Growing Fossils: “Embodied Land”

Now and then there is debate about the land use of energy generation, usually around renewable energy. But these are often only addressing the direct ‘ visible’ land use and efficiencies; For instance: A energy Atlas in the Netherlands counts only the land needed for a fossil fuel power plant itself. Not the land involved in producing fossil fuels. Nor land impacts from the materials for the power plant , or for a wind turbine , that have an impact as well, which can be transferred to land occupation. Which I call Embodied Land in a new Exergy Based methodology.

Recently Victor Smil published a book on the land use impacts of renewable energy sources. [Smil 2015 ] Which is one of the few publications adrressing real land use. However, also in this book, fossils are just fossils, and no calculation of their land use impacts for “processing” are included.

In a paper a few years ago I estimated electricity generation from coal and oil via biomass in terms of Embodied Land: kWh generated per m2 of original biomass. [Rovers 2011]

I divided the total recoverable stock ( used and unused) and divided that by time ( 65 million years of processing and land occupation) and land ( the whole earth surface biomass in terrestrial and marine sediments of which only part became fossil fuel) to come to these data.[x] It was a very rough calculation to get an estimate, as a time space relation. It turned out that the production of oil was in the order of 14000 liters per year globally. The average for electricity production by oil coal and gas together , came down to 0,00000138 MJ per m2-year (from Solar radiation to electricity, compared to 432 MJ/m2-year for PV panels. Who says that PV panels are not efficient…?)

Not completely surprising these are very low figures for the efficiency and output, though a somewhat pessimistic estimate, since no double use of the land was considered.

Recently I discovered a paper that analyses this process of fossil fuels in more detail, [Dukes 2003] with more fundamental data research to use instead. Dukes calculates the several steps in the fossil fuel forming by means of preservation factors and recovery factors:

a preservation factor (PF): is defined as the fraction of carbon that

remains at the end of a transition from one fossil fuel precursor to the next

A recovery factor (RF): is defined as the proportion of original photosynthetic product recovered as fossil fuel.

Recovery factors are the product of the PFs of each transition and additional terms for extraction efficiency

After calculating these factors , like for coal the step from biomass to peat, from peat to browncoal, and from brown coal to bituminous and anthracite, it comes out at ~9 % efficiency . From biomass to coal that is. To be added should in fact be the steps from solar to biomass, and from coal to electricity. (coal to heat is bit different, here we compare electricity)

Recalculating Dukes data for these steps data gives a 0,082% efficiency from solar to electricity. (see the table 3 by Dukes). For the route to oil and gas similar calculations are made. Oil and gas are around a 1000 times less effective.

With 1997 as a reference year Dukes calculates that to generate the fossil fuels burned in 1997, it required ( in original biomass) 422 times the amount of carbon that is fixed globally each year!

Or 36 times the sum of solar energy striking the earth in one year.

If recalculated backwards , Dukes could estimate that the turning point was 1888, in which year for the first time in history more was consumed via fossil fuels than reproduced globally.

If I transfer Dukes’ data to “ solar powered Land” , we get the following:

coal : 2,3 MJ/ m2

oil : 0,0027 MJ/m2

gas: 0,0024 MJ/m2

for comparison: the PV route provides 432 MJ /m2 . (NL)


However, these data are only for space, not yet for time compensated. It takes millions of years to cook and pressurize these resources. Dukes does not address this, only assumes they require a storage (ie process) time of over 10 million year . In that case we have to divide his land data by this 10 million, to get the time related production figures. (in m2-year). (just for comparison: for coal this would give 0,00000023 MJ/m2/year , which is in the same order of magnitude as my previous calculations, ‘only’ a factor 6 different from my estimates : no surprise, since I calculated for 65 million years of processing…)

If not divided by this time factor , its assumed that the land is only occupied for direct production of biomass and does not claim land afterwards for processing, so the output in energy is on a real time basis, a yearly result.

The question is therefor: is the land during the processing period available or not available for other use ? Which decides if we have to include this time land occupation? Above we have two extremes: in my rough calculation its assumed all land is all time occupied and required, Dukes assumes indirectly that the land is needed once, to produce the biomass only.

Of course, when the biomass is continuously stacked upon the previous years layer a continuous process takes place and the processing itself does not occupy additional land, since underneath the new yearly additions.

At some point in the process, supplementing the process with biomass will stop and continue ‘underground’, and from one point of view still claims space , be it underground. However, the top soil could then be used for other activities like living area on top. ( as is the case in The Netherlands, where the natural gas is mined underneath the living area in the province of Groningen- which is now undergoing earthquakes as a result of subsoil earth settling; in a way proving that the land should have been preserved for the process and not for living on top! Or the other way around, when living on top is allowed, the gas should have stayed subsoil.) Of course, mankind is only around for a few years in serious numbers, and processing has been able to continue for millions of years without the question of land use by people.

A practical outcome in this discussion is to give fossils the benefit of the doubt, for as long as this problem is not studied in depth, and use the continuous process preference, and the production data per m2 by Dukes become production data per m2-year.

space use of energy sources.

The data used above, both by Dukes and me , are still direct data, only for energy generation. Not included is the land claim for energy and materials to process the energy, like power plants, grids, PV panels, inverters etc. These require Embodied Land as well., A first calculation for polycrystalline PV panels shows that 1 m2 of PV panel requires in fact 3323 m2-year of Embodied Land, to (re-) generate the resources and energy that went into the production of the panels. (see ) Suppose its life expectation is 25 years, each m2 PV panel requires an extra 133 m2 of land continuously in production for that 25 years, to compensate for the panel production (in terms of solar power driven conversions in mass and energy) The net-power output per m2 land reduces from 432 MJ to a net 2,51 MJ per m2 . Which is still a little more compared of fossil fuels, for which however the grids and power plants are not yet included, and would lower the fossil data significantly.

There are a few other studies using land and space use of renewable energy, one of the latest is by Smil in his book Power Density . This shows that renewable energy requires much more land then usually estimated, calculated in terms of power output per m2 land. It has some good data , but the point is that in these calculations, like Smil’s , the land use by fossils, as we calculated above, is seldom incorporated. What usually happens is that the consequences of renewable energy are seen as too heavy, too costly, to ineffective , to much land, or else, which then is used as reasoning for continuing using fossils. But arguing this way, it would be a fair comparison to address fossil fuels the same, as if produced-processed today ( which in fact they still do) : the consequences of fossils are more huge, and with the real effects in land use and efficiency compared for both, the advantage of renewables will be clearly seen before fossils. ( see also the article in Lowtech magazine by Kris de Decker, based on Smil, and my reactions ) ( In terms of cost there is a difference of course, but only since the processing of coal is left out of the equation).

Any way, in either case, globally , we will have to reduce our demand and consumption drastically, (see also blog on CO2 budget) to continue ‘life’, although maybe in a less luxurious form. For the industrialized countries that is. Just changing to renewables, or even “new grown fossils”, will not work. Renewables , though much less as fossils, have their impact as well: not in the source, but in their land use and impacts from conversion devices.

* A working environment for a  tool  for testing calculations with “Embodied Land” has been launched past weeks. If you are interested have a look at


Rovers R. et all, 2011 , Space-time of solar radiation as guiding principle for energy ánd materials choices , World Renewable Energy Congress 2011 Sweden May 2011, Linköping, Sweden (sea )

Dukes, J.S. 2003. Burning buried Sunshine: Human consumption of ancient solar energy, Climatic Change 61:31-44

Smil, V., 2015, Power Density: A Key to Understanding Energy Sources and Uses, MIT Press ISBN: 9780262029148

lowtech: The article in in dutch om this website:

Author: ronald rovers