I've been mulling something of late, and it hasn't left me in a tremendously good mood.

Take a look at these two sets of graphs:

The first one is from the US Energy Information Administration, a group within the US Department of Energy tasked with coming up with independent statistics and analysis on US and world energy use. This chart is from the "International Energy Outlook 2011" report, released last September. It shows the breakdown of fuels used to generate electricity, given fairly conservative projections of growth and changing energy mix.

It shows that, by 2025 -- a little over 10 years from now -- coal will provide 10,200 terawatt-hours (TWh) out of a total of 28,700 TWh produced around the world, annually. By 2035, it's up to 12,900 TWh out of 35,200 TWh.

The second graph is from an article by David Roberts in Grist last year, "The Brutal Logic of Climate Change." Based on work done by leading energy/climate researcher Kevin Anderson (former head of the UK's Tyndall Energy Program), it shows how soon we as a planet need to start reducing carbon emissions, and how rapidly they need to decline given different "peak emissions" points. That's to avoid a 2 degree C increase in global temperatures, now understood to be a potentially catastrophic level of warming.

Here, we see that if we have peak emissions of around 65 gigatons of CO₂ equivalent in 2025, we have to be down to under 20 GtCO₂e by roughly 2035, and to zero GtCO₂ shortly thereafter. In energy terms, we'd have to go from this:

to this:

Basically, we have to replace over 21,000 TWh of electricity generation from coal and natural gas (yes, natural gas is less-harmful than coal, but still has a greenhouse impact) with an equivalent amount from some mix of renewable, hydro, and nuclear. And do it in 10 years.

Except it will have to be more than that, at least another 15,000 TWh more, because we'll have to replace all of the gasoline and diesel-powered vehicles on the roads around the world with alternative forms of transportation, all of which has to be electric (or human/animal-powered). And also add however much new power is required to run the various production lines day and night to make all of the needed photovoltaics, wind turbines, electric buses, and such.

For comparison, the world added... 15 TWh in solar in 2010.

Set aside issues of politics and economics, and simply look at raw logistics: is it even possible to undertake that kind of shift in 10 years?

As the second set of graphs above suggests, if we start before a 2025 peak, we'll have somewhat less carbon-based energy production we'd have to replace, and somewhat more time in which to do it. Not much, though -- even peaking in 2015 only pushes the deadline(!) out to 2050, if we're lucky (the red & blue lines in the graphs show alternative scenarios from the IPCC, none of which are very pleasant).

But given the current global political environment, it's difficult to imagine a real agreement to eliminate carbon emissions, taken seriously by all parties, showing up before the end of this decade.

So here are our three scenarios:

1) We manage to get a real global agreement in place within the next five-eight years, and spend the subsequent 25 or so years undertaking the largest industrial transformation imaginable. Politically implausible.

2) We don't get a real global agreement in place before 2025, and have to cut emissions by 10% per year (as Roberts notes, the biggest drop we've seen is 5% after the USSR's economy collapsed). Physically implausible.

3) Neither of those happen, and we start to see truly awful impacts, mostly in the developing world at first, all of which make the world politically more hostile and economically more fragile -- and make it more difficult to cut carbon emissions effectively.

This is why I think geoengineering is going to happen. Desperate people do desperate things, and when you hear sober scientists say things like population "carrying capacity estimates [are] below 1 billion people" in a world of 4 degree warming, it's hard to argue convincingly that the uncertainty and risks around geoengineering are worse.

Anyone who thinks that geoengineering is a way to avoid cutting carbon is an idiot. Geoengineering is a tourniquet, a desperate measure to stop the bleeding when nothing else can work in time. If Anderson's analysis is accurate (and, if anything, it may be optimistic), it's hard to see how we can avoid taking these desperate measures.