Throughout this blog, I’ve been
talking about models. Not ones that go down the catwalk, nor ones made out of
paper and glue. I’m talking about computer models that attempt to represent the
natural environment in some logical way and predict what is going to happen in
the future. Predicting the future is an extremely difficult task, with huge
uncertainties. It’s time to talk about the art of modelling to provide some
context for the variety of studies I have examined.
Most of the models in studies
I’ve been talking about are extremely complex, using a huge amount of data and
making a lot of calculations to make predictions about the future. Other types
of environmental model can be much simpler (less data and making less
calculations), but sometimes are not suited to the complex nature of future
climate change research. However, there are examples of these models being used
to discuss the general direction of change.
Just getting the data alone to
run large environmental models can be a considerable issue – the requirements
can be very hard to meet, particularly in a scientific climate in which a lot
of meteorological and hydrological data is not free to access and hidden behind
barriers (Alliance for Permeant Access, 2011). The high data requirements also
may increase uncertainty in predictions – more data means more parameters (changeable
values that alter the results of a model), and so more uncertainty in the
predictions the model makes as the range of possibilities is higher (Beven 1993). This uncertainty is exacerbated by the fact often only the bare minimum
of data needed to run models is available – there would be less uncertainty if
there were fuller data sets, but these are rarely available.
Complex models used require huge
amounts of computing power, expertise and most of all time to construct and
run. I would personally argue that the
massive uncertainties associated in complex climate and hydrological modelling
mean we certainly need to take results with a huge pinch of salt – in most
cases all we can tell is the direction and general magnitude of change, and
even this can be unclear in some models due to huge uncertainties. I hope this
does not put everything that has come before in doubt, but the modelling
studies I have talked about really do only represent our best guesses, and by
no means prescribe with accuracy what will happen in the future. This is not to
deride modelling work though – it is our best guess, and a necessary guess. It
is with some urgency that we need to attempt to quantify future changes to
runoff and discharge – I will explore why in the next post.
Surprise, surprise, it’s
extremely hard to predict the future. But we are doing the best we can.
0 comments: