For the alarmist mind climate can not been more simple. Carbon dioxide levels go up, temperatures goes up. Whatever weather event we encounter is caused or influenced by it. Nothing can even disprove this, there is no room for doubts with this simple logic.
This logic is based on several misconceptions. In some next posts I will explore some of misconceptions I had and how they changed.
The first misconception (being adrressed in this post) is: the earth has a global temperature, this is measured and it is going up in a way that is causing alarm. It even seemed to be accurate enough to capture an 0.8 °C increase in temperature over 160 years.
Just a couple years ago I had no doubt that this was feasible and that the science was mature enough to achieve this kind of accuracy. In my believer years I especially looked at the NASA-Giss dataset. Not really a surprise: this dataset is extensively used by alarmist minds and it had an aura of being trustworthy. Let’s look more into it.
Strange things start to happen when a person start to think logically about the things that surrounds him. I came to the realization that in reality the concept of a Global Temperature does not exist and it seems absurd claiming we could measure it accurately.
To begin with, temperature varies a lot. Not only in location, but also in time. In humans, taking a temperature is really simple. Stick a thermometer in your mouth, read the value and you will have an accurate measurement of the temperature inside the body.
Not so in Earth. There is not one convenient place where the temperature of the earth can be measured. For example in Belgium the South-East part (The Ardennes) has the highest elevation and in general has colder temperatures than the rest of the country. In the North-West there is the North sea and temperatures are moderate there. In the North-East there are more extremes in highs and lows. So even in a tiny country as Belgium there are several different influences on temperatures.
Even on a more local scale there are differences. I live near a hill, smack in the middle of the country. On that hill there is woodland and this has a slightly different temperature than its surroundings. Also a few kilometers from where I live there is another hill with a micro-climate where it is warm enough to cultivate grapes, something which is not possible in the place where I live, even being within walking distance.
There is not only a huge variation according to the location, each point will vary throughout the day and night. It will be coldest in the morning just before sunrise and warmest in the afternoon. Also there will be variation throughout the year (coldest in winter, warmest in summer and spring/autumn in between). And probably also longer cycles of 30, 60, 200 years,…
So, no place on earth will have the same temperature for very long during the day and temperatures will change constantly. Measuring the mean temperature will be quite a challenge. It is not possible to measure temperature at all those places, so the next best thing will be to measure as many points as possible. As been done in surface temperature datasets as GISS and HadCrut.
If all those stations were kept in the same way, this would give us some more idea of the temperature evolution over time (at least for the measured spots), but this is not the case. Stations are dropped, moved, instruments changed, surroundings changed,… Inevitably, the mean temperature will be the result of a statistical analysis, hopefully a good representation of the real temperature.
When one wants meaningful results, samples must be representative of the population. Bias in sampling will influence the end result. The problem here is that surface stations are situated in specific places. In or near cities, airports and other places where people most likely live. Excluding places where people normally don’t live (mountains, deserts,…). In the GISS dataset, most of the samples are taken from the United Stated, some in Europe and Asia and only very few in Africa and Australia.
This is called Convenience sampling. This means there is no real random sampling. Not all points have the same chance of being measured. Although convenience sampling has it merits, it is definitely not the right way to sample for a mean temperature. Especially when instruments/locations/… change over time.
Sampling in convenient places means sampling in/near cities and airports, therefor attributing to Urban Heat Island effect. Due to pavements/asphalt/buildings more heat is accumulated during the day and irradiated at night, therefor leading to higher temperatures than without these constructions. This could be compensated, but this will mean starting from assumptions. The more the assumptions agree with reality, the more accurate the result. But how to correctly compensate for all this bias?
This is not the only bias. I already learned about other siting biases like weather stations located next to air conditioner units, close to buildings and parking lots, even one on the roof of a building. These things undoubtedly will have an influence on the temperature reading and on the results after the calculations. Discovering this measurement bias was my first turning point from a believer to a skeptic view.
The ultimate question will be: how much does this non random sampling matters? That is an open question. Maybe the rest of the potential measurements cancels the bias of measurements out. But then, maybe not. Systematic bias is very unlikely to cancel out. If one want to have a result from this incomplete data it will necessary to make assumptions about the quantity of the bias.
Look at how the GISS dataset morphed over a couple decades from a cycle to almost a straight line. Which gives the impression that the scary result is dependent on new assumptions, not new measurements.
That is only land temperature. Earth is covered 75% by water. Measuring temperatures was first done by sticking a thermometer in a bucket of water drawn from the ocean over automatic systems of measuring the temperature of water in the intake port of large ships to buoys. It went from very scarce data in the past to more detailed information from 2003 (Argo).
What about satellite data? Coverage is much better, although not 100% of the surface (there are slices that aren’t covered and there is a gap at the pole). But these are not the datasets being used by alarmists and only 30 years worth of data.
But, but, doesn’t the Giss dataset is temperature anomaly, not absolute temperatures? Sure, it is and has it advantages and disadvantages. Maybe more on this in a later post. In Giss the result is the difference between the measured temperatures against the average temperature between 1951-1980. Smack in a period when there was a new ice age scare. Compare a current temperature with a average low temperature and this current temperature will be over accentuated.
Ultimately, why did I took it for granted? Every time I heard about it, I was used as something evident: “the temperature of the earth is rising”. This made me think it was evident. Science made quite some progress, why wouldn’t it possible that the temperature of the earth could be determined? But the temperature of earth is incredibly complex and ever changing. Now when someone tells me that the temperature of the earth (510 million square kilometers) could be measured with an accuracy of 0.1 °C from biased samples containing the data of a couple thousand stations, I would think it is ridiculous, something not to be taken seriously.