There is an ever-present call for greater accuracy and confidence and added value in weather forecasts and climate predictions to achieve applicability in decision-making for societal resilience and for economic and ecosystem sustainability. Climate modelers are being asked to provide predictions of future temperature and precipitation under climate change scenarios not just at continental scales, but for individual states and cities that may be at risk for overall warming, diminished water supply, and extreme events such as heat waves and floods. Weather forecasters issue timely warnings on severe weather using models that rely on a growing awareness of the interactions between the land surface and the atmosphere over short time scales, such as the growth of a thunderstorm, and over longer periods such as the development of a seasonal drought. Hydrologists apply knowledge of weather and the land surface to the understanding and forecasting of numerous events with similarly broad societal impacts, from urban flash floods to regional river floods, from rainfall deficits on local farms to seasonal droughts affecting entire countries, and from the effects of a forest fire on a city’s water supply to the long-term impacts of climate change on global water resources. Ecologists and numerous specialized communities are concerned about the potential impacts of climate change and extreme events on the health and spatial distributions of forests, animal species, agricultural lands, and the human-built environment.Again, feedback is welcome!
From their individual perspectives, each of these communities of practice has developed sophisticated tools and methods, both for their own understanding of the system under study and to deliver their results for application and use by decision-makers and the public. In some cases additional value and accuracy may be achieved by re-evaluation of both overt and implicit assumptions that accompany these modeling and forecast efforts. In many areas of the modeling and forecast endeavor, those methods originally devised for the parameterization of observed phenomena can now be supplemented or replaced with more recent empirical and analytical datasets. With the growth of specialized understanding in many of these subjects, some parameterized forecast system components can now be abstracted entirely to employ physical representations of observed processes. The analyses to support that reformulation can now be provided by aligned communities such as remote sensing specialists, foresters, ecologists, etc. The development of a comprehensive and physically accurate understanding of the natural world based on both structure and function of the ecosystem requires contributions from numerous specialized fields. This multi-disciplinary approach is recognized as the most viable path to deeper understanding and greater accuracy when attempting to gauge impacts both of natural systems on human decision-making, and of humans on their natural environment.
Historically, many such modeling systems have augmented a detailed treatment of the central problem with coarser representations of “external” processes, those aspects of the physical system that are essentially outside the modelers’ focus. However, an examination across multiple fields of inquiry into these various processes shows that one model’s “externalities” have been addressed as another’s central focus, and vice versa. The long-term development of climate and weather forecast models is one example of the improvement in accuracy that may be obtained, not just in forecast skill but also in the accessibility of physical representation and process understanding, by the combination of process models from different communities and approaches. In that development, atmospheric models were originally developed with a coarse representation of surface conditions, while land surface models originally treated the atmosphere as an external source of “forcing” conditions. The coupled land–atmosphere model is now a staple of forecast centers around the world, demonstrating accuracy in the representation of the natural system, and predictive skill, that far exceed earlier separate and uncoupled modeling efforts. Coupled atmosphere–ocean general circulation models (AOGCMs) are employed for the prediction of climate change and its impacts, and are expected to become more accurate and even more useful to decision-makers with improvements to the component representation of land surface processes.
It is a persistent challenge for modelers to reduce a problem to a tractable scope and scale, while also allowing for the emergence of detailed response patterns, using present computational methods for fully nonlinear systems. Earth’s atmosphere and land surface are tied in dynamic mutual feedback processes over multiple spatial and temporal scales, the full scope and detail of which remain difficult for us to formulate. Land–atmosphere models typically represent only a fraction of the complexity that is observed in the real system. Modeling methods attempt to address this issue from a conceptual and computational viewpoint with simplifying assumptions and parameterizations. The spatiotemporal scales of interest remain important in helping to shape the model dynamics: climate models are oriented on long-term simulations of conditions, while surface-based models must consider the rapid changes that come with the persistent fine-scale redistribution of water and the actions of humans on their environment. Some of the highest-resolution global climate models employ simulation grid cells at a scale of 10-100 kilometers; the entire domain of a land surface process model, formulated for an area that might be of interest to natural resource planners and policy-makers, could fit into a single climate model grid cell several times over. These spatial and temporal scales of interest overlap at the domain size and scope of weather modeling over land areas for both forecasting and system understanding.
Accordingly, the development of such coupled process models is ongoing. Among the goals of continued work in this area is the application of land–atmosphere models to predictions of extreme hydrologic events such as droughts and floods. At the same time, advances in ecological process modeling and remote sensing can provide valuable additional information to these efforts where knowledge gaps exist. Data assimilation efforts in weather forecasting already employ remote sensing and other observational products at various levels of processing and accuracy, demonstrating one way for models to maintain fidelity with natural systems. Another avenue for continued improvement, especially by the incorporation of methods from different specializations, is found in the way the land surface is represented in these climate- and weather-oriented modeling systems. Decades of remote sensing technology and datasets now support landscape-scale ecological models that describe the states and dynamics of land cover and human land use, as well as disturbances to those from both natural and anthropogenic sources. All of these aspects of the land surface provide feedback to the atmosphere through various parameters and processes, with a range of impacts on the surface radiative balance, land–atmosphere heat exchange, and the local and regional hydrologic cycle.
08 December 2012
Dissertation Proposal Excerpt, part 2
A few more paragraphs excerpted from my Ph.D. Dissertation Proposal, currently in preparation. This is a continuation of the narrative that I began posting yesterday.