fall each inch, and then using those times to claim one actually took those measurements while the ball was falling.

No, we don’t really need a global average temperature. What we NEED is to know if it’s warming. If we can’t determine that from nearly a thousand stations, then it’s an exercise in self-delusion. At the very least, one can admit that accuracy to thousandth of a degree is impossible to achieve from measurements in tenths of a degree.

It has to be used. The global average is the estimate for the whole globe. You have to use the best knowledge that you have of the Sahara. Otherwise you don’t have an estimate for the globe. And then you don’t have anything. Just a few special points. Of course, the poor coverage of the Sahara is a negative, which shows up in total error. But Sahara is only a few % of world.

To expand on my previous brief statement that “we don’t NEED a global average temperature,” If we look at the idea from a purely scientific perspective, we might start by asking “I wonder what the Earth’s average temperature is?”, purely from a state of inquiry. One of the first caveats that might come to mind is “How do we determine a time frame to establish this average?”, because it’s obvious that the average global temperure is in a constant state of flux. The Earth is rotating and constantly showing a different face to the Sun — who knows by how much that average temperature might change at every instant, if we had some way of knowing exactly what it was for every instant. Any effort on our part to determine an average is going to be like trying to pin a butterfly to a wall by throwing a dart at it from 50 feet away.
So we start by looking at the data we have: thousands of weather stations around the world. Some are better quality than other; they’re thinly spread in some parts of the world, and heavily clustered in others. There are none to speak of across the 75% of the globe that’s ocean, and very few at the poles and in the large deserts of the world. Still, there is SOME data, so we have to figure out how to use it.
The spread of stations is too uneven to pretend to be able to get any kind of a global average temperature — but we can certainly determine a local average annual temperature, especially if we use once-a-day events like TMAX or TMIN. It doesn’t matter what time of day they occur, they’re the extremes of the day.
The other option is to take the temperature at a given time every day — say at local noon or local midnight — and determine the movement of temperature around those specific times. The altitude doesn’t matter, the humidity doesn’t matter; the only thing that matters is what was the temperature at that place at that time, or alternatively, what were the max and min temps for the day, regardless of when they occurred.
Now we’ve got real, solid, data. We can’t get a global average temperature from it, but we can see what’s happening at thousands of locations around the world.

It isn’t just semantics. Think about a poll of US opinion on say, who to vote for Pres. Now you can ask around for a number of people – then you have the opinion of those people. But if you carefully design a sampling scheme, you can ask 1000 people and say that represents the opinion of the US. There are two tests of that:
1. Do people who ask a different lot of people (other pollsters) get a similar result?
2. Does it predict elections?
Of course, people can point to failures and discrepancies, but the fact is that poll results are meaningful as an estimate for the whole US, even though they only ask 1000. They interpolate to the other 149,999,000 (my WAG) and it works.

Those polls have an error of +3/-4 percent, too. In a global average temperature of 16.0 C, that would be +0.48C /-0.64C — an entire 1.12 C worth of error. How does one square that with reporting an anomaly to three decimal places?
One also can’t take those polling results and drive out to Ottomwa, IO, find Ted Undecided, and predict which candidate he’ll be voting for, and then use that prediction as another sample of the population to improve the accuracy of the poll. — which is exactly what is being done by interpolating temperatures between Tripoli and Algiers for Ben Guecha, and then using that temp for Ben Guecha as a reporting station for global temperature data. It’s hoisting itself by its own bootstraps, and that’s not allowed.

Taking individual anomalies is part of the process of proper temperature sampling. For historic data, unlike polling, we have very little choice about sampling. But we can take out stuff that would be hard to interpolate, and we don’t really want to know about. That is climatology – altitude, latitude etc, but expressed as historic normals. Then we are sampling something much more uniform, but which still has the change that we want to track.
One other thing about interpolation – it doesn’t have to be always right, because it is only an intermediate in getting an average. So in your dropping ball example, all you want to know in the end is the time it took to drop. You could notionally do all those interpolations, and they might have errors, but when you add them up, you’ll still get the same total time.

But one can’t take all those interpolated values and count them as actual polling samples to decrease the error in the measurement. Besides, the rate of a falling body is part of the Law of Gravitation, which NASA uses to launch spacecraft and hit a dinner plate at the orbit of Pluto. Guessing the temperature of El Oued based on temperatures taken 400 kilometers away doesn’t have near the predictive ability as Newton’s Laws.