Page 89 - Climate Change and Food Systems

(Collins, 2007; Gleckler, et al., 2008). Recent model simulations are likely to be more reliable than earlier vintages due to continued modelling improvements.
There are two key sources of uncertainty in GCM climate projections. The first stems from the practice of “parameterizing” sub-processes in GCM models.6 Parameters are used to describe the average or expected effects of sub- processes on the larger climate processes that can be modelled explicitly. They can number in the hundreds in some models. Different models incorporate different processes and parameters, and any or all of the models may not capture
all of the known and unknown influences that determine the future climate (IPCC, 2007). To address uncertainty about the values assigned to parameters, they have typically been formulated as the mean effect of alternative parameter values, averaged across many model runs, and assuming the same final equilibrium state for the
larger process. To better describe uncertainty about parameter values, GCMs have begun to incorporate stochastic parameter values that
are sampled from a known probability of their occurrence for a given final state, and model results are then expressed in terms of probabilities of outcomes across a range of parameter values (McFarlane, 2011; Flato, 2011).
A second source of uncertainty is due to natural variability. Because the temporal scale of GCMs is so long-term, even small differences in a model’s initial climate conditions can lead to large differences in results for specific future dates, such as 2050 or 2100. To address this uncertainty, climate modellers have begun to develop ensembles in which the same simulation is run in a single model across a range of initial conditions that are randomly sampled from observations of the atmosphere. This practice reduces potential model error associated with natural variability
due to the effects of short-term weather event phenomena, such as El Nino. Results thus describe probabilistic outcomes with respect
to initial conditions in a chaotic weather system (Clark et al., 2010; McFarlane, 2011).
chapter 3: economic modelling of climate impacts and adaptation in agriculture: a survey of methods, results and gaps
6
Sub processes include: radiation, water vapor, aerosols, clouds, precipitation, temperature, oceans, soil moisture, biological processes, permafrost, miscellaneous (IPCC, 2014).
figure 3
Projected emissions in the representative concentration pathways (RCPs) and extended concentration pathways
Source: van Vuuren et al., 2011
69