Numerous statistical issues arise from dispersed data and diverse data types and
sources for agricultural processes and production across and within countries. As one
example, given current data restrictions estimates of agricultural contributions to
greenhouse gas emission are inconsistent. Accounting for the spatial variation in
agricultural practice at various scales (e.g. between country heterogeneity and within
country variations) and the temporal dynamics associated with both seasonal and year-
to-year variations presents significant challenges of statistical analyses to estimate not
only current greenhouse gas (GHG) inputs but longer-term projections as well. Similar
issues arise in projection of agricultural resource demands when taking account of
potential economic variations in the costs of inputs such as fertilizer, and associated
spatial changes in production practices.
These present opportunities for analysis of spatial stochastic processes with
dynamics operating on longer time scales (e.g. crop system responses to climate
changes over decades) and shorter ones (e.g. between seasons) to estimate current
agricultural impacts spatially averaged across the world. Development of agreed upon
methods to estimate these impacts can provide useful inputs to a variety of international
policy decisions as well as serve as potential planning tools for agribusiness. Methods to
effectively compare the spatial distributions of production to models would provide
confidence that the models can be effectively used to compare impacts of alternative
management practices and longer-term policies. Confidence developed through
mathematical analysis of spatially-structured models for emissions are essential for
providing evidence that mitigation strategies for pollutants can work, the time scales
these might require, and whether regulation or incentives associated with such mitigation
strategies can be effective.
An additional challenge is to develop methods to elaborate equitable allocations
of resources to meet increasing world demands arising from population growth,
economic growth of the developing world and associated potential changes in caloric
intake and animal protein consumption. This includes developing methods to handle
spatial variation in demand and production so as to evaluate alternative assumptions
about future consumption demand. The large spatially-disaggregated economic sector
models tend to be sensitive to uncertainties in demands arising from social systems
models, presenting significant computational challenges. In addition, this complexity
requires new impact estimators derived from the spatiotemporal model outputs.
Integrated assessments of economics linked to biological production models and
alternative models for social system response can reflect the strengths and weaknesses
of different worldviews by comparing alternative models. A mathematical challenge
concerns whether rankings of impacts of alternative scenarios derived from differing
management plans or policies are robust to uncertainties; such uncertainly stems from
human system responses, environmental conditions derived from climate assumptions,
and also from the parameterization of the models. The robustness of such relative
assessments of scenarios has been applied in ecological evaluations (Fuller et al., 2008)
in a computational framework, but it is possible that new mathematical approaches could
lead to generalized results on the robustness of rankings.
The complexity of interactions between components of agricultural systems
models and the associated need for extensive parameterization naturally lead to
concerns about data availability and the utility of these models to project the implications
of current trends. The implications for human food consumption are sufficiently important
barry
(Barry)
#1