Mass. Institute of Technology (MIT) Greenhouse Gas Emissions Scenarios for 1995-2100 (CO2, CH4, N2O, SF6, PFC, HFC, NOx, SOx, CO, NMVOC, NH3) at 5 yr intervals

A set of three (low, high, reference) emissions scenarios for CO2,
CH4, N2O, SF6, PFC, HFC, NOX, SOX, CO, NMVOC, NH3 and carbonaceous
particulates from 1995 through 2100 at 5 year intervals, provided on a
1 degree by 1 degree latitude-longitude grid.

The scenarios were generated using the Massachusetts Institute of
Technology (MIT) Emissions Prediction and Policy Analysis (EPPA)
model, developed by the MIT Joint Program on the Science and Policy of
Global Change. The known probability characteristics were generated
using an uncertainty technique known as the deterministic equivalent
modeling method (DEMM). The emssions values were downscaled from
aggregated EPPA regions using population as weights.

Briefly, the DEMM approach requires distributions for uncertain
parameter inputs and then uses a statistical sampling technique to
generate parameter sets. The full EPPA model is run with these
parameter sets and then an nth order polynomial is fit to the outcomes
of interest. A Monte Carlo analysis of 10,000 runs was then conducted,
again using the input distributions, but simulating the polynomial fit
rather than the underlying complex model to limit computation
requirements. We then select parameter sets and using the underlying
complex model to produce scenarios with a known probability for the
outcomes of interest.

The 3 scenarios were selected as follows: We chose parameter sets that
produced the median (50 percentile value) and 97.5 percentile value
(upper) and 2.5 percentile value (lower) limit for CO2 emissions in
2100 (i.e. a range covering 95 percent of the
distribution). Conditional on these values, we then chose parameter
sets that produced 50 percentile values for each of the other
emissions. Conditioning the other emissions on the likelihood of the
CO2 emissions scenarios means that emissions of the non-CO2 substances
are higher for the scenario conditioned on the 97.5 percentile than on
that for the 50 percentile which are, in turn, higher than those
conditioned on the 2.5 percentile value scenario of CO2 emissions
because of generally positive correlation among emissions of these
substances and CO2. This scenario selection design was chosen so that
the resulting scenarios are approximately 2.5, 50, 97.5 percentile
outcomes yet retain the characteristic that the scenarios for all
substances are the result of internally consistent scenarios given the
structure of the EPPA model.

Some important caveats: These scenarios do not include emissions from
natural sources or sinks of carbon, other GHGs, or other
substances. They include emissions of carbon and other substances from
land use change (deforestation) and agriculture (waste burning,
livestock, rice production, soils) but do not include carbon sinks due
to forest regrowth. Also, note that we include uncertainty in
emissions for most of these substances in our base year of 1995. This
means that, for a predetermined level of natural emissions and/or
uptake, any particular anthropogenic emissions scenario in our set may
not be consistent with an observed trend in concentrations of gases in
the atmosphere (e.g. CH4, N2O).

Units: Gigagram (Gg) of compound (CO2,N2O,etc.) emitted in that cell
per year. PFC is in Gg of CF4 equivalents. HFC is in Gg of HFC-134a
equivalents.