How Long Will the Hiatus Period Last?

[skierinvermont]

1. Okay. So we agree that models substantially overestimated the warming since 1993, and this was documented correctly in the Fyfe et al. paper. The paper lists some possible explanations, but notes that ultimately, we have to wait and see with additional temperature observations. The question is why this discrepancy occurred. I (along with many others here) would argue that it's due to an incorrect response to a change in radiative forcing (ie the models are too sensitive to a change in RF).

 

2. There is no signal with the PDO/AMO in the model mean. Since this oscillation is relatively predictable on a 30 year timeframe, most of the models should be able to accurately reproduce this modulation, and thus the model mean should roughly reflect this. But we see a exponential increase in the rate of warming in the models.. not a multidecadal shift every 30 years. 

 

This is probably one of the reasons why we continue to fall out of the range of modeled predictions fairly rapidly.

 

 

3. Models generally do not have these hiatus periods lasting more than 15 years. Since the PDO/AMO is an oscillation that modulates the centennial warming on a multidecadal basis, they are not correctly simulating these oscillations, which may partially explain why they cannot properly simulate Antarctic Sea Ice increases and underestimate Arctic Sea Ice decreases.

 

From the NOAA State of the Climate Report in 2008:

 

"The simulations rule out (at the 95% level) zero trends for intervals of 15 yr or more, suggesting that an observed absence of warming of this duration is needed to create a discrepancy with the expected present-day warming rate.”

 

My point still stands though. Since the PDO/AMO is a predictable oscillation, the models should reflect this in their simulation, but there is no evidence of this. In addition, the horseshoe of cool water is missing in the spatial pattern of warming in the modeled simulations, which is a signature of the transition from the +PDO to the -PDO. This is further evidence that the models do not accurately reproduce the PDO well in their hindcasts.

 

 

 

2. Absolutely false and your response demonstrates your oversimplified incomplete understanding of both internal variability and climate modelling. First of all, when the models are designed to integrate with current internal variability they are able to replicate slower warming at present and in the near future. So on its face, your statement is false. When internal variability data is integrated, even the model mean correctly shows slow warming at present. 

 

Second, there isn't a whole lot of evidence that the PDO or AMO occur on exactly a 30 year timescale. The cycles tend to be of different lengths. So even from a purely physical perspective, you can't be sure when the cycles begin or end. Of course, the models aren't designed to integrate such simple statistical statements as "PDO cycles occur every 30 years." They are designed around physics. And the physical understanding of the PDO and AMO are both weak. So it's not surprising that models cannot predict the timing of the phase changes.

 

 

However, the models do show PDO and AMO variability. This is one of the reasons why they show long periods of more rapid warming and of slower warming. There is a large subsection dedicated to this. 

 

3. Another basic logical error. Just because the models do not show cooling periods longer than 15 years (actually very rarely they do), does NOT mean they do not sure -PDO periods. They do show -PDO periods. You can read all about it in the AR5. 

 

 

And finally, the large graphic you posted has clearly been manipulated to show the largest possible discrepancy. As far as I am aware, the 30 year trends are in relatively decent agreement with models. How about providing us with data not created on Spencer's blog?

[skierinvermont]

Models are typically initialized in 1850 which is why they are unable to predict a change in the PDO in exactly 2007 (or 1998). However, models initialized in the 1990s predict a phase change in the PDO around 2000 and subsequent lower temperatures by about .05 to .1C. 

[ORH_wxman]

I think a more interesting question is how long the models and obs keep diverging from eachother versus whether the trend line remains negative or positive. I do not think the trend line will keep a negative value from 2001 by 2020, however, if the warming is very slow, then it still is diverging from models which forecast an increased rate of warming. That is the bigger area of interest to me. The models are failing at a time when they should be most correct (i.e. natural variation should be the least amount of influence versus any time in the past). CO2 forcing is more dominant than any other period in the observation period, yet the biggest discrepancy is occuring in the past decade.

[chubbs]

I also think the hiatus will last into the 2030s based on the length of cooling cycles that ended around 1910 and 1975. However this period is different than previous cooling cycles in that temperatures are steady or slowly increasing instead of actually declining.  Likewise I expect the warming after this hiatus ends to be more rapid than the 1975 - 2005 warming. 

[ORH_wxman]

It will be a good test to see how well the new Met Office 5 year forecasts work out.

 

fcst_global_t4.png

 

 

Their forecast looks a lot more realistic than some others....though a 5 year forecast really isn't that risky...since they aren't initializing it from 1990 or 2000 or 2005 like some of the various CMIP5 model studies.

[chubbs]

Yeah, I think that the next +PDO era this century could see a rapid .6 to .8C rise in about 25 years compared to .6C from 77-02

and .4C from around 1915 to 1940. If I had to guess now I would say about 1.3C of additional  warming by 2100.

 

We are in general agreement. Need to recognize the uncertainty however. Recent cycles could completely breakdown if warming continues..

[tacoman25]

Well then he is directly contradicting himself. He specifically states that the observed warming is due to UHI. 

 

Sounds like a hack trying to make some money. 

 

Looking at the greater context of what he has said, and not just some little snippet you cherry-picked, it sounds like Spencer thinks some of the NH warming trend is due to UHI. He also stated that he believes humans could be responsible for anywhere from 10-90% of warming...so clearly, he is not attributing observed warming just to UHI.

[ORH_wxman]

 

Richard Betts from the Met Office had a great explanation of how the new model works compared to the old one.

 

 

Why are forecasts from the computationally-intensive HadGEM3 so much lower than the corresponding forecasts from HadGEM2?

The difficulty is that the question itself is not well-posed, because the HadGEM3 and HadGEM2 ensemble simulations are not “corresponding forecasts”.

The HadGEM2 simulations were started from a climatological mean state with near ‘pre-industrial’ (actually 1860) conditions, and then run through the late 19th & 20th Centuries and first few years of the 21st Century with observed changes in greenhouse gas concentrations, aerosols and land use, with the model simulating its own natural internal variability. Since internal variability is unpredictable after a few years, we don’t expect such simulations to capture the specific timing periods of relatively faster or slower warming. This is what I mean by an “uninitialised” simulation – the initial conditions are not for a specific date and/or are so far back in the past that they have no effect on the internal variability simulated by the model.

In contrast, the HadGEM3 5-year simulations were an intialised forecast, initialised with observational data in late 2012.

There are 2 reasons why this difference makes it not useful to compare the 2 simulations over a 5-year period of specific dates (in this case 2012-2017):

1. For any particular 5-year period, HadGEM2 doesn’t ‘know’ what climatic state it should be starting at (eg: 2012 in this case). It’s got it’s own simulated version of the 2012 climate, but we’d never expect that to correspond to the real 2012, due to chaos. Since the mean of the HadGEM2 runs does not include the ‘hiatus’ of the 2000s, the model’s 2012 is warmer than the observed 2012. In contrast, HadGEM3 does ‘know’ the real observed climate of 2012, so it starts at a cooler state than the equivalent portion of the HadGEM2 run.

2. The runs in the HadGEM3 ensemble are all initialised with observational data from 2012 so are simulating internal variability starting from the same conditions. If there is any signal in the internal variability, the different runs should agree to some extent. In contrast, the runs in the HadGEM2 ensemble all have their own individual versions of 2012 and hence their subsequent variability after 2012 will probably be very different. Use of the ensemble mean tends to cancel these out to just give the GHG-forced warming trend over the 5-year period.

So, with HadGEM2 being uninitialised, the ensemble mean starts at a warmer 2012 than in observations and then merely follows a GHG-forced warming trend until 2017, with internal variability in the different runs cancelled out. However HadGEM3 starts at a cooler 2012, as observed, and is able to capture internal variability better.

Hence it’s not at all surprising that 2017 in HadGEM2 started in 1860 is noticeably warmer than in HadGEM3 initialised in 2012.

5-year forecasts could well be dominated by internal variability so don’t particularly tell us anything about the long-term trend. We’ve not yet done future projections with HadGEM3 so don’t know how much warming it will project in the long term.

Incidentally, HadGEM2 was not the only model submitted to CMIP5 / IPCC AR5 by the Met Office. We also submitted initialised forecasts with HadCM3 – see here.

Hope this clarifies my tweet!

Cheers

Richard

 

 

Hopefully Hadgem3 does better overall further into the future than Hadgem2 did. Unless this past 10-15 years is like a 2-3 sigma natural variability event in the cold direction, then hadgem2 obviously over-estimated the GHG trend in relation to natural variability.

[Jonger]

Well then he is directly contradicting himself. He specifically states that the observed warming is due to UHI.

Sounds like a hack trying to make some money.

Probably more than a quarter or it is.

[Sundog]

Probably more than a quarter or it is.

 

False. 

[Jonger]

Probably more than a quarter or it is.

 

False. 

Oh really? So what test planet did you use to come to this conclusion?

[tacoman25]

I don't see the difference.

 

He thinks future warming will be minimal. He thinks human contribution so far has been minimal. He thinks there is a liberal conspiracy to stifle those that disagree such as himself. He's also made statements suggesting no warming has occurred at all. But at other times, he has contradicted those statements.

 

Saying "Oh maybe people have contributed .1C" doesn't mean that he acknowledges AGW. That's like a dumb cop-out to pretend he's being objective.

 

1. It depends on how you define "minimal". My guess from looking at larger context is that he means minimal in relation to current projections.

 

2. If he believes there are efforts to stifle those that disagree from the mainstream, that doesn't make him a denier of AGW.

 

3. From what I've seen, overall he believes there's a lot more uncertainty as to the amount of warming due to humans than most scientists. Sorry, but that does not make him a denier as you claimed.

[ORH_wxman]

Oh really? So what test planet did you use to come to this conclusion?

 

 

There are no studies that support UHI being more than 25% of the warming in the peer reviewed temperature datasets. In the raw data for certain areas...sure. But that has been adjusted for.

 

If there's a big nitpick in the surface data, you would be better served scrutinizing the TOBS adjustments which in more recent years have strayed away from the peer-reviewed values expected post-1985. But even this wouldn't account for more than about 15-20% of the temperature trend at most.

 

 

Otherwise you are just spouting off conspiricy junk with no scientific evidence to back it up.

[tacoman25]

I also think the hiatus will last into the 2030s based on the length of cooling cycles that ended around 1910 and 1975. However this period is different than previous cooling cycles in that temperatures are steady or slowly increasing instead of actually declining.  Likewise I expect the warming after this hiatus ends to be more rapid than the 1975 - 2005 warming. 

 

Well part of the reason this -PDO period is not seeing a real decline in temps (although the 1946-75 period was only a slight decline) could be because there were more aerosols in the air during the last -PDO period.

 

For some reason, the same people who have trumped up aerosols as the main cause behind global temp fluctuations since 1950 (and not oceanic phases) ignore that logic when talking about how with this -PDO phase temps are still slowly warming or flat instead of declining.

[tacoman25]

We are in general agreement. Need to recognize the uncertainty however. Recent cycles could completely breakdown if warming continues..

 

That's what they were saying in the 1990s. There was a lot of speculation that AGW was causing more El Ninos. Oops.

[Jonger]

Oh really? So what test planet did you use to come to this conclusion?

There are no studies that support UHI being more than 25% of the warming in the peer reviewed temperature datasets. In the raw data for certain areas...sure. But that has been adjusted for.

If there's a big nitpick in the surface data, you would be better served scrutinizing the TOBS adjustments which in more recent years have strayed away from the peer-reviewed values expected post-1985. But even this wouldn't account for more than about 15-20% of the temperature trend at most.

Otherwise you are just spouting off conspiricy junk with no scientific evidence to back it up.

In a science where tenths of a degree are debated, its quite difficult to accept that the easily observable heat islands wouldn't contribute a couple tenths to global temp trends. Most long term measurements originate in areas that were most prone to surrounding growth.

I'm on my phone so my citing ability is difficult, but the adjustments I have seen to compensate for UHI appear quite paltry.

[bluewave]

Hopefully Hadgem3 does better overall further into the future than Hadgem2 did. Unless this past 10-15 years is like a 2-3 sigma natural variability event in the cold direction, then hadgem2 obviously over-estimated the GHG trend in relation to natural variability.

 

Maybe when they do a long range run of the HadGEM3, it will show the smaller amount of warming like the NCAR CCS.

 

http://www.sciencedaily.com/releases/2011/09/110918144941.htm

 

To track where the heat was going, Meehl and colleagues used a powerful software tool known as the Community Climate System Model, which was developed by scientists at NCAR and the Department of Energy with colleagues at other organizations. Using the model's ability to portray complex interactions between the atmosphere, land, oceans, and sea ice, they performed five simulations of global temperatures.

The simulations, which were based on projections of future greenhouse gas emissions from human activities, indicated that temperatures would rise by several degrees during this century. But each simulation also showed periods in which temperatures would stabilize for about a decade before climbing again. For example, one simulation showed the global average rising by about 2.5 degrees Fahrenheit (1.4 degrees Celsius) between 2000 and 2100, but with two decade-long hiatus periods during the century.

[ORH_wxman]

In a science where tenths of a degree are debated, its quite difficult to accept that the easily observable heat islands wouldn't contribute a couple tenths to global temp trends. Most long term measurements originate in areas that were most prone to surrounding growth. I'm on my phone so my citing ability is difficult, but the adjustments I have seen to compensate for UHI appear quite paltry.

 

 

They do contribute, and its been adjusted for. You have to keep in mind that there are reverse adjustments too. Many sites moved from downtown to airports outside of the city which made for cooler readings.

 

The adjustments are not going to be perfect and they will have to adjust in the future too, however, there is very little evidence that the adjustments are so bad that you would have an error of over 20%. Probably even 10%.

 

TOBS is more questionable than UHI adjustments...but even those are not a large portion of the trend.

 

 

If the surface trends were so erroneous, then they wouldn't be close to the satellite trends since 1979 when they went online.  

[Jonger]

They do contribute, and its been adjusted for. You have to keep in mind that there are reverse adjustments too. Many sites moved from downtown to airports outside of the city which made for cooler readings.

 

The adjustments are not going to be perfect and they will have to adjust in the future too, however, there is very little evidence that the adjustments are so bad that you would have an error of over 20%. Probably even 10%.

 

TOBS is more questionable than UHI adjustments...but even those are not a large portion of the trend.

 

 

If the surface trends were so erroneous, then they wouldn't be close to the satellite trends since 1979 when they went online.  

 

You're right, these adjustments far dwarf any issue UHI could add....

 

[Snow_Miser]

Well then he is directly contradicting himself. He specifically states that the observed warming is due to UHI. 

 

Sounds like a hack trying to make some money. 

 

I guess anyone who disagrees with you then is a hack trying to make money. Good grief.

[Snow_Miser]

1. It depends on how you define "minimal". My guess from looking at larger context is that he means minimal in relation to current projections.

 

2. If he believes there are efforts to stifle those that disagree from the mainstream, that doesn't make him a denier of AGW.

 

3. From what I've seen, overall he believes there's a lot more uncertainty as to the amount of warming due to humans than most scientists. Sorry, but that does not make him a denier as you claimed.

 

For the record, here are Roy Spencer's future predictions relative to other predictions.

 

[tacoman25]

For the record, here are Roy Spencer's future predictions relative to other predictions.

 

 

 

Wait, so a "denier of AGW" still predicted future warming?? That's weird.

[Snow_Miser]

It will be a good test to see how well the new Met Office 5 year forecasts work out.

 

fcst_global_t4.png

 

I don't think we should have much confidence in the UKMO decadal predictions, after their previous two predictions in 2010 and 2011.

 

[tacoman25]

That's a valid point, bluewave. 

 

The 2012 forecast calls for more of the same. Friv and others are wanting/expecting something more like the 2011 forecast from here on out. HEAT EXPLOSION FROM THE OCEAN IMMINENT!!!

[skierinvermont]

Probably more than a quarter or it is.

 

False. This has been rehashed a million times on this forum and in academic journals. UHI is accounted for in the observational records. BEST (done by skeptics) is the latest study to confirm this fact.

[skierinvermont]

Lol @ citing Spencer and Braswell. The paper contained so many egregious errors the editor of the journal resigned and apologized for its publication. 

 

Predicting .5C of warming by 2100 is a denier. It is complete denial of the physics of GHGs and the evidence that feedbacks are at the very minimum slightly positive.

[PhillipS]

Is there agreement on what will constitute an end to the 'hiatus'?  It may have been discussed earlier in this thread but I didn't see anything in a quick review.

[Snow_Miser]

Is there agreement on what will constitute an end to the 'hiatus'?  It may have been discussed earlier in this thread but I didn't see anything in a quick review.

 

I'm going with the 2030s. This is roughly consistent with both the Li et al. paper and the Curry/Wyatt paper. That would be around the time the PDO flips to it's positive mode. Anything sooner than that is not possible IMO unless the PDO flips to it's positive phase early. 

[skierinvermont]

Is there agreement on what will constitute an end to the 'hiatus'?  It may have been discussed earlier in this thread but I didn't see anything in a quick review.

 

I would define it as any time a 15 year trend of +.15/decade or greater appears. Or a 10 year trend of +.25 or greater. Even if the trend were Nina to Nino, it would probably indicate substantial warming is occurring (faster than the current rate). Right now even starting in 1999 the trend is not .15C/decade.

[chubbs]

Well part of the reason this -PDO period is not seeing a real decline in temps (although the 1946-75 period was only a slight decline) could be because there were more aerosols in the air during the last -PDO period.

 

For some reason, the same people who have trumped up aerosols as the main cause behind global temp fluctuations since 1950 (and not oceanic phases) ignore that logic when talking about how with this -PDO phase temps are still slowly warming or flat instead of declining.

Well wether it is GHG alone or GHG+aerosal this "cooling" cycle is underperforming compared to the past 2.