environmental change; for example, a politician is usually in office for
only a few years, while the effects of his or her decisions on the
environment may not be seen for decades. New mathematical
methods are needed to characterize this discrepancy and to develop
strategies for managing it, including subtle uses of discounting to better
understand the tradeoff between short-term economic gain and long-
term environmental degradation and the development of the science of
uncertainty to better understand the likelihood of long-term impacts of
current activities.
- HESs often involve very complex datasets that are then reduced to
many fewer variables in order to make the problems tractable. The
health of an individual forest, for example, emerges from the health of
all the species in the forest and the way they interact – a very complex,
high-dimensional dataset – but in a model of the world’s forests, that
health might be captured in a single number. This process of reducing
a high-dimensional dataset to a lower-dimensional one is called
projection geometry, and it needs further development in the context of
HESs.
- Using ensembles of models to improve prediction has emerged as an
especially important tool in forecasting HESs. This technique is poorly
understood theoretically and needs more development.
