468
a wider set of calibration data from across a wider range of regions, potentially
using clustering algorithms to identify comparable regions for analysis.
Both of these challenges raise a final fundamental challenge, which is how to
create meaningful sub-state IAMs in the absence of sufficient data. Research on
data integration (the lower right quadrant in Fig. 12.4), will help to bridge data gaps,
but many gaps can be expected to remain, particularly at local scales and in developing countries that may not have the same tradition of data collection and management as the USA. For this reason, an important scientific challenge and focus of
FEW systems modeling research will be creating multiple meaningful scenarios
that effectively span the parameter uncertainty space in the context of limited data.
In addition to linkages to data research, this will necessarily also rely on the codevelopment of parameter sets with stakeholders (2) and visual analytics methods
(3) as part of that co-development process.
Third, and related, the development teams for models and for the use of models
in decision-making must consist not only of modelers, but also of social scientists
that understand how different actors frame the problems and potential solutions and
how scientists, modelers, and decision-makers interpret key concepts such as uncertainty; data development experts that can create usable information from often
incomplete sources and characterize the uncertainty in this information; visualization and other communication experts to bridge the divide between modelers and
decision-makers.
These challenges are particularly acute in developing countries with less data
and modeling infrastructure. Overcoming these challenges will be critical if the
world is to address FEW system needs while simultaneously making regional/local
decisions that fulfill other related needs such as sustainable development goals
(Sachs 2012) and climate change commitments.
17.3.3 Visual Analytics Methods
Understanding future FEW system conditions and their impacts is based, in part, on
the ability to model the key drivers and the underlying processes of complex interactions between FEW systems and human systems. Various food, energy, water,
climate, and socioeconomic scenarios can reveal different projections of future
FEW system conditions and their cascading impacts. It is important for analysts and
policymakers to design strategies (e.g., for sustainability purposes) based on different scenario results.
These scenarios can be modeled by the interaction between ensembles of FEW
system models with varied model parameters. This assumes that those scenarios can
approximately describe such interactions between natural and human systems given
limited observations, imperfect assumptions, and finite model choice. Thus, it
becomes necessary to explore differences in modeled FEW system scenarios and
the resulting variability in supply and demand. Most visual analytics tools for modeling focus on both spatiotemporal pattern identification and uncertainty analysis of
E. Eftelioglu et al.
a wider set of calibration data from across a wider range of regions, potentially
using clustering algorithms to identify comparable regions for analysis.
Both of these challenges raise a final fundamental challenge, which is how to
create meaningful sub-state IAMs in the absence of sufficient data. Research on
data integration (the lower right quadrant in Fig. 12.4), will help to bridge data gaps,
but many gaps can be expected to remain, particularly at local scales and in developing countries that may not have the same tradition of data collection and management as the USA. For this reason, an important scientific challenge and focus of
FEW systems modeling research will be creating multiple meaningful scenarios
that effectively span the parameter uncertainty space in the context of limited data.
In addition to linkages to data research, this will necessarily also rely on the codevelopment of parameter sets with stakeholders (2) and visual analytics methods
(3) as part of that co-development process.
Third, and related, the development teams for models and for the use of models
in decision-making must consist not only of modelers, but also of social scientists
that understand how different actors frame the problems and potential solutions and
how scientists, modelers, and decision-makers interpret key concepts such as uncertainty; data development experts that can create usable information from often
incomplete sources and characterize the uncertainty in this information; visualization and other communication experts to bridge the divide between modelers and
decision-makers.
These challenges are particularly acute in developing countries with less data
and modeling infrastructure. Overcoming these challenges will be critical if the
world is to address FEW system needs while simultaneously making regional/local
decisions that fulfill other related needs such as sustainable development goals
(Sachs 2012) and climate change commitments.
17.3.3 Visual Analytics Methods
Understanding future FEW system conditions and their impacts is based, in part, on
the ability to model the key drivers and the underlying processes of complex interactions between FEW systems and human systems. Various food, energy, water,
climate, and socioeconomic scenarios can reveal different projections of future
FEW system conditions and their cascading impacts. It is important for analysts and
policymakers to design strategies (e.g., for sustainability purposes) based on different scenario results.
These scenarios can be modeled by the interaction between ensembles of FEW
system models with varied model parameters. This assumes that those scenarios can
approximately describe such interactions between natural and human systems given
limited observations, imperfect assumptions, and finite model choice. Thus, it
becomes necessary to explore differences in modeled FEW system scenarios and
the resulting variability in supply and demand. Most visual analytics tools for modeling focus on both spatiotemporal pattern identification and uncertainty analysis of
E. Eftelioglu et al.