attention to the maximum technically-feasible reductions for the
relevant sectors (Amann et al., 2007). The use of positive and
negative variations with the same absolute value (90%/90%,
50%/50%) was made in order to minimize the nonlinearity effects
of the advection algorithm according to what is discussed in
Bott (1989). Due to limitations in hardware resources, TMs were
built from experimental runs using the meteorology of year 2007
(BS) and thus assuming that the meteorology of this year is
representative of future situations. It is worth noting that
obtaining a TM is a complex and computer-intensive process,
being this an additional motivation for considering only those
sectors that accounted for the greatest emissions of a given
pollutant (Table 3). Due to computing limitations and to the
sector and pollutant-specific nature of TMs, conducting the
parameterization with only five simulations was deemed suffi-
cient for guaranteeing statistical consistence without compromising
the laboratory’s computing capacity. This fact is
particularly true since it was observed that increasing the number
of simulations for the construction of TMs slightly increased
the quality of the statistical regression (Vedrenne et al., 2013).
Additionally, it should be kept in mind that a TM is a statistical
parameterization of model performance so limitations in the
robustness of AERIS diagnostic capabilities should definitely be
expected when compared to a deterministic AQMS.