Transcript Monitoring and Assessing the Condition of Aquatic Resources: Role

Statistical Perspective on the Design and Analysis
of Natural Resource Monitoring Programs
Anthony (Tony) R. Olsen
USEPA NHEERL
Western Ecology Division
Corvallis, Oregon
(541) 754-4790
[email protected]
Web Page: http://www.epa.gov/NHEERL/ARM
National Water Quality Monitoring Council:
Monitoring Framework
•
Applies to all natural resource monitoring
•
Monitoring pieces must be designed and
implemented to fit together
•
View as information system
•
National monitoring requires consistent
framework
•
Reference: Water Resources IMPACT,
September 2003 issue
Impact Article Contributors
• Framework Overview
 Charles A. Peters
 Robert C. Ward
• The Three C’s
 Abby Markowitz
 Linda T. Green
 James Laine
• Monitoring Objectives
 Charles S. Spooner
 Gail E. Mallard
• Monitoring Design
 Tony Olsen
 Dale M. Robertson
• Data Collection
 Franceska Wilde
 Herbert J. Brass
 Jerry Diamond
• Data Management
 Karen S. Klima
 Kenneth J. Lanfear
 Ellen McCarron
• Assess and Interpret
 Dennis R. Helsel
 Lindsay M. Griffith
• Report Results
 Mary Ambrose
 Abby Markowitz
 Charles Job
Monitoring Program Weaknesses
1. Monitoring results are not directly tied to management
decision making (monitoring objectives)
2. Results are not timely nor communicated to key audiences
(convey results)
3. Objectives for monitoring are not clearly, precisely stated
and understood (monitoring objectives)
4. Monitoring program not viewed/implemented as an
information system (data management, overall)
5. Monitoring measurement protocols, survey design, and
statistical analysis become scientifically out-of-date
(field/lab methods, monitoring design, data
analysis/assessment)
Communicate, Coordinate, Collaborate
• Communication: process of
conveying information; can be
one way or an exchange of
thoughts, messages, or ideas
• Coordination: process in which
two or more participants link,
harmonize or synchronize
interaction and activities
• Collaboration: process in
which two or more participants
work collectively to deal with
issues that they cannot solve
individually; partnerships,
alliances, teams
Convey
Results
and
findings
Develop
monitoring
objectives
Design
monitoring
program
• Kish (1965): “The survey objectives should determine the
sample design; but the determination is actually a two-way
process…”
• Initially objectives are stated in common sense statements –
challenge is to transform them into quantitative questions that
can be conveyed precisely to intended audience.
• Statistical perspective is key
 Know whether a monitoring design can answer the question
 Know when the question is not precise enough – multiple interpretations
Identify Monitoring Objectives
• Objectives determine the monitoring design (yet monitoring
design constrains objectives that can be met)




Usual to have multiple objectives
Precise statements are required
Objectives must be prioritized
Objectives compete for samples
• Statistical perspective helps identify






Target population
Subpopulations that require estimates
Elements of target population
Potential sample frames
Variables to be measured
Impact of precision required
Example: From Question to Objective
• What is the quality of waters in the United States?
• What is the quality of streams with flowing water during
summer in the U.S.?
• What is the biological quality of streams with flowing
water during summer in the U.S.?
• How many km of streams with flowing water during the
summer are impaired, non-impaired, or marginallyimpaired within the U.S.?
 How is impairment determined?
 What is meant by summer?
 Are constructed channels, canals, effluent-dominated streams
included?
Develop
monitoring
objectives
Design
monitoring
program
• Key components of monitoring design
 What will be monitored? (target population)
 What will be measured? (variables or indicators)
 When and how frequently will the measurements be taken?
(temporal design)
 Where will the measurements be taken? (site selection)
• Statistical perspective
 Sample frame and target population
 Survey design
Collect
field and
lab data
What is a Target Population?
• Target population denotes the ecological resource for
which information is wanted
• Requires a clear, precise definition
 Must be understandable to users
 Field crews must be able to determine if a particular site is in the
target population
• More difficult to define than most expect.
• Includes definition of what the elements are that make up
the target population
Target Population, Sample Frame, Sampled Population
We Live in an Imperfect World…
Ideally, cyan, yellow, gray squares would overlap completely
Basic Spatial Survey Designs
• Simple Random Sample
• Systematic Sample
 Regular grid
 Regular spacing on linear resource
• Spatially Balanced Sample
 Combination of simple random and systematic characteristics
 Guarantees all possible samples are distributed across the resource
(target population)
 Generalized Random Tessellation Stratified (GRTS) design
Generalized Random Tessellation Stratified
(GRTS) Survey Designs
• Probability sample producing design-based estimators
and variance estimators
• Give another option to simple random sample and
systematic sample designs
 Simple random samples tend to “clump”
 Systematic samples difficult to implement for aquatic resources
and do not have design-based variance estimator
• Emphasize spatial-balance
 Every replication of the sample exhibits a spatial density pattern
that closely mimics the spatial density pattern of the resource
Spatial Balance: 256 points
Why aren’t Basic Designs Sufficient?
• Monitoring objectives may include requirements that
basic designs can’t address efficiently
 Estimates for particular subpopulations requires greater sampling
effort
 Administrative restrictions and operational costs
• Natural resource in study region makes basic designs
inefficient
 Resource may be known to be restricted to particular subregions
• Complex designs may be more cost-effective
National Wadeable Stream Assessment 2004
Survey Design & Response Design
• Survey design is process of selecting sites at which a
response will be determined
 Which sites will be visited (spatial component)
 Which monitoring season will sites be visited (temporal
component, panel design)
• Response design is process of obtaining a response at a
site:
 When site is to be visited within a monitoring season
• A single index period visit during a monitoring season
• Multiple visits during monitoring season: e.g. monthly, quarterly
 Field plot design
 Process of going from basic field measurements to indicators
Design
monitoring
program
Collect
field and
lab data
Compile
and manage
data
• Components





Field methods (response design)
Laboratory methods
Measurement quality objectives
Quality assurance & quality control
Logistical plan and gaining site access
• Statistical perspective
 Experimental designs to determine cost-effective and scientificallydefensible response designs
 Statistical quality control
 Methods for minimizing non-response
Response Design - Fish
100
• 1-pass sampling
• Spread effort throughout
reach
• Get “common” species in
approx. relative
abundance
80
Species
Richness
60
40
(% of Maximum)
20
0
0 10 20 30 40 50 60 70 80
Stream Length
(Channel Width Units)
Response Design: Benthos and Periphyton
D
C
K
X-site
J
I
L
H
G
L
E
F
C
B
L
R
R
R
C
Distance between transects=4 times
mean wetted width at X-site
SAMPLING POINTS
• L=Left C=Center R=Right
• First point (transect B)
determined at random
• Subsequent points assigned in
order L, C, R
Total reach length=40 times mean wetted width at X-site (minimum=150 m)
C
A
US Forest Service
Forest Inventory
and Analysis (FIA)
Plot Design
Minimizing Non-Response: Prairie Potholes
• Landowner contact procedure
 Obtain owner list from USDA ASCS local office
 Cover letter explained study, random selection, measurements,
walking access only, timing/duration visit, offer to honor special
owner conditions
 Consent form
 Map of identifying wetland to be visited
 Telephone contact 2-4 weeks after letter – list of FAQs and
answers provided to personnel
 Second letter 5-6 weeks after initial letter
• Access rates: private land 42%
• 25% of access approvals required multiple contacts
• From Lesser et al (2001)
Compile
and manage
data
Collect
field and
lab data
• Components: compile and manage data





Data entry
Database development
Metadata
Data preservation
Data discovery and retrieval
• Statistical perspective




Statistical QA checking of data
Access to auxiliary data used in statistical analyses
Influence retrieval and database design
Importance of preserving design information
Assess
and interpret
data
Examples
• STORET modified to include survey design information
 Which sites are part of the survey design
 Stratification, weights, cluster variables
• USGS NWIS and NWISWeb
 NWIS focus on input/site specific (typically time focus)
 NWISWeb focus on retrieval (typically spatial focus)
• National Resource Inventory’s analysis database
 Statistical imputation for missing data
 Statistical creation of pseudo points
• Incorporate known information
• Link across years for consistency
 Determination of single weight for each point in database
 Results in a single, consistent database for 1982, 1987, 1992, …
that is easy to use for statistical analyses
Compile
and manage
data
Assess
and interpret
data
Convey
Results
and
findings
• Derived indicator construction
• Statistical Design-Based estimation
• Statistical Model-assisted and model-based estimation
 Inference to unsampled locations
 Spatial pattern inference (or where is the map!)
• Semi-empirical modeling
 Incorporating physical processes
 Empirical statistical modeling using auxiliary data
Design-Based Population Estimation
• Scientific inference from sample to population
• Minimizes assumptions used in the inference process
• Relies on principles of statistical survey design and
analysis
• Natural resource programs who use





Forest Inventory & Analysis
National Resource Inventory
National Wetland Status and Trends Program
National Agricultural Statistics Service programs
Environmental Monitoring and Assessment Program (EMAP)
Estimating Site Occupancy Rates
• MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A.
Royle, and C. A. Langtimm. 2002. Estimating site occupancy rates
when detection probabilities are less than one. Ecology 83:22482255.
• Likelihood based model for estimation
 Assumes simple random sample of sites
 Similar to closed-population, mark-recapture model
 Estimate probability of occupancy and probability of detection
• Estimation with complex survey designs
 Maximum likelihood as before
 Likelihood must incorporate survey design
• Stratification
• Unequal probability of selection
• Cluster sample
Northeast Lake Fish Species PAO
Brook Trout
Chain Pickerel
Northern Pike
White Sucker
Adj PAO
Unadj PAO
American Eel
Brown Bullhead
Yellow Bullhead
Rock Bass
0
0.2
0.4
0.6
Proportion Occupied
0.8
Statistical Model-assisted and Model-based
Estimation
• Improve estimation based on complete coverage
information
• Adjustment for non-response at the site level
• Small area estimation
• Spatially-explicit model of probability of impairment
 Identification of “hot spots” likely to be impaired
• Will see increased use of these techniques
Semi-parametric Small Area Model:
Northeast Lakes ANC prediction for HUCs
• J. Breidt, J. Opsomer, G.
Ranalli, G. Claeskens, G.
Kauermann
• Colorado State University
STARMAP research program
sponsored by USEPA STAR
grants program
Semi-empirical Modeling: USGS NAWQA
Estimated nitrogen export
(kg/km2/yr) for watersheds
of the conterminous United
States.
• SPARROW relates in-stream
water-quality measurements to
spatially referenced
characteristics of watersheds,
including contaminant sources
and factors influencing
terrestrial and stream transport.
• The model empirically
estimates the origin and fate of
contaminants in streams, and
quantifies uncertainties in
these estimates based on
model coefficient error and
unexplained variability in the
observed data.
Assess
and interpret
data
Convey
Results
and
findings
Develop
monitoring
objectives
• Questions to ask when planning reporting




What is objective for communicating the results?
Who is the target audience?
What is message want to convey?
What formats will be used to convey the message?
• Statistical perspective




Clarity on scope of inference: target population/sampled population
Reporting of precision for results
Construction of statistical tables
Construction of presentation quality statistical graphics
IBI Results
Geographic Distribution
3%
10%
35%
14%
31%
28%
(Insufficient
Data)
41%
26%
Western Appalachians
North-Central Appalachians
22%
21%
10%
15%
28%
34%
Valleys
11%
35%
Ridge and Blue Ridge
Estuarine Stressor Comparison
Benthic invertebrate condition
Louisianian Province
Virginian Province
Degraded
18 ± 8%
Degraded
30 ± 6%
Undegraded
82 ± 8%
Undegraded
70 ± 6%
Condition
Unknown
10%
Habitat 14%
Metals 42%
Unknown
39%
Low Dissolved
Oxygen 49%
Low D.O.
Contaminants 28%
Contaminants 10%
Toxicity 4%
Both
2%
Stressors Associated with Degraded Condition
MAIA: Relative
Risk Assessment
RR 
Pr(Poor BMI, given Poor SED)
Pr(Poor BMI, given OK SED)
“The risk of Poor BMI is
1.6 times
greater in streams with
Poor SED
than in streams with OK
SED.”
Lake Ontario Diporeia Spatial Pattern
Summary
• Statistical perspective is pervasive throughout the
monitoring framework
• Substantial advances in incorporating statistical
perspective in monitoring have been made during the last
half of the 20th century
• Many statistical methodology advances are on the
horizon that will improve monitoring cost-effectiveness
• Incorporating a statistical perspective throughout the
development and implementation of a monitoring
program is no longer optional – it is essential
When will natural
resource monitoring
programs be able to
support an Environmental
Statistics Briefing Room?