Forests vs. solar: which delivers more carbon benefit?

As solar booms and an increasing variety of land is in its crosshairs, it’s worth doing quick thinking and quick math to compare the two. Spoiler: both are awesome, and they aren’t really in conflict.

A friend recently asked me: “What’s better for fighting climate change: cutting down a stretch of forest to build a solar plant, or leaving that forest to do its thing?”

Quick takeaways:

Solely in net carbon terms, it’s worth cutting down trees to install solar plants in most plausible U.S. scenarios. Fortunately, we have a lot of land in forest relative to our land-related solar energy needs, and we can surely be a bit picky about which land we use.

Forests sequester and store carbon in trees, other plants, and soils. Solar is emerging as a primary alternative to burning coal and natural gas, major sources of greenhouse gas emissions. From the perspective of fighting the climate emergency, which is more important?

There are three essential elements with which to frame this seemingly simple but surprisingly multifaceted question, as I see it:

• Forests are important for many things – carbon sequestration is just one of them. I’m looking at this question here solely through the lenses of energy and carbon.
• To generate all of the energy consumed in the U.S. in a year with solar would require roughly 0.5-1.0% of all land. Perhaps quite a bit less land. And about 34% of U.S. territory is in forest land (USDA 2014).
• Therefore, we will never have to eliminate much of our forest for the sake of solar.

Let’s do a little bit of math

Beyond that framing, we have some quick math to do. How much CO2 you avoid with solar energy depends on several quantitative assumptions:

• the energy density, or land efficiency, of a solar plant (measured in acres or hectares per MW)
• the baseline carbon intensity that is your point of comparison, such as the surrounding electric grid (measured in pounds of CO2-equivalent per MWh)
• the amount of sun that falls on the land under consideration (or the “capacity factor” of solar in a given geography)

I’m largely skipping other considerations, such as the relevant details of solar cost structure, even though those elements – such as the energy density of individual panels or the emergence of bifacial panels – are steadily improving.

Even with those simplifications, we can combine those ranges

And then the extent to which a forest sequesters carbon depends on which forest you’re considering. According to the USDA’s Carbon Storage and Accumulation in United States Forest Ecosystems (1992):

Another way of saying this: the stock of carbon in forests is roughly 120,000 to 210,000 pounds per acre. The associated flows are more complicated: forests at different stages grow at different rates, and plantation silviculture (i.e., tree farms) may appear to sequester carbon quickly at certain times, even if they never become high-biomass natural forests. But let’s just consider the stocks for now.

So now we have a comparison to make. I’ve combined these apples and oranges in a single graph – take a look, and then I’ll explain.

The left-most pair of bars shows the range of carbon stored in forests. Big range, but forests are diverse, as the previous bar chart demonstrated. The three other bar pairs show how much carbon is avoided by an acre of solar, in different grid regions and in areas with different solar potential. In reality, these factors can be correlated, but we’re presenting general sensitivity analysis here, rather than individual cases.

Wait, why apples and oranges? Consider the following: the forest bars are stocks, while the others are annual flows. In other words, you cut down a forest and lose something in the range of forest values presented here, but then you get back an annual flow in the range of solar outcomes. That means we can calculate a “payback” period, i.e., how many years it will take solar in different situations to “pay back” the carbon lost by destroying the forest where the solar plant is built.

The results are striking: the payback is fast in two notable ways. First, the payback happens well inside a decade so it’s on a timeframe that’s relevant for action. And second, the payback is dramatically shorter than the expected life of current solar facilities – indeed, most solar modules come with 25- or 30-year warrantees. (These payback calculations do not take into account the inevitable slight degradation, on the order of 0.25% to 0.50% per year of output. Incorporating those numbers wouldn’t change much.)

I’ve included all of my calculations in this Google Sheet. Let me know if you have questions.

Note #1: Biological carbon, other forest value, and solar boom zones

One could quite straightforwardly use this methodology to consider very specific scenarios, using eGRID data for a particular grid region, forest carbon data specific to a particular region and forest management regime, and the details of a proposed or existing solar generation facility. The possibilities are endless, but they are highly likely to fall into the ranges defined here, simply because I’ve chosen well-distanced bookends for all of the variables.

It is crucial to stress that this “analysis” implies a highly reductionist view of the value of forests. I do not hold that view. Natural forests are not merely stocks of carbon; they house biodiversity and furnish habitat, provide clean water and other watershed functions, offer recreational value, and represent unique cultural importance. These factors should play a role in any assessment involving forests.

Specifically, it’s worth looking at the regions in which solar is taking off. In arid and semi-arid ecosystems, carbon per acre is much lower, so rapid solar development in California, Arizona, and Texas suggests a fast payback.

Note #2: How much land would we need for solar for all U.S. electricity?

This calculation (that I cite near the start) is fun and interesting, but it has some substantial caveats. We must assess those before proceeding:

• Simplification of time: I calculate energy (in MWh) for an entire year, but I don’t worry about how to get energy from July and August (more sun in most places) to December and January (less sun). That sort of long-duration storage isn’t cheap or easy right now.
• Simplification of space: I also consider the U.S. as a whole without worrying about where the best places for solar are and where the energy use is.

Still, it’s a good sense-of-scale calculation. You can find others on the web, but I’ve built this from the ground up so you can see the math. The ingredients:

• Total electrical energy use in the U.S. in 2018: 3,860,119 MWh (Source: Electric Power Annual (2019), EIA, Table 1.2)
• Land intensity of solar capacity: 6-8 acres per nameplate MW (Source: I figured out the ratio from various utility-scale solar plants, such as the Villanueva Solar Park in Mexico, the Desert Sunlight Solar Farm in California, and the Mesquite Solar project in Arizona. There’s an excellent Wikipedia page of the world’s largest solar power stations for starting points to build additional comparisons, but the range of land intensity values is fairly tight.)
• Capacity factor for solar: 1200-1700 MWh/MW per year (Source: NREL’s PVWatts calculator. There are endless possibilities here, but I used a low end that is slightly below some places where solar is happening, such as Minnesota and New York, and a high end that represents some sunny places where solar is booming, such as Arizona, California’s Central Valley, and West Texas.

The rest is conversions, but for thoroughness:

• These numbers produce a range of 150-283 MWh generated per acre.
• With 2018 electricity usage, that means a land area needed of 13.6-25.7 million acres, or 21,000-40,000 square miles.
• Given a U.S. land area of 3.8 million square miles, the resulting need is 0.56% to 1.06% of total U.S. land area for solar plants in order to generate all U.S. electricity usage from solar.