• Paris Agreement: PDF (hosted on unfccc.int website)
  • Paris Agreement: Website (hosted on un.org website)
  • Paris Agreement: Process (hosted on unfccc.int website)

It should be noted that the “Paris Agreement” text says:

Holding the increase in the global average temperature to well below
2°C above pre-industrial levels and pursuing efforts to limit the temperature
increase to 1.5°C above pre-industrial

It should be noted that:

Paris Agreement does NOT state what pre-industrial means

  • Many earlier documents talk about a baseline of 30 years centred on the 1750 baseline, which makes sense as “Pre Industrial” times, given when the industrial revolution started.
  • Using 1750 baseline may mean we are a few tenths higher, but there is greater uncertainty, including the possibility that the current global temperature is one tenth lower.
  • The climate data improves significantly from 1850.
  • It does make a lot of sense to use 1850-1900 baseline, which has become the standard.
  • The IPCC Special Report on 1.5C uses 1850-1900 baseline, and states what effects are expected for 1.5C and 2.0C above that baseline. This is of course the important thing: “what happens, and at what temperature value”.

Paris Agreement does NOT define measurement of 1.5°C and 2°C

  • There is neither a standard global definition of how to measure the global temperature anomaly, nor what 1.5C means in the Paris Agreement (daily value, monthly value, yearly average, 10 year average, etc…)
    • There are many different sources of data, which average to give pretty consistent values (thermometers, ship temperature measurements, satellite measurements, bouys descending through the water column) and plenty of proxies: tree rings, animal migrations, “temperature of catch”, “centre of mass of species”, bloom dates etc.
    • As per Climate Reporting – Why so many different values:
      • On a single month / single year basis the numbers jump around by about 1C
      • There are many different data sets
      • Therefore a “smoothing” technique is needed, to allow the trend to shine through. The choice of the smoothing technique can allow claims which are up to 15 years / 0.4C apart.
  • To my mind, it is intolerable that we can spend so much energy on “1.5C” and “2C”, but not know exactly what the “Pais Agreement Defined 1.5C and 2.0C” are.

Range of Believability

Despite my reservations above the Paris Agreement, all the GMST Data Sets I have worked with:

  • Are always within 0.2C of each other, on the same month
  • Are within 7 years of each other.
    • E.g. range of time between the slowest dataset to reach a temperature value (NOAA / GISSTemp), and the fasted to reach that value (Berkeley Earth).

The bigger source of temperature / timescales is “smoothing” applied, as per Climate Reporting – Why so many different values.

This means that the GMST Data Sets for key institutions are fairly well aligned on what the year-averaged GMST Temperature actually is, and the question is only how long it takes us to believe when the “GMST Trend Temperature” (or colloquially “The Climate”) has reached that point.

When climate moved slowly it was reasonable to take a 30 year running average. However now that global temperatures are rising at 0.2C+ per decade, using this backward-looking average means you are up to 15 years out of date, and likely 0.3C below reality. More believable smoothing uses 30 years centred on the current date, using real data for the previous 15 years and using climate models to project the future 15 years.

I don’t have access to climate models, so I am just using:

  • 181 / 241 month centred smoothing. Using linear / quadratic best fit get the middle value
  • At the start/end where I can’t get 90 / 120 months into the future / past, I use the full 90 / 120 months on one side, and as many months as possible on the other. This means the last 90 / 120 months is just pure linear / quadratic best fit. (E.g. pure straight line or pure quadratic curve).

See Climate Milestones – NOAA, GISS, Copernicus, HadCRUT, Berkeley Earth and Climate Milestones Copernicus – ERA5). These are very basic, but still just looking by eye, they are more believable than the pure 30-year-running-average values (which just run well below the actual data points). Surprisingly the linear-fit is less jumpy than the quadratic-regression. I will improve my smoothing algorithms as time / knowledge improves.