Hector temperature values (plus GCAM results!)

[rplzzz]

Good news, everyone! I've got Hector running with GCAM. Here are some plots from the GCAM data viewer. The red line in each is Hector; the blue line is MAGICC. Note also that MAGICC starts ingesting GCAM emissions after 2005, while Hector starts after 2010.

Now, for the question promised by the issue tag. What's the story with the temperatures? They look awfully high. For example, in 2000 the temperature anomaly is nearly 1.25 C. I pulled the outputstream from a stand-alone run and found the same result:

2000,rcp45,0,temperature,Tgav,1.17862,degC
2000,rcp45,0,temperature,Tgaveq,1.39939,degC

(sub-question: what is the difference between T and Teq?)

I just looked back at figure 4 from the paper, and it seems we see it there too. We track pretty well through sometime shortly after 1950, and then we sort of take off abruptly. Do we have any idea what's causing that? Everything else looks good, but that temperature result is bound to generate some push-back from GCAM users.

[bpbond]

Wow, that integration was fast work--nice job, that's great. Corinne and I will take a look at the temperature issue and come visit early next week.

[cahartin]

awesome! That is exciting. Will take a look at temperature. Thanks!

[rplzzz]

Thanks. I credit the ease of getting it up and running to hector's high-quality design. Virtually everything just worked on the first try. I'm pretty excited about it. I think it's going to enable a lot of things we've been wanting to do with GCAM that we just can't easily do today.

[rplzzz]

For what it's worth, the corner in the temperature seems to be caused by a leveling off in SO₂. Here is a plot of the total forcing, SO₂ forcing (direct+indirect), CO₂ forcing, and forcing ex. SO₂.

So, you can see that for much of the century CO₂ and SO₂ forcing more or less offset each other. Then, right around 1970 SO₂ emissions plateau, and the associated forcing abruptly flattens. This produces an apparent corner in the total forcing. If you back out the SO₂, the residual curve looks quite smooth.

This all seems pretty reasonable. In fact, it's fair to wonder why the other models don't see a similar effect. Presumably they include some sort of offsetting effect that we're missing. It's worth noting that the HADCRUT temperature series doesn't show the abrupt takeoff, so that would argue that we need to track down the missing component and add it.

However, there is another thing to consider. Looking back at the plots I posted earlier, MAGICC and Hector have nearly the same forcing between 2020 and 2040. However, the temperatures produced by the two models are quite different over that period. MAGICC's temperature equation is pretty much the same as ours, so either GCAM has MAGICC configured to use a different feedback, or MAGICC and Hector are getting substantially different ocean heat flux. I'm trying to track down where MAGICC gets its feedback factor (it doesn't seem to be in any of the files in GCAM's input directory). Maybe if someone else has time they can look into the flux issue.

[cahartin]

that's interesting about the SO2. I wonder if there are multiple radiative efficiency values for calculating the radiative forcing of SO2?

Good point that we match MAGICC in RF but our temperatures are higher. Ocean heat flux is an easy parameter to change and would certainly affect the temperature response. I'll take a look at that.

[rplzzz]

I tracked down the climate sensitivity in MAGICC (it is under the somewhat obscure name DT2XUSER), and it's the same 3.0 that we are using, so ocean heat seems like the way forward. What is the scientifically plausible range of values for the heat flux parameter?

[cahartin]

the flux parameter can go up to about 1. So we have a lot of room to work with.

[rplzzz]

Sounds good. I will experiment with the GCAM version and let you know what I come up with.

[bpbond]

Ah yes, good old DTXUSER. I'm happy to help with this next week, if I can.

[rplzzz]

Changing the ocean heat flux doesn't really get us where we want to go. Here are results for a few values of the heat flux parameter. The key for the line colors is

red : hector, heatflux=0.2 blue: hector, heatflux=0.3 green: hector, heatflux=0.4 yellow: hector, heatflux=0.7 orange: MAGICC

The summary is that the effect of the heat flux parameter is stronger when the temperature anomaly is higher, so raising it affects values late century much more than early century. As a result, if you crank the heat flux down enough to get agreement in the early part of the century, Hector's results are well below MAGICC's at the end of the century. I don't know that that's a great tradeoff, since Hector and MAGICC both seem to be at the low end of the model range in the 22nd century and beyond.

We should probably look at what happens in policy scenarios, and then we might want to convene a panel of GCAM users to see what they think.

[bpbond]

So per Corinne's email earlier today, it seems likely that the O3 component is causing the big problem here?

[bpbond]

Closing this, as it's fixed AFAIK.