The concept of a Master Sample: Lower Columbia Example

1
The concept of a Master SampleLower Columbia
Example

• Phil Larsen
• Pacific States Marine Fisheries Commission
• c/o USEPA
• 200 SW 35th St.
• Corvallis, OR 97330
• Email Larsen.Phil_at_epa.gov
• Phone [phone]

2
Background

• Many agencies are interested in regional scale
  monitoring of stream networks and watersheds,
  using similar attributes and similar/same
  protocols
• Cant afford to monitor everywhere (i.e., cant
  census)
• Conducting a sample survey a sound way to
  characterize a region by monitoring a
  representative set of sites

3
LCFRB Habitat Monitoring Key Entities

• Federal
• 5 agencies
• State
• 5 agencies
• Local
• At least 12

4
From Stevens and Olsen, 2004Spatially balanced
sampling of natural resources. JASA 99 262-278
Sampling the gamut of natural resources requires
a technique that can

• select a spatially balanced sample of finite,
  linear, and areal resources with patterned and
  possibly periodic responses
• use arbitrarily variable inclusion probability
  with imperfect frame information, in the presence
  of substantial nonresponse.

5
Spatially Balanced Sampling

• GRTS Generalized Random-Tessellation Stratified
  design
• Incorporates randomization
• Is spatially balanced
• Creates an ordered list of sites
• See www.epa.gov/nheerl/arm for details

6
(No Transcript)
7
Combining Data

• Data from different sample surveys can be
  combined if certain design principles are
  followed
• Similar target populations are well defined
• The same frame represents the target populations
  (i.e., use the same digital stream trace
  coverage)
• Randomization in site selection
• Common protocols are used to measure stream
  attributes

8
A Desirable Goal

• Create a design that allows and facilitates
  integration up front rather than after the fact.
• Current GRTS design allows selection of a
  master sample that can be subset to meet
  specific needs at a variety of spatial scales

9
Washington

• Frame USGS 1100,000 scale digital hydrography
  (perennial and intermittent)
• Selected a statewide sample of about 125,500
  points on the network as an ordered list
• Any subset drawn in order is a spatially balanced
  sample
• Distance between points is approximately 1 km.
• Assigned a set of classification variables for
  sorting the sample

10
The Master Sample(An ordered list of sites)

• Ecoregion
• Level 2
• Level 3
• Level 4
• USGS Hydro units
• By number
• By name
• WRIA
• Name
• Area
• other geo-referenced variables

• Site ID
• Location (Lat/Long)
• County
• Quad 100
• Strahler order
• Perennial/intermittent

11
LCFRB Stratification SchemeTable 6

• Ecoregion
• WRIA
• Subbasin (different from USGS subbasin?)
• Physiographic zone (different from ecoregion?)
• Stream order

12
WA Master Sample File
13
Example

• Pick a statewide sample
• 50 sites in each of the 5 OR DEQ reporting
  regions
• Equal balance among
• headwater (1st order)
• wadeable (2nd, 3rd order)
• non-wadeable (4th order) streams
• About 17 sites per stratum

14
Example

• Solution
• Create stream groups headwater, wadeable,
  non-wadeable from stream order assignments
• Sort the list by DEQ region and stream group
  within DEQ region, maintaining the original order
  of sites
• Pick the first 17 sites on the list within each
  DEQ region and stream order group

15
(No Transcript)
16
Intensify in three sub-basins
17
Intensify in one National Forest and one BLM
District
18
Combine stream network and watershed selection
19
Washington Master Sample Example Designs
20
TEMPORAL DESIGNAUGMENTED SERIALLY ALTERNATING
Note although not indicated here, within year
revisits to sites are important to evaluate
seasonal variation and field crew performance.
Figure courtesy of Scott Urquhart.
21
Example Design Characteristics

• Equal of sites by Strahler order categories
  1st, 2nd, 3rd
• State-Wide design
• 1054 unique sites allocated to 6 panels
• Panels 1 to 5 visit once every 5 years
• Panel 6 visit annually
• 1550 site-visits over 5 years
• 310 sites per year

22
State-Wide 5 Year 6 Panel Design
23
State-Wide Panel 1
24
Example Design Characteristics

• WRIA design
• 17 sites each WRIA for total 1054 sites
• 1054 unique sites allocated to 6 panels
• Panels 1 to 5 visit once every 5 years
• Panel 6 visit annually
• 1550 site-visits over 5 years
• 310 sites per year
• Intensive design
• 50 total sites in 3 WRIAs
• 17 sites from WRIA 5 year design and 33
  additional sites

25
WRIA 5 Year Design
26
WRIA Design Panel 1
27
Intensive Studies WRIA 5 Year Design
28
LC Master Sample
29
Number of sites by WRIA and Stream Order
30
Panel 1 54 status sites (from table 8)
31
Panel 2 18 annual sites (from table 8)
32
Lewis 30 special interest
33
Diagonstic 30 Germany
34
LCFRB Habitat Status Monitoring (p. 2)

• Landscape Census/complete coverage
• Watershed
• Uplands/Hill slopes
• Wetlands
• Stream Corridor Sample survey
• Channel Conditions
• Riparian zone
• Floodplain
• Water
• Quantity ?
• Quality

35
LCFRB Stratification SchemeTable 6

• Ecoregion
• WRIA
• Subbasin (different from USGS subbasin?)
• Physiographic zone (different from ecoregion?)
• Stream order

36
Salmon Recovery Priority Tiers (p. 24)Some
questions

• Four tiers
• Are these mapped?
• Can they be identified in the office or field?
• Tiers as strata?
• Tier 1 as an index stratum for trend sites?

37
LCFRB Sample Type (p. 21 table 8)

• Survey Status
• Master sample
• Index Trend
• Master sample via an index stratum, e.g., Tier 1
  sites
• Diagnostic
• Master sample possible if diagnostic strata can
  be identified
• Focal site specific project evaluation
• Probability vs. judgmental (or hand selected)
  sites

38
Stream Habitat Sampling LevelsTable 7 (p. 26)

• Indicator
• Remote/office
• Reconnaissance
• On the ground rapid assessment
• Inventory
• Reach/habitat
• Intensive
• Site
• Concept of Nested samples

39
Landscape Scale

• Compile at 12 yr intervals
• Metrics
• Road density
• Mass wasting
• Impervious surfaces
• Land use/land cover
• Channel migration
• Wetland availability
• Floodplain connectivity
• Key question who will oversee the collective
  landscape data base and its updates?
• Need to improve the stream network frame
• 1100 K NHD
• 124 K DNR

40
Utility of a Master Sample

• An exploratory tool to examine different site
  allocations easily.
• A framework for an actual integrated,
  multi-agency state-wide or regional monitoring
  program.

41
(No Transcript)
42
(No Transcript)
43
Flexibility/Potential

• Add classification variables by assigning sites
  to new classes
• Requires georeferenced data
• Facilitates integration across programs
• What sites already being monitored by another
  agency?
• Encourages
• Communication about monitoring designs among
  agencies
• Development of common databases to share data
  easily
• Agreement on common protocols

44
To Make This Work, We Need

• A standard frame that everyone would use, even if
  imperfect
• A process for correcting the frame as errors
  arise (i.e., frame maintenance)
• Institutional Management
• Database
• Frame and corrections
• GIS coverages for classification/stratification
• Monitoring programs who is doing what where?
• Monitoring results

45

• Phil Larsen Larsen.phil_at_epa.gov
• Tony Olsen Olsen.tony_at_epa.gov
• Don Stevens Stevens_at_stat.orst.edu
• Oregon State University
• www.epa.gov/nheerl/arm

46
(No Transcript)
47
(No Transcript)
48
(No Transcript)
49
Three Monitoring Scales

• Landscape
• Stream Corridor
• Water

50
(No Transcript)
51
(No Transcript)
52
(No Transcript)
53
(No Transcript)
54
(No Transcript)