Department of
Land Resources and Environmental Sciences
Montana State University
P.O. Box 173120
Bozeman, MT 59717-3120
Dr. Geoffrey Poole
Tel: (406) 994-5564
Fax: (406) 994-3933 gpoole@montana.edu
Dr.
Robert Payne
Tel:
(406) 994-7197
rpayn@montana.edu
Department of
Computer Science Undergraduate Research
Isaac
Griffith Stuart Winslow Ryan Nix
Computational
Ecology Research Group
Welcome
Computational
Ecology
The
computational ecology research group is concerned with the advancement
of scientific processes and research of new techniques to model
environmental processes. We focus on modeling scientific
information in the fields of hydrology and ecological landscapes.
An
on-going collaboration between the Department of Land Resources and
Environmental Sciences and the Computer Science Department is exploring
new challenging techniques and algorithms to model and represent
complex ecosystem processes such as nitrogen cycling in streams.
Many
opportunities exist for ecologists and computer scientists.
In particular, we are interested in:
Visualization
techniques of complex, multi-dimensional flow networks
Dependency
tracking algorithms
Efficient
multi-threading models
Multi-scale
representation of ecological data
The
following is a visual representation of a hydrological landscape.
It shows a simple network of various (multi-dimensional)
currencies fluxing through intrconnected nodes.
An Introduction
to NEO
Network
Exchange Objects (NEO) is a new software framework designed to
facilitate simulation modeling of ecosystems, where the ecosystem is
represented as a Scape,
through which Resources (e.g., water, heat, nutrients, and energy) are
fluxing.
A
current example of a NEO-based model is the Water Resources Exchange
Network model (WREN), which is designed for modeling river floodplain
ecosystems. First, water movement and exchanges are simulated
through surface and subsurface flows. This water movement is
organized into exchanges among a network of storage cells, where each
cell represents specific location upon or within the
floodplain.
Second, additional resource networks link carbon, nitrogen, and oxygen
movement to the water movement, because water tends to be the dominant
media for resource movement in rivers. Biological processing
of
these resources is also simulated in these networks, where carbon,
nitrogen, and oxygen in the water are consumed or produced by
biochemical reactions in river microbial communities. In this
way, combined resource simulation within the WREN model provides a
holistic view of water quality and quantity across a floodplain,
benefitting both managerial and scientific pursuits regarding rivers.
Carbon,
nitrogen, and oxygen are strong influences on water quality and
subsequently influence the ecosystem services a river can provide, such
as: general recreation; drinking water supply; or recreational and
commercial fisheries. Scientists use models such as WREN as a
highly detailed definition of how we think the physics and life within
the system support these services. The scientific goal is to
find
locations or times that the model does not work, and use this
information to design experiments that eventually lead to model
enhancements. Meanwhile, managers may use the same models to
assist with decisions about river systems. For example, WREN
may
be used to determine how controlled changes to water flow (perhaps from
a dam outlet), might affect the quality and quantity of water across a
downstream floodplain. Of course, managers and scientists
need to
communicate about scenarios where or when a given model is not likely
to work, in order to know where or when that model is useful for making
sound decisions.
Because
these models are highly detailed and often need to be executed
thousands or tens of thousands of times, their application requires
substantial computational power. For example, though WREN is
usually applied to individual floodplains (perhaps ~5 km long and ~1 km
wide), it simulates resource behavior at sufficient detail to require
processing power similar to global-scale water supply models.
We
are currently using the RMSC and WREN to simulate the Nyack floodplain
near Glacier National Park and Flathead Lake in northwest
Montana. Our current model executions focus on simulating
floodplain response to changes in river flow expected from multiple
climate change scenarios. The goal of this work is to
understand
what aspects of water quality and quantity may change in the upcoming
decades, and thus provide information on what aspects of this change
are most likely to influence management decisions. This
effort is
likely to require a sensitivity analysis, where multiple concurrent
executions across parallel processors will explore which combinations
of inputs have the largest effect on model output. For
example, a
sensitivity analysis may provide information on which aspects of
climate change are most likely to change the amount of water, carbon,
nitrogen, and oxygen across the floodplain. The RMSC provides
parallel computing power that would otherwise be unavailable to execute
these simulations, thus greatly improving the efficiency of this work.