Selection of 2014 models to serve on the Commons

I still believe it's important to have representation of the RCP2.6
futures in our set of models. Here are three reasons.

First, we are not capable of predicting the human side of the future out
to the year 2099. If instead of forward, you were to go back 85 years to
the year 1929, and imagine life in 2014, I don't think you'd come up
with GPS satellites and smartphones. You might also have trouble
imagining the history of totalitarian regimes that would soon unfold
(mind you, this is standing prior to October '29!). We simply do not
know what is going to happen over the century in terms of technological
developments, social change, shifts in governance patterns, and
financial and economic crises. This is not to say I think RCP2.6 is
likely -- I don't, and the scenarios I can conjure up to lead to it
aren't happy ones -- but the IPCC felt it was plausible enough to
include it in AR5 and came up with from the literature 20 different
scenarios that might lead to that outcome.

The second reason is a tactical one. The dimension of RCP choice is much
easier to explain to people than the differences between individual
models. The latter involves highly technical distinctions of model
physics and parameterization, the former just different emissions
trajectories, which I think is easier for people to grasp. The point is
this - if you include an RCP2.6 future in your set of models, even (or
especially) as you explain it's not very likely, your users gain a sense
of agency when they _choose not to select it_. They may quickly look at
the RCP2.6 model outputs for grins and giggles, but in so doing they
also gain a better understanding how emissions levels relate to
projected climate. Allowing our users to select from the full range of
RCPs engages them - in a sense it moves them towards co-collaboration.
Conversely, emphasizing model choice at the expense of RCP diversity
throws away an opportunity for science communication and education
because it highlights the dimension that isn't very clear and looks like
a confused muddle (model choice) instead of the one that gives a clear
portrait (different emissions levels). Communication of the science
behind climate change is difficult enough already, as we well know.

Finally, we don't know what the decision spaces of the eventual users of
the data will be. In particular, if the data get used by people coming
from a mitigation perspective rather an adaptation angle, these people
may be quite interested in illustrating the most optimistic future as
they advocate for their own policy agendas. Perhaps this person is an
environmental advocate pushing for energy deintensification, or maybe
the person is a transportation planner lobbying for improved bicycle
infrastructure for mitigation purposes who wants to display the
difference between the most optimistic trajectory and the one we're on
as they lobby for their cause. In short, our user base may end up being
quite diverse, and some of these people will want to see the full RCP
spectrum.

Best,

Allan

I agree with Allan’s comments here. Although we are on a particular emissions trajectory, many plausible futures could unfold in which we get on a new one.

Just using the Bay Area models as an example (the interpretation of these models might change as you go to the state level), the GFDl b1 model does get you into the slightly warmer/drier future. But dropping the RCP26 models means you don’t have representation in the slightly warmer (1C) and wetter (25%) space. CSIRO a1b is close but is over twice as warm.

So if you wanted to keep the number of models you are serving equal to what’s in the document but include the RCP26 examples, I might choose only one of the MPI45 and GFDLb1. These have about equivalent warming (within 0.5C of each other) and only different in direction of precip (+- 10% from historic). But +- 10% from historic may not really be that big of an absolute difference in precip in the models.

That’s my 2 cents.

Cheers,
Sam

Thanks everyone.

Due to the diversity of uses and the difficulties associated with trying to guess what people will need, we've decided to serve all 18 futures.