RealClimate logo

Two graphs show the path to 1.5 degrees

Filed under: — stefan @ 21 April 2021

In the Paris Agreement, just about all of the world’s nations pledged to “pursue efforts to limit the temperature increase to 1.5 degrees Celsius above pre-industrial levels”. On Saturday, the top climate diplomats from the U.S. and China, John Kerry and Xie Zhenhua, reiterated in a joint statement that they want to step up their climate mitigation efforts to keep that goal “within reach”.

But is that still possible? Here are two graphs.

Global temperature trend (relative to mean 1880-1910, NASA data). The colored curve shows the moving average over 12 months, the black line the linear trend over the last 50 years. Transient warmth following two strong El Niño events in the tropical Pacific is indicated by arrows. If everything continued like this, the 1.5 degree limit would be exceeded around 2040.

The first graph shows the global temperature trend. Warming has progressed essentially linearly for fifty years in response to increasing CO2 emissions. Although the latter accelerate the rise of CO2 in the atmosphere, on the other hand, radiative forcing (which causes warming) increases only with the logarithm of CO2 concentration, and therefore roughly linearly since the 1970s. Any acceleration of warming over the last decade is not a significant trend change. It is linked to two El Niño events in recent years, but that is part of natural variability. Does anyone remember the discussion about the supposed “warming pause” in the early 2000s? It also never was statistically significant, nor did it signify a trend change.

Therefore, if emissions continue to grow, we expect a further roughly linear increase in temperature, which would then exceed 1.5 degrees around 2040. If we lower emissions, the trend will flatten out and become roughly horizontal as we reach zero emissions. Therefore, these observational data do not argue against the possibility to still keep warming below 1.5°C.

Exemplary emission trajectories with CO2 emission budgets that, according to the IPCC, correspond to limiting warming to 1.5 °C with 50% probability (solid) or limiting it to 1.75 °C with 67% probability. The same emissions as in 2019 were assumed as the starting point in 2021, assuming the “corona spike” in 2020 is likely to be temporary.

The second graph shows global CO2 emission trajectories with which we can still limit warming to 1.5 °C, at least with 50:50 probability. This means: given the uncertainties, this could also land us at 1.6 degrees, but with a bit of luck, it could land us a bit below 1.5 degrees. The core conclusions:

  •     It is not yet impossible to keep warming below 1.5 °C.
  •     This requires roughly a halving of global CO2 emissions by 2030 (as already stated in the IPCC 1.5 degree report).
  •     If the world dithers for another ten years before emissions fall, it will no longer be possible (red curve).

It should be noted that I have not assumed net-negative emissions here. Many scenarios assume that we first emit too much and that our children then have to pull CO2 out of the atmosphere after mid-century – I think this is not very realistic and also ethically questionable. I think we will probably not be able to achieve more than reducing global emissions to net zero. Even that would require CO2 sinks to compensate for unavoidable residual emissions, e.g. from agriculture.

Conclusion: The limitation to 1.5 degrees is still possible and from my point of view also urgently advised to avert catastrophic risks, but it requires immediate decisive measures. I am curious to see what the climate summit scheduled by US President Joe Biden will bring in the coming days!


Fact check by Climate Analytics to the claim that we can no longer limit warming to 1.5°C.

This article originally appeared in German at KlimaLounge.

166 Responses to “Two graphs show the path to 1.5 degrees”

  1. 151
    nigelj says:

    Victor @145 ” V: Yes, they (aerosols) can spread great distances, but as they do, they dissipate. The farther from the source of the pollution, the less the cooling effect.”

    True. And the source of aerosols was strongest in the northern hemisphere mid last century and it had a pronounced cooling trend. The cooling trend mid last century was less dramatic overall in the southern hemisphere. Exactly as you would expect as aerosols dissipate.

    Other factors also contributed to the cooling trend mid last century. There were more volcanic eruptions than normal mid last century.

  2. 152
    MA Rodger says:

    Victor Grauer @145,
    I don’t know whether it is because you are still head-down under your bridge or clattering around above trying to intimidate the passing goats with your trollish aggression and maybe blinded by the glare of the sunlight, but “I fail to see” is entirely what we have come to expect from you. Again and again and again.
    Do you wish the incredible level of troll-headed blindness you have exhibited down this thread-so-far pointed out to you?

  3. 153
    Barry Finch says:

    Re: Various exchanges about localization of aerosols from human activity. It is not only the aerosols which move with the wind (and disperse, having less effect). (Ahem) the wind also moves with the wind, the air moves and takes its temperature with it. The air temperature does not disperse so much as the aerosols. It is the air temperature that is being measured for GMST, only partly dependent on the solar radiation at the place where it is being measured unless there is a flat calm, and also very much dependent on where the air blew in from. I recall a beautiful sunny afternoon here with air at -25 degrees arriving from the north. It would have been above zero if there’d been a flat calm all afternoon because the Sun was beaming down from a clear sky. This is why surface-air temperature measurement is only partially related to local conditions, because air moves (as stated by others). Air takes energy ~32S-38N northward to 38N-90N at a power of ~6,000,000 gigawatts.

  4. 154
    CCHolley says:

    Richard Caldwell @150

    Victor is an entertainer.

    Well, I suppose some people considered Adolf Hitler’s speeches entertaining too.

    And yes, jokes are meant to be entertaining, but not sick jokes done in bad taste. Sorry, Victor is just a bad joke and not entertaining in the least to anyone except to himself.

    Perhaps you should ponder the difference between science and entertainment?

    Oh, I get the difference, but unfortunately, Victor really does not. And perhaps only a fool would think otherwise.

  5. 155

    #138, 140, 141 (MAR & Barry F)–

    Thanks for the comments. It’s interesting about the regional differences. In one sense, maybe a bit unsurprising, since in general the smaller the physical scale, the more variable the record–but hemispheric scale is not all that small! And yes, TLT and SAT are not the same thing. (Might the land-sea ratio difference between hemispheres be relevant to this RSS-UAH hemispheric temperature trend difference?) By the way, MAR, your link didn’t work–got a ‘verboten’ message when I clicked.

    I’d only add that in defense of the significance of the trend agreement between RSS and UAH that it’s my understanding that the growing discrepancy between RSS and the other datasets that became so marked a few years back was a leading indicator that something wasn’t right with their algorithms, and that this helped lead them to the diurnal drift/time of observation issue that was the key to the new version. If Mears et al. took TLT/SAT consilience that seriously, then I’m certainly content to do so!

    Widening focus just a bit, it’s interesting that both satellite records have had to make such major adjustments. (The biggest for UAH was the 2005 correction, I think.)* It really highlights the non-trivial nature of using microwave radiance measurements to tease out temperature at various altitudes.


  6. 156

    #148, RC–


    Well, if I’m right about a ‘matter metaphysical’, then it may just be ‘the very first time.’

  7. 157
    MA Rodger says:

    Kevin McKinney @155,
    Sorry about the ‘verboten’ URL. I’m still wrestling with what the new GoogleSites does-&-doesn’t do. The old version was silly enough. The new one appears to be stepping further into the silliness. So perhaps this URL will prove ‘zulässig’ – fingers crossed.

    The analysis comparing GISS SAT, UAH and RSS TLT does become more complex if the globe is cut into more complex bits than hemispheres but this is worth the look. You suggest Land/Ocean which can be done by comparing NOAA with UAH.
    Global Land/Ocean shows NOAA SAT & UAH TLT pretty-much in step for the Ocean anomalies, both showing +0.45ºC warming (1979-2019 data) which is roughly matching GISS/UAH SH (+0.4ºC).
    The NOAA/UAH Land comparison again mirrors the GISS/UAH NH finding, the warming over the full period matching even more exactly, both SAT measures showing +1.0ºC warming while both UAH TLT measures showing +0.6ºC, the difference a pretty linear affair if you don’t ignore the big Land/Ocean wobble due presumably to the 2016 El Niño.
    A more-exact timing of these divergences is perhaps important. In the GISS/UAH/RSS NH comparison, the three are in reasonable agreement since 2013 before which the TLT both diverge from GISS with the UAH/TLT divergence all prior to 2005, this also the timing for the NOAA Land/UAH Land divergence.

    It is possible to dig a bit further as GISS gives anomalies by latitude (while NOAA could be split by NH/SH as well as Land/Ocean).

    The comparison by latitude shows GISS, RSS & UAH in rough agreement in the Tropics showing +0.5ºC while in the Northern Extra Tropics GISS & RSS show +1.0ºC while UAH shows only +0.5ºC, this divergence occurring prior to 2000. In the Southern Extra Tropics, GISS shows least warming +0.2ºC, with the divergence (quite wobbly) from UAH’s +0.38ºC prior to 1996 and divergence from RSS’s +0.55ºC prior to 2007.

    While we might expect some divergence as SAT & TLT are measuring different things, it is the timing of the divergences that surely suggest there area calibration issues with TLT data.

  8. 158
    MA Rodger says:

    Kevin McKinney @155,
    Further to #157, I find now our friends at Google are determined that GoogleSites must be accessed through their front door and not have images displayed through the windows.
    So I have uploaded the graphic of GISS SAT, UAH and RSS TLT trends by Hemisphere onto a less cluttered webpage. See Graph 2 down THIS LINK.

  9. 159

    #157, MAR–

    Sorry, still ‘not permitted’–as you seem to be troubleshooting the issue, the exact error message was:

    403. That’s an error.

    Your client does not have permission to get URL [redacted] from this server. (Client IP address: [redacted])

    Forbidden That’s all we know.

    I don’t suppose it would be as simple as setting the permissions to share with everyone?

    On the analysis, let me summarize some of your comments, as the discussion can be tricky to follow.

    NOAA SAT Ocean/UAH TLT Ocean: 0.45 C
    GISS/UAH SH: 0.4 C
    NOAA Land/GISS NH: 1.0 C
    UAH Land/UAH NH: 0.6 C

    (These are all for the period 1979-2019.)

    So the takeaway here is that the divergence between SAT and TLT measurements are indeed associated mostly with the land/NH components, while ocean/SH are relatively closer in value.

    While I’m not feeling so ambitious just now as to download all the data you consider, I do note that a WFT comparison on the land side supports this point:


    CRU slope = 0.0279975 per year
    RSS slope = 0.0263645 per year
    UAH slope = 0.0136027 per year

    Essentially, UAH shows about half the warming as either of the other two measures.

    I won’t address the more detailed comparisons, other than to agree that calibration issues probably still exist; but then, that’s rather been the satellite story all along. A question though: have you considered whether baseline is relevant to your analysis? Baseline doesn’t affect trend, but it can affect which part of which curve appears divergent.

    Case in point, perhaps–here’s a comparison of the same three data sets as in my above WFT graph–that is, the land components for UAH TLT, RSS TLT and CRU:

    The two satellite measures are aligned to rough parity at the beginning of the span. It’s far from definitive, I think, but it appears to me that the UAH response is ‘muted’, with regard to both trend and short-term variability. (Note the relative amplitudes of the last two ENSO excursions.) Perhaps there’s more in play than simply calibration?

    Here’s another view of the same data, with all three datasets aligned to ‘cross’ trends at about 2009; fortuitously, this also allows pretty good intercomparison of those ENSO ‘wobbles’. One can see that there’s pretty good agreement on amplitude between CRU and RSS, but UAH not so much.

  10. 160
    nigelj says:

    MAR & KM, my understanding is RSS updated its temperature record a couple of years ago by making corrections for satellite drift etc, etc and is now similar to NASA GISS and all the others, but UAH didn’t do this and is now an outlier. Has Roy Spencer given any reason why he hasn’t done the same as RSS? Or do you know of any reasons? The man is a lukewarmer but I’m trying to resist the temptation to put it down to this.

  11. 161
    MA Rodger says:

    Kevin McKinney @159,
    Regarding the timing of divergences, I think the wobbles of the rolling annual averages are a distraction and would suggest 5-year averages provide a clearer view.
    @157 I described (perhaps confusingly backwards through time) the Northern Hemisphere divergences from GISS:-

    “In the GISS/UAH/RSS NH comparison, the three are in reasonable agreement since 2013 before which the TLT both diverge from GISS with the UAH/TLT divergence all prior to 2005,”

    The WFT engine allows a CRUTem4/RSS(land)/UAH(land) comparison which also shows the three “in reasonable agreement” 2013-21, at which point the “TLT both diverge” from CRUTem4 (the blue trace), remaining in step 2005-13 at which point they diverge from each other. The WFT graphic thus provides pretty-much a carbon-copy of the NH traces in GRAPH2 linked @158 (if that is now ‘zulässig’).
    But perhaps the best way of demonstrating the TLT calibration problems is to use this WFT graphic which shows RSS(RED) plotted against three offset plots of UAH. It shows RSS & UAH are a brilliant fit but only if you can ignore the periods 1987-95 and 2001-03 where they manage to diverge first by 0.15ºC and then by a further 0.09ºC which for a 4-decade record adds up to a difference in trends of 0.06ºC/decade. (The SKS Trend Calculator puts the global difference in linear trend between UAH & RSS at +0.079ºC.)

  12. 162

    #158, MAR–

    Interesting graph comparing RSS, UAH & GISS. What surprised me was to see the close consilience between RSS (southern hemisphere) and UAH (northern hemisphere)! My candid reaction, after years of living in the South, was “You know *that* ain’t right!” Or more soberly, “got to be coincidence!”

    But given how robust the relative hemispheric differences are across datasets, it dramatizes once again how much of an outlier UAH really is.

  13. 163

    #160, nigel–

    You’re close. Actually, UAH has done a revision relatively recently–2015–but its effect was to *decrease* the warming trend further.

    More here:

  14. 164
    MA Rodger says:

    nigelj @160,
    Actually UAH has indeed revised its TLT record, from v5.6 to v6.0 in 2015. The result was a reduction in the warming trend, thus the reverse effect of the RSS revision from v3.3 to v4.0 which showed increased warming.
    Globally UAHv5.6 (which runs 12/1978-to-7/2017) was actually a pretty good match for HadCRUT4, this with the exception of the first decade of UAH. So while a calculated trend over the full period gives HadCRUT at +0.175ºC while UAHv5.6 shows +0.155ºC, from say 1988-on, the trends become very similar (1/1988-to-7/2017 HadCRUT shows +0.179ºC while UAHv5.6 shows +0.182ºC).

  15. 165

    MAR, #161–

    But perhaps the best way of demonstrating the TLT calibration problems is to use this WFT graphic…

    Yeah, that’s pretty compelling.

    (Here it is again for convenience: )

  16. 166
    MA Rodger says:

    Kevin McKinney @165,
    Mind that this “compelling” outcome doesn’t advance our understanding but only reinforces my message back up-thread @138.

    “There is a view that RSSv4 presents a better measure of TLT than UAHv6 because its global values show a trend close to the SAT global trend. But I think we should be cautious with satellite TLT data. … if you start looking at RSSv4 numbers at a NH/SH level (or some other sub-global level), the fit with SAT becomes less ‘similar’.”

    We see what looks for all the world like a calibration problem but, importantly, we don’t know if it is RSSv4 or UAH.6 or both that has the problem.

    I can perhaps add to that “compelling” outcome but still without being able to show where any calibaration problems lie.
    Delving into my spreadsheet calculations, the differences in rate-of-warming for NH & SH for RSS & UAH have been quite close since 1995 (indeed very close for 1995-96 and 2007-on). So why are they not “close” before 2005? Somehow, in RSS the difference in NH/SH warming 1979-95 is 0.17ºC and roughly equals the NH/SH difference for 1995-2019 (so 0.34ºC over the full RSS record to 2019). While in UAH (which shows the same difference to RSS 1995-2019) this difference 1979-95 is zero, both NH & SH warming at roughly the same rate through the period (so only 0.17ºC over the full UAH record to 2019).
    I would perhaps be saying this lack of NH/SH differential warming (through a period of evident SAT warming complete with an increasing NH/SH SAT differential) was indicative of calibration problems within UAH, except this period is also one of strong volcanic activity which may well be significant.
    So I’m not happy to say this is indicative of UAH.6 calibration problems.

    But I note today that WFT give an RSS Land record which I haven’t spotted before. The WFT graphs show a good fit between CRUTem4 & RSS[Land] and putting the RSS[Land] numbers into my spreadsheets, take off +0.025ºC/decade of warming and the NOAA[Land]/RSS[Land] fit is very very similar to the NOAA[Ocean]/UAH[Ocean] fit with a divergence of +0.08ºC ‘peaking’ in the late 1990s. (The divergence is +0.10ºC in the NOAA[O]/UAH[O] comparison.)
    Given that is the sort of thing I would expect if there was no calibration problems, thinks, maybe something significant could be dug out from that.

Leave a Reply

Comment policy. Please note that if your comment repeats a point you have already made, or is abusive, or is the nth comment you have posted in a very short amount of time, please reflect on the whether you are using your time online to maximum efficiency. Thanks.