Geoscience Reference
In-Depth Information
rise by 55% on the other end. For policymakers, not only do these estimates provide a very broad
range of possibilities in absolute terms, these models cannot even predict the direction of change
with any real certainty.
Another example are global economic land use models, which make predictions about land cover
and land use in the future. A recent study has shown that current global estimates of the amount of
land under crop production vary by as much as 300 Mha (Fritz et al. 2011). These estimates were
derived from different global land cover products that have been developed from satellite remote
sensing. The significance of this becomes clear when this land cover information is used as an
input to other models such as economic land use models. In a recent model comparison exercise by
Smith et al. (2010), the global change in cropland area (and other land uses) in 2020 and 2050 was
compared across the different models under various scenarios including climate change. Many of
the predictions of cropland change in the future were less than the 300 Mha, which means that the
predictions fall within the uncertainty band of current baseline cropland information. Other stud-
ies have shown that model outputs and analyses can vary substantially depending upon which land
cover product has been used (Ge et al. 2007; Quaife et al. 2008; Linard et al. 2010).
These examples are only the tip of an iceberg, and as Maslin and Austin (2012) clearly articulate,
with increased computing power and the ability to run larger and more complex models, this
hidden
uncertainty will increasingly come to the surface.
18.7 CONCLUSIONS
The focus of this chapter has been to discuss the current limits to GC, which most people might per-
ceive to be computational power. For those of us who have experienced the changes since the first
edition of this topic was published, we can now run models on our desktop that were not possible
just 15 years ago. At the same time, our models have become much more complex, so it is not dif-
ficult to build models now that take what we might consider to be an unreasonably long time to run.
Our tolerance to acceptable run times has probably also decreased over time. However, the biggest
limit is actually not computational power itself but the need to make better and more efficient use of
what resources already exist. Maybe 15 years from now, when we all have access to quantum com-
puters and our code is optimised for this new parallel environment, we may look back and wonder
how we ever managed to model anything complex with any degree of realism.
There are still, however, real limits in predictability and understanding due to the presence of
chaos, events that cannot be anticipated, the amount of complexity that we can model, the over-
whelming amounts of spatial data now being collected and the widening range of uncertainty esti-
mates. As scientists, we will keep pushing these limits as far as possible, as our computational
environment continues to improve. For the field of GC and spatial sciences more generally, many
of these limits can be viewed as continued challenges that will lead to new research questions and
ultimately drive these disciplines forward in the future.
REFERENCES
Abbott, M.B., J.C. Bathurst, J.A. Cunge, P.E. O'Connell, and J. Rasmussen. October 15, 1986. An introduction
to the European hydrological system—Systeme Hydrologique Europeen, “SHE”, 1: History and
philosophy of a physically-based, distributed modelling system.
Journal of Hydrology
87(1-2): 45-59.
Adnan, M., P.A. Longley, A.D. Singleton, and I. Turton. 2014. Parallel computing in geography. In
Geocomputation
, 2nd edn., eds. R.J. Abrahart and L. See, pp. 49-68. Boca Raton, FL: Taylor & Francis.
Albert, A. 1995.
Chaos and Society
. Sainte-Foy, Quebec, Canada: IOS Press.
Antsaklis, P.J. and A.N. Michel. 2007.
A Linear Systems Primer
. Boston, MA: Birkhauser.
Baker, S. 2011.
Final Jeopardy: Man vs. Machine and the Quest to Know Everything
. Boston, MA: Houghton
Mifflin Harcourt.
Barrow-Green, J. 1997.
Poincaré and the Three Body Problem
. Providence, RI: American Mathematical
Society.
