acedp november 2010 beijing workshop bond
TRANSCRIPT
Additional information to support river health assessment Dr Nick Bond
Talk outline
• Some additional ideas to think about in relation to river health assessment. – Things we may have overlooked in other talks
– Not necessarily linked to one another
1. Quality assurance
2. Site selection
3. Pressure indicators
4. Classification
5. Refinement and adaptation
1. Quality assurance
• Managers and the public expect a high level of
confidence in assessments.
• Requires
– A scientific basis for the assessment
– High standards of quality control & quality assurance (QA/QC)
• Field and laboratory work
• Data analysis and data storage
• Data interpretation
– Careful evaluation of the results (do they make sense?)
Sources of error in assessment
Scientists like to
make things
complicated
QA/QC – staff training
• Appropriate training and testing of
staff involved in field and
laboratory work
• EPA (Australia) ID 20% of
invertebrate samples twice for
consistency.
• Development of standard
operating procedures
• May involve collaboration between
organisations - universities,
research groups, central &
provincial govt.
Site selection
• Site selection strongly
influences assessment
results
• River health assessment
best served by random or
stratified random site
selection
– Guided by classification
and assessment of threats
• Worthwhile to develop
clear guidelines before
going into the field
found on the EMAP website:
(http://www.epa.gov/nheerl/arm/designpages/design&analysis.htm).
FIGURE 3-1. Examples of two-dimensional probabilistic sampling
designs.
Simple random samplingSimple random sampling
Stratified random sampling
strata
Stratified random sampling
strata
Quantifying trends in resource condition is often an important objective for regional assessments.
Although there are different approaches for allocating sampling effort over time, only two are
covered in this document: permanent station and serially alternating (Rathbun 1999). Permanent
station approaches use a random sample of n sites that are all sampled during each time interval.
This option provides the least spatial coverage but may provide the highest temporal resolution
of trends, if temporal autocorrelation is weak. It is noteworthy that if resources allow sampling
the entire population of large river segments, a permanent station temporal design is appropriate
3-8 Concepts and Approaches for the Bioassessment of Non-wadeable Streams and Rivers
Chapter 3.0
found on the EMAP website:
(http://www.epa.gov/nheerl/arm/designpages/design&analysis.htm).
FIGURE 3-1. Examples of two-dimensional probabilistic sampling
designs.
Simple random samplingSimple random sampling
Stratified random sampling
strata
Stratified random sampling
strata
Quantifying trends in resource condition is often an important objective for regional assessments.
Although there are different approaches for allocating sampling effort over time, only two are
covered in this document: permanent station and serially alternating (Rathbun 1999). Permanent
station approaches use a random sample of n sites that are all sampled during each time interval.
This option provides the least spatial coverage but may provide the highest temporal resolution
of trends, if temporal autocorrelation is weak. It is noteworthy that if resources allow sampling
the entire population of large river segments, a permanent station temporal design is appropriate
3-8 Concepts and Approaches for the Bioassessment of Non-wadeable Streams and Rivers
Chapter 3.0
sediment plume
Data analysis and storage
• Relatively complex datasets
– Large # of variables
– Taxonomy may change over
time
– Biological indicators often
derived from a series of
calculations
Importance of good data
management typically
overlooked whereas good
QA/QC demands it
Evaluating results
• Indicators are not perfect – sometimes results
will conflict with expectations
– Requires transparent process of review and
refinement.
– Often several possibilities
• Sampling error, natural disturbance effects (e.g. floods),
localised pollution event.
– Local expertise/input is valuable
– Look to explain conflicting patterns rather than
simply discarding the result.
2. Pressure indicators
Disturbance (pressure)
Ecolo
gic
al in
dic
ato
rs
• Good indicators respond predictably to
disturbance gradients
Pressure indicators
• Good indicators respond predictably to disturbance gradients
• Why not just measure the disturbance gradient to predict river health?
Disturbance (pressure)
Ecolo
gic
al in
dic
ato
rs
Pressure indicators
• Good indicators respond predictably to disturbance gradients
• Why not just measure the disturbance gradient to predict river health?
Disturbance (pressure)
Ecolo
gic
al in
dic
ato
rs
Two answers:
1. Sometimes we do
Pressure indicators
• Good indicators respond predictably to disturbance gradients
• Why not just measure the disturbance gradient to predict river health?
Disturbance (pressure)
Ecolo
gic
al in
dic
ato
rs
Same
pressure
different
management
Improved
management
Two answers:
1. Sometimes we do
2. Ecosystem health can
improve if best practice
management actions are
implemented
Example - urbanisation
• Impacts of urbanisation best
predicted by % effective
imperviousness
• Water sensitive urban design
reduces imperviousness
Forest
Grasslan
d Wetland
agricultur
e urban
Pressure indicators as an element of river
health assessment
• Generally desktop based
analyses
• Data increasingly
available at fine scales
• May incorporate a range
of threats (e.g.)
– Population density
– Agricultural production
– Agricultural water use
– Upstream/downstream
impoundments
Pressure indicators as an element of river
health assessment
• Help Identify ‘high risk’ areas
• Guides site selection – stratification by land-use
• Testing indicators
• Effective communication tool
River Disturbance Index =
Upstream Storage Volume
weighted by Catchment Area
3. Classification
• River health assessment must account for
natural variation in rivers
• River classification an important step in most
assessment programs
• Identifying appropriate indicators
• Scoring (targets and thresholds)
Requirements of a River
Classification to support
river health assessment
1. Discriminates ecologically meaningful variation in indicators
2. Based on data unaffected by human disturbance
3. Stable groupings that assist indicator selection & target setting
4. Derived using explicit, repeatable and transferable methods
5. Applicable to a range of stream types
Other applications of classification
• Identifying ‘comparable’ catchments outside of
the study basin
– Assessing transferability of published data to/from
different regions based on similarity in classification
Caveats on classification results
• Important to confirm classification is biologically
meaningful
– Most software will produce ‘classes’ even if these
are extremely similar to one another.
– Possibly try a range of classification algorithms and
input variables to determine sensitivity of the results
to particular attributes (e.g. soils, temperature,
rainfall).
USGS Hydro 1K global dataset
• DEM stream network
• Nested subcatchment codes (Pfaffstetter,
1989) • Allows linking of catchment data without complex
geoprocessing
• not as fine-scaled as some catchment
delineations so less well suited to local
analyses.
• Climate, ecosystem type layers also available
at global/national scales
Minimum catchment resolution
Supporting datasets – ecosystem types
4. Refinement and adaptation
• River health monitoring and environmental flows
assessment strategies will evolve over time
• 15+ years in Australia (>20 years in USA)
– Changes in indicators
– Refinement of targets and scoring systems
– Continues to evolve
• Many good Chinese examples of applying these
methods in a research context
• Incorporating into management is the next step
The process
Field trial Assess indicator
sensitivity to
disturbance
gradient
Identify suite of
potential indicators
Land-use
assessment to
define
disturbance
gradient
Adopt appropriate
standard
Did the indicator
respond as
expected
Consider for
inclusion in
scorecard
Can thresholds and
targets be established
from the data?
Do standards already
exist (chinese or
international)
Review
indicator
Include in
scorecard
River
Classification
yes
No
Consider for
future
programs
No
Yes
yes No
Conceptual
models