measuring A framework to develop useful landscape indicators for aquatic responses

advertisement
A framework to develop useful
landscape indicators for measuring
aquatic responses
David Theobald,
John Norman, Erin Poston,
Silvio Ferraz
Natural Resource Ecology Lab
Dept of Recreation & Tourism
Colorado State University
Fort Collins, CO 80523 USA
2 February 2005
Project context

Challenges of STARMAP (EPA STAR):
Addressing science needs Clean Water Act
 Integrate science with states/tribes needs



From correlation to causation
Tenable hypotheses generated using
understanding of ecological processes
Goal: to find measures that more closely represent our
understanding of how ecological processes are
operating
Typology of Landscape Context for
Aquatic Response Indicators
Sample/site
1.
a.
b.
covariate at x,y location (e.g., geology, elevation, population density)
covariate nearby x,y location (e.g., housing density at multiple scales)
Watershed-based
2.
a.
Summary or average of covariates within watershed defined from “pourpoint on up”. “lumped”, overlapping, hierarchical,
Spatial relationships between locations
3.
a.
b.
Euclidean (as the crow flies) distance between points
Euclidean (as the fish swims) hydrologic network distance between points
Functional connectivity between locations
4.
a.
b.
c.
d.
Direction of important ecological process (e.g., flow direction)
Scale dependent
Distances not symmetric, stationary  violate traditional geostatistical
assumptions!?
From watersheds to tessellation of Reach Contributing Areas (RCAs):
local vs. Amount of contributing area, flow volume, etc.
From watersheds/catchments as
hierarchical, overlapping regions…
River continuum concept (Vannote et al. 1980)
“Lumped” or watershed-based analyses





% agricultural, % urban (e.g., ATtILA)
Average road density (Bolstad and Swank)
Dam density (Moyle and Randall 1998)
Road length w/in riparian zone (Arya 1999)
But ~45% of HUCs are not watersheds
EPA. 1997. An ecological assessment of the
US Mid-Atlantic Region: A landscape atlas.
Southern Rockies Ecosystem Project.
2000.
Benda et al. BioScience 2004
Dominant downstream
process
Upper and lower Colorado Basin
Flows to downstream HUCs
… to Tessellation of Reach
Contributing Areas (RCAs)
Automated delineation

Inputs:
stream network (from USGS
NHD 1:100K)
 topography (USGS NED, 30
m or 90 m)


Process:
“Grow” contributing area
away from reach segment
until ridgeline
 Uses WATERSHED
command

“true”
“adjoint”
catchments
catchments
Reaches
(segments)
Reaches are linked to catchments
1 to 1
relationship
 Properties of
the watershed
can be linked
to network for
accumulation
operation

RCA example

US ERF1.2 & 1 km DEM: 60,833 RCAs
Hydrologic
distance:
-Instream
-Up vs. down?
FLOWS
Overlapping
watersheds
Accumulate
downstream
FLOWS (and
SPARROW)
Stand-alone
watershed
Watershed-based
analyses (HUCs)
Tesselation of true,
adjoint catchments
?
Zonal
Accumulate
Process/Functional
Up/down (network)
Land (basins)   Stream
Watersheds
HUCs/WBD
Reach Contributing Areas (RCAs)
Grain (Resolution)
Challenges: conceptual & practical



Definition of a watershed
Overland surface process vs. in-stream flow process
Scale/resolution issues


Artifacts in data



Attribute errors, flow direction, braided streams
Linking locations/points/events to stream network


E.g., different answers at 1:500K vs. 1:100K vs. 1:24K
Reach-indexing gauges, dams?
Very large databases
GIS technology innovations and changes
Asymmetric Kriging for Stream Networks

Developed by Jay Ver Hoef, Alaska
Department of Fish and Game (Ver
Hoef, Peterson, and Theobald, In
press)

Spatial statistical models for stream
networks
 Moving average models
 Incorporate flow and use
hydrologic distance
 Represents discontinuity at
confluences

Important for pollution monitoring
Flow
Need for network datastructure
within GIS  Landscape Networks!




Need to represent relationships between
features
Using graph theory, networks
Retain tie to geometry of features
Implementation in ArcGIS
Geometric Networks (ESRI – complicated, slow)
 Landscape Networks (GeoNetworks): Open, simple,
fast

Feature to Feature Relationships via Relationship Table
Up
Down
RCAs are linked together
– but spatial configuration within an RCA?
1. Ignore variability
2. Buffer streams
3. Buffer outlet
2 major hydro. processes w/in RCA
A
B
A'
B'
Legend
C'
C
#
0
outlet
nhd_rivers
catch
^_
Points
Distances
AA'
0
0.5
1
2
Miles
1. Overland (hillslope): Distance (A to A’)
2. Instream flow: Distance (A’ to O)
BB'
CC'
Flow distance: overland + instream



Hydro-conditioned
DEM (e.g., EDNA)
FLOWDIRECTION
FLOWLENGTH
Flow distance: overland



Hydro-conditioned DEM
(e.g., EDNA)
Burn stream into
FLOWDIRECTION
FLOWLENGTH
Flow distance: instream



Hydro-conditioned DEM
(e.g., EDNA)
FLOWDIRECTION
FLOWLENGTH from
outline – overland
FLOWLENGTH
Why are functional metrics important to
understanding effects of land use
change on freshwater systems?

Clearer relationship between ecological (aquatic,
terrestrial) process, potential effects (e.g., land
use change) and response


Huge (insurmountable?) challenge is that we cannot
develop traditional experimental design (manipulated
vs. controlled) because landscapes are so large and
human activities so dominant
More direct relationship between process and
measure, biologically meaningful
Tools needed
to enable
“network
thinking”
- FLOWS v0.1: ArcGIS v9 tools
-Higher-level objects 
faster coding!
- Open source
- Integrated development for
documentation
Thanks!

Comments? Questions?

Funding/Disclaimer: The work reported here
was developed under the STAR Research
Assistance Agreement CR-829095 awarded
by the U.S. Environmental Protection Agency
(EPA) to Colorado State University. This
presentation has not been formally reviewed
by EPA. The views expressed here are solely
those of the presenter and STARMAP, the
Program (s)he represents. EPA does not
endorse any products or commercial services
mentioned in this presentation.

STARMAP:

FLOWS: davet@nrel.colostate.edu
www.stat.colostate.edu/~nsu/starmap
CR - 829095
Laramie Foothills Study Area and
Sample Points
Accessibility:
travel time along
roads from urban
areas
Planned future activities

Papers


Presentations



Theobald GRTS Sept. 23
Poston
Products



Completing draft manuscripts on: GIS-GRTS, RCAs,
overland/instream flow, dam fragmentation, GeoNetworks
FLOWS tools
Datasets: RCAs (ERF1.2)
Education/outreach

Training session for FLOWS tools
Possible future activities

Dataset development




RCA nationwide with involvement for USGS NHD program
Reach indexing dams (for EPA, Dewald)
Discharge volume
Symposium: “At the interface of GIS and statistics for
ecological applications” (~January 2005)





What are the strengths and weaknesses of GIS-based and statisticalbased tools?
How can/should statisticians respond, direct, and utilize GIS-based
types of tools?
How can/should statistical tools be best integrated with GIS?
What are the needs of agencies if statistical-based tools are to be used?
When should GIS-based tools be used?
How can these two approaches best complement one another?
Landscape ecology and freshwater
systems

One consequence of this interplay [between pattern and process] is
the form of functional connectivity found in a landscape. The
landscape pattern-process linkage produces spatial dependencies in a
variety of ecological phenomena, again mediated by organismal
traits. All of the components of this framework change with changes
in scale, often in different ways. It is through the integration of these
features of landscapes and of organisms that landscape ecology can
offer new insights to freshwater ecologists, fostering a closer linking
of spatial patterns with ecological processes (Wiens 2002)

Hierarchies
Spatial heterogeneity
Processes








Overland
Instream
Functional connectivity

Constraints
Watershed to stream
Reach to reach (stream network)
Network
Spatial & temporal scales, processes
Poff, N.L. 1997
Download