In addition to assessing
vulnerability to current climate, there is also likely to be interest
in assessing how vulnerability might change into the future. Although
this embodies additional uncertainties, to do so requires understanding
of how the climate is projected to change in the future. Projections of
future climate that are applied in VIA assessment are conventionally
referred to as
climate
scenarios (Mearns et al., 2001), which distinguishes them
from climate predictions or forecasts, to which probabilities can be
attached. However, this distinction is becoming blurred as climate
scientists have moved towards expressing future climate in terms of
conditional probabilities. A useful recent comparison of different
methods of climate scenario development for use in VIA is provided by
Wilby et al. (2009). Table 3-2 combines elements of that review into a
summary of different scenario construction methods, their resource
needs and potential applications.
The most
credible and sophisticated tools for simulating the response of the
Earth’s climate to increasing emissions of greenhouse gases
and aerosols are coupled atmosphere-ocean general circulation models
(AOGCMs). There is agreement among all models that the planet will
warm, globally, though the magnitude varies from model to model. There
is less unanimity in the projected regional pattern of changes in other
climate variables such as precipitation, radiation or windspeed, and
the spatial resolution is quite coarse (grid box dimensions are seldom
finer than 150 km). Since most impacts of climate change will be
manifest locally, there have been great efforts to downscale AOGCM
projections to a finer spatial resolution (Fowler et al., 2007), either
using numerical models (Mearns et al., 2003; Rummukainen, 2010) or
statistical techniques (Wilby et al., 2004), and sometimes involving
the use of stochastic weather generators (Wilks, 2010). There have been
several major research projects conducted to this end in Europe (e.g.
PRUDENCE (Christensen et al., 2007), and ENSEMBLES (van der Linden and
Mitchell, 2009), North America (e.g. NARCAP – Mearns et al.,
2009), and globally with a current focus on Africa (CORDEX –
Giorgi et al., 2009).
Table 3-2: Selected methods of climate scenario development classified according to theirresource needs and potential applications for adaptation planning (based on tables in Wilby etal., 2009: and amended, with major additions in italics).
Level of
resource needs | Methods |
Spatial application and
input requirements | Applications
for adaptation
planning |
| Limited |
Sensitivity analysis
Climate analogues
Trend extrapolation | Local (site/area)
(Observed climate data) | Resource management,
Sectoral
Communication, Institutional, Sectoral
New infrastructure (coastal)
|
| Modest | “Delta”
change
Pattern-scaling
Stochastic weather
generation
Empirical/statistical
downscaling | Regional
(AOGCM and simpler
global model outputs) | Most adaptation activities
Institutional, Sectoral
Resource management,
Retrofitting, Behavioural
New infrastructure, Resource
management, Behavioural |
| High |
Dynamical downscaling
(RCM)
Coupled AOGCMs Regional-global
Probabilistic |
Regional-global
(AOGCM outputs)
Regional-global
Global-regional-local
(Multiple sources) |
New infrastructure, Resource
management, Behavioural,
Communication
Communication, Financial
New infrastructure, Resource
management, Communication |
An alternative method used to
generate climate scenarios involves identifying spatial analogues
(climates in other regions) or temporal analogues (climates from the
past) that may resemble anticipated future conditions in a region (Ford
et al., 2010). Other simple techniques involve adjusting present-day
climate by fixed increments (e.g. warming in increments of 1°C;
precipitation changes in increments of ±5%) to explore the
sensitivity of exposure units to a changing climate (IPCC, 1994), or
applying simple extrapolation of past trends (Wilby et al., 2009).
Perhaps
the most common technique for applying climate scenarios in VIA studies
is the so-called “delta change” method, whereby
changes between modelled reference and future periods are appended as
factors (or “deltas”) to the climate observed
during the reference period. This technique recognises the common
biases found in model representations of present-day climate (e.g.
Fronzek and Carter, 2007). Pattern-scaling is a method often applied in
integrated assessment models (IAMs) for relating the regional patterns
of changes in climate derived from individual AOGCM simulations to
global mean annual temperature (which can be computed in simple climate
models). The same pattern can then be scaled up or down according to
the simple model’s temperature projections for a wide range
of emissions scenarios and future time periods (Mitchell, 2003).
Finally, as computer power has improved, so multiple ensemble
simulations with climate models have become feasible, allowing
different model uncertainties to be explored and encouraging climate
scientists to attach likelihoods to climate projections. The UKCP09
projections are probabilistic (Murphy et al., 2009) as are recent
projections for Finland (Räisänen and Ruokolainen,
2006), Australia and southern Africa (Moise and Hudson, 2008) and
Europe (e.g. Harris et al., 2010).
Case Study: Use of GCMs to
determine climate futures in New York and the
Metropolitan East Coast region
As part of the US
National Assessment of the Potential Consequences of Climate
Variability and Change, an assessment of climate change and the
Metropolitan East Coast (MEC) region - covering the 31 countries of the
New York City metropolitan region and a total population of 19.6
million in the states of New York, New Jersey, and Connecticut - was
undertaken. The goal was to understand the impacts of climate
variability and change on the physical and human systems.
The
assessment used five GCM scenarios: one based on current trends; two
from the UK Hadley Centre and two from the Canadian Centre for Climate
Modeling and Analysis, both of which consider greenhouse gases
individually, and then a combination of greenhouse gases and sulphate
aerosols that are emitted through industrial activities. Typically
sulphate aerosols create a cooling effect by reflecting and scattering
solar radiation, and thus they offset greenhouse gases to a certain
extent. As a result, using these scenarios forecasts lower temperatures
than scenarios that include only greenhouse gases. This gives a good
estimate of the envelope of potential change. Linear interpolation
between GCM grid boxes meant that scenarios were obtained for several
of the cities within the region. However, because the cities are
relatively close, there is little variation between them, and so the
study used the mid-point of the study region.
While
each of the five future scenarios provide a distinct projection of
precipitation change, it is important to note that the precipitation
projections of the GCM scenarios do not agree either in magnitude or
direction (as opposed to the projected temperature changes, which agree
in direction, but not magnitude). The Hadley Centre’s
scenarios show increasing levels of precipitation while the Canadian
Centre projects decreasing precipitation over time.
Through
the use of a range of plausible scenarios, the MEC assessment
researchers are able to project possible impacts created by climate
variability and change as well as to evaluate the MEC
region’s responses. An assessment exercise such as the MEC
study is useful in enabling preparedness for extreme climate events in
the present as well as readiness for a changing future climate.
Source:
Metropolitan East Coast Assessment website (http://metroeast_climate.
ciesin.columbia.edu/index.html)
and Clean Air Partnership (2007)
|
