National Forest Health Monitoring Riparian Pilot John Day Basin

1
National Forest Health Monitoring Riparian
PilotJohn Day Basin

• Mary Manning and Vicente Monleon
• USDA Forest Service

2
Partners

• USFS Forest Inventory and Analysis (FIA)
• EPA Environmental Monitoring and Assessment
  Program (EMAP)
• Oregon Department of Environmental Quality (DEQ)
• USFS Pacfish/Infish Biological Opinion
  Effectiveness Monitoring Team (PIBO)

3
Purpose

• Due to the small area and narrow linear features
  of most riparian systems west of the Mississippi
  River, many riparian systems are under sampled or
  un-sampled using the FIA grid and their
  definition of forest land use.
• There are boundary problems, definition problems,
  specific problems to riparian area.
•  

4
(No Transcript)
5
Objectives

• Develop and test a survey design and two response
  (plot) designs for the purpose of estimating
  riparian attributes. This includes condition and
  extent. Not restricted to forest as defined by
  FIA.
• Collaborate with other agencies who collect
  instream, hydrologic and vegetative data
  (particularly those federal programs, such as EPA
  EMAP, that use probabilistic sampling).

6
Objectives (cont)

• Long term, repeated measures, change detection
  monitoring
• Monitor change in amount, condition and extent of
  riparian vegetation
• Broad scale monitoring (Forest, State)
• Need identified by WY State Dept. of Forestry in
  1999. Not currently part of a broad scale
  inventory (FIA)

7
Data Quality Act Daubert Principles

• Population estimates must have known confidence
  (or error) associated with them (SE, CI etc.)
• Method must be repeatable, transparent to an
  independent, third party user (QA/QC),
• Statistically rigorous (high SN)

8
Target Population

• Need concise, non-ambiguous definition to monitor
  changes through time and space with a random
  sample.
• Use of probabilistic design critical for making
  inferences about the population.
• All items within the population must have a
  known probability of being selected.
• Definition is statistically based-what can be
  monitored over time with no bias.

9
John Day pilot definition

• The vegetation from the edge of the bankfull
  channel or greenline (whichever is closest to the
  active channel) to 200 ft. horizontal (with slope
  correction) distance perpendicular to the
  channel.

10
Survey and Response Design (EMAP)

• Survey design process of selecting sites at
  which a response will be determined
• Probability model for inference is based on the
  randomized selection process
• -Spatial and temporal component
• Response design process of obtaining a response
  at a site
• A single index period during a year
• Plots or reaches (with subsamples)

11
Methods

• Survey Design Sites were selected from a
  spatially balanced list frame using the
  1100,000 USGS stream layer (EMAP procedure).
  Co-located with EPA EMAP and OR DEQ sites.
• Response DesignTwo modified FIA Phase 3 plot
  designs were tested 1) circular and 2)
  rectangular subplots- both types placed end to
  end, and aligned perpendicular to the stream at
  each site.
• Data collected Tree count and measurements, down
  wood, species composition by layer to 1 cover

12
Generalized Random Tessellation Stratified
Design(EMAP)

• Spatially balances the sample across the resource
  (improved precision)
• Overview of process for linear resource
• Use stream layer (RF3 or NHD)
• Weight by stream order to balance sample
• Hierarchically randomize points to place them on
  line, assigning each point unit length
• Select a systematic sample from the points using
  a random start

13
2001 Catskills Pilot Plot designs
14
Edge of subplot at greenline or bankfull,
whichever is closest to stream
Circular Plot Design
15
Rectangular Plot Design
Edge of subplot at greenline or
bankfull, whichever is closest to the stream
3 ft
16.5ft
16.5ft
16
Constraints

• Must have fixed area plot with area of the 4
  subplots totaling 1/6th acre
• This is necessary for consistency in area
  estimation (amount and extent)
• Must map condition classes within the 4 plot
  cluster
• Must sample the national core variables
• (e.g., tree list, dbh)

17
Methods

• The crew sampled 40 EMAP sites. EMAP research
  ecologists provided stream reach sampling
  locations and detailed vegetation data from the
  same sites.
• Oregon DEQ sampled the sites using EMAP Surface
  Waters Physical Habitat sampling protocols, which
  include coarse vegetation sampling stream and
  floodplain sampling.

18
(No Transcript)
19
Methods (cont.)

• PIBO crewperson sampled 15 reaches using their
  riparian vegetation protocols
• Greenline and valley cross section transects (to
  augment the P3 plot data to characterize riparian
  extent, composition, structure and function).

20
PIBO design
20 x 50 cm
21
Analysis

• Comparing plot shapes.
• Comparing 1 versus 2 sides.
• We considered that sampling both sides takes more
  time - time that could be used to increase the
  sample size by sampling other sites.

22
Analysis (cont.)

• Compare the results from the riparian inventory
  with the regular, grid-based FIA and FHM
  inventories.
• The latter inventories provide a population
  average, since they are not geared toward any
  particular subpopulation.

23
Analysis

• Met with partner statisticians (12/03)
• Enter and analyze data (Spring 04)
• Comparative analysisspecies richness
• Which design captures the greatest number of
  species
• Publish findings in report (Summer 04)

24
Analysis (cont)

• Define population parameters (e.g., species
  richness, diversity, etc)
• Develop algorithms to estimate population
  parameters. 
• Collaborate on comparative analysis of the 4 data
  sets.
• Sample unit for comparison is the support
  reach.

25
Initial Findings

• Very preliminary results suggest
• Vascular plant species richness is much greater
  in the riparian plots (gt150 species),
• Human impacts (in particular, roads) are more
  severe. This will argue for the need for a
  riparian inventory.

26
Limitations of the data set

• Using the EMAP sample frame, the scope of
  inference federal lands adjacent to permanent,
  wadeable streams.
• Doesnt include large rivers, intermittent and
  ephemeral streams
• Cant pool these data with other data sets and
  make inferences about the population (unequal
  selection probabilities, different target
  population).
• Broad scale-not appropriate at fine scale without
  intensification.
• Over samples narrow and under samples broad
  valley bottom riparian vegetation.

27
Options-Next Steps

• WY pilot with Intermountain West FIAdevelop more
  accurate aerial detection methods (working with
  RSAC)
• Add ons to base design (e.g., PIBO
  greenline)
• Blind intensification of survey design (more
  randomly placed sample points) or stratified
  random design (by landformvalley bottom)