HomeMy WebLinkAboutsummary
Ecologi B) cal (EFSA Flows Science Advisory Board
Meeting Summary – February 21, 2012
Archdale Buildin , Rg
X
aleigh NC
APPROVED for distribution 4/21/2012
Attendance
sion
line)
Members
Donnie Brewer, Environmental Mgt Commis
ice (on
y
Mark Cantrell, US Fish & Wildlife Serv
sitBob Christian, East Carolina Univer
John Crutchfield, Progress Energy Carolinas
mission
Tom Cuffney, U.S. Geological Survey
Linda Diebolt, Local Governments
Chris Goudreau, NC Wildlife Resources Com
Jeff Hinshaw, NC Cooperative Extension
esources
d
Jim Mead, NC Division of Water R
Sam Pearsall, Environmental Defense Fun
Judy Ratcliffe, NC Natural Heritage Program
aime Robinson, NCAWWA‐WEA
ay Sauber, NC Division of Water Quality
esources
J
J
Bill Swartley, NC Division of Forest R
Alternates
Cat Burns, The Nature Conservancy
Vernon Cox, NC Department of Agriculture
n
Sarah McRae, US Fish & Wildlife Service
teve Reed, NC Division of Water Resources
an Stancil, NC Wildlife Resources Commissio
, NC Division of Water Resources
S
V
Fred Tarver
C Division of
on Rayno
arah Young
N Water Resources
D
S
Duke
Guests (Onsite)
ature Conservancy/
n of Water Quality
Phillip Jones, RTI
Kimberly Meitzan, The N
oIan McMillan, NC Divisi
ennifer Phelan, RTI
ichele Cutrofello, RTI
J
M
Cynthia van der Wiele
he Nature Conservancy
ature Conservancy/SARP
Guests (Online)
, T
N
Mark Anderson
Mary Davis, The
whorn
le
g
Dan McLa
Linwood Pee
Jeffrey Mannin
tt
on
Kyle Hall
David Ellio
ars Harm
lorence
L
F
Haywood
nstitute
NCSU Cooperative Extension Facilitation Team
Mary Lou Addor, Natural Resources Leadership I
(NRLI)
Patrick Beggs (WECO) Watershed Education for
Communities and Officials
Christy Perrin (WECO) Watershed Education for
Communities and Officials
Nancy Sharpless, Natural Resources Leadership Institute
(NRLI)
The purpose of the Ecological Flows Science Advisory Board:
he Ecological Flows Science Advisory Board will advise NC Department Environment and Natural
sins
T
Resources (NCDENR) on an approach to characterize the aquatic ecology of different river ba
and methods to determine the flows needed to maintain ecological integrity.
, and background information about the E‐Flows SAB are available at: Presentations, reports
www.ncwater.org/sab
at the
ddress and map ar
The EF SAB will meet April 24, 2012. 9:45 am 4:15 pm Stan Adams Education Center
e on the last page.
A
EFSAB Meeting Summary ‐ 2/19/2012 Page 2 of 37
I. Executive Summary
Below are highlights of the meeting summary sections, including actions to be taken (in bold).
III. Fidelity Testing:
The Environmental Defense Fund and Research Triangle Institute (RTI) are working
together to develop a strategy for fidelity analysis. They provided on overview of the work.
An objective of the project is to adopt a stream classification system that represents the
distribution of aquatic biota in North Carolina. Fidelity is the term used to describe the
likelihood that an organism or guild is found in one class, versus another class. The more
faithful an organism or guild is to a class, the more fidelity it shows. The project will
compare fidelities of aquatic biota to different stream classification systems, starting with
the EFS classification system. RTI will
• class – pair 185 USGS gages to biota at biological monitoring sites to determine stream
biology assignments
• generate 500 “virtual gages” with WaterFALL hydrologic data to assign stream
classes to biological monitoring stations without gages
• use Random Forest non‐parametric analyses to determine probability of species
occurrence and biological fidelity to stream classes
Potential future project steps have been identified but were not presented in detail.
A question was posed concerning a discrimination analysis or analysis of similarity to
incorporate the entire assemblage information and come up with an overall assessment of
whether these communities are following the classifications. It was suggested this should
be done before an analysis using random forest of each individual species. Board
members Sam Pearsall and Tom Cuffney will lead this exchange of data required to
erform this analyses. p
V. The Nature Conservancy Classification System for Northeast US:
Mark Anderson presented on their project to create a standardized, 13‐state aquatic habitat
classification and mapping system to provide a foundation for state and regional
conservation in Northeast and Mid‐Atlantic. Products were a standardized NE Aquatic
Habitat Classification System (ANAHCS), and a GIS dataset of aquatic habitat using the
AHSCS. Four key habitat variables used include stream size, stream gradient geology, and
stream temperature. They developed a mapped stream system with each stream type as
some combination of these 4 variables, which was simplified into 92 stream types. Mary
Davis explained that the Southeast Aquatic Resource Partnership (SARP) hired TLC to
expand the classification to the Southeast. SARP is starting with the NE classification and
adding additional variables.
Mary concluded that hydrologic classification has not been demonstrated to be a consistent
factor in improving ecological and instream flow relationships. The ELOHA process should
be revised to use classification to improve flow‐ecology relationships, but not necessarily as
a first step that constrains the development of these relationships. She suggests picking a
classification and running with it, while explaining it can improve as more information
becomes available.
EFSAB Meeting Summary ‐ 2/19/2012 Page 3 of 37
VI. Rive s: r basin hydrologic model update and visual representation of stream classification
Steve Reed, NCDWR provided a brief update of the status of river basin hydrologic model
development, and then introduced Michelle Cutrofello, RTI, to share a demonstration of
color coding to individual stream reaches to illustrate stream classifications at work. She
showed a couple examples (in Little Tennessee and Tar River basins) where using
WaterFALL she determined the streamflow class for ungaged segments using the the state’s
classification software . The stream segments were color coded to visually represent the
resulting classifications and points at which classes change along the length of a river.
VII. Update on habitat modeling scenarios:
Jim Mead shared the changes in running and illustrating habitat modeling scenarios that
he’s made since the last time he showed them to the EFSAB in October. Based on feedback
from today, he is leaning towards using 11 shallow guilds and 8 notshallow guilds
and have 2 plots at all the sites (rather than one plot of 19 guilds). He will use box
plots, and a symbol convention that discriminates by stream class. He will
contemplate how to address the proportional habitat issue that was raised. Tom
Cuffney offered to assist with this if needed, but needs to hear from Jim if that is the
case.
Table of Contents / Feb 21 agenda (click to go to section)
I. Executive Summary .................................................................................................................................................... 2
II. Welcome, Agenda Review, Introductions, Logistics ...................................................................................... 4
III.Review of February 21, 2012 Meeting Summary ........................................................................................... 4
. IV Fidelity Testing ............................................................................................................................................................. 4
V. Small Groups Discussion on Fidelity Testing ................................................................................................. 17
VI. The Nature Conservancy Classification System for NE .............................................................................. 19
VII. River basin hydrologic model update and visual representation of stream classifications ....... 24
I. VII Habitat modeling for evaluating ecological flows ‐ Update ...................................................................... 29
. IX April 24 Agenda .......................................................................................................................................................... 36
......................................................................... 37 X. Directions .............................................................................................
EFSAB Meeting Summary ‐ 2/19/2012 Page 4 of 37
II. Welcome, Agenda Review, ntroductions, Logistics
Patrick Beggs welcomed everyone to the 10th meeting of the NC Ecological Flows Science Advisory
Board. Everyone attending the meeting, in person and online, introduced themselves and their
affiliation. The EFSAB was reminded about the ground rules, the process for raising questions, and
reminded to speak into the microphone when raising questions. Everyone was reminded that the
I
session was being recorded.
The remaining 2012 meeting dates and meeting locations are listed below and posted online at
www.ncwater.org/SAB. The facilitation team is looking into new meeting locations with free
parking and ample facilitation meeting space whenever possible and thus the following meeting
locations may be subject to change:
April 24, 2012 ucational State Forest ‐ Stan Adams Training Center, Jordan Lake Ed
St enter (tentative)
June 19, 2012 ng
May 29, 2012 ‐ an Adams Training C
August 28, 20
‐ Wake County Agriculture Services Buildi
12
Sept 25, 2012 ‐ St
‐ Archdale Building
an er (tentative)
October 23, 2012 ‐ W ure Services Building
Adams Training Cent
ake County Agricult
November 27, 2012 ‐ Archdale Building
III. Review of February 21, 2012 Meeting Summary
The January 17, 2012 Meeting Summary was distributed for a final review, is approved, and is
posted on the DWR website. All approved Meeting Summaries of the EFSAB are located at:
ww.ncwater.org/sabw
IV. Fidelity Testing
. Sam Pearsall addressed the Board first
The sereem to be 3 separate concerns
1.we should have a biology based classification
2.we don't understand the classes we are using in relation to their biological charac
3. we should look at more and different variables that take us beyond flow and into
ter
substrate, climate, morphology, etc.
Of these 3 concerns, Environmental Defense Fund (EDF) has decided to look at 2 of them, #s 2
and 3. We are not looking at #1 because it could lead to circular logic and because we simply
don't have the data available to develop a set of biological classes. Questions to ask include:
How do the classes we have represent biology? How does biology respond to the classes we
have? Should the classes be subdivided or changed as a result of the biological analysis? Are
there other variables that we should incorporate into our analysis?
EDF has negotiated with RTI to develop a strategy for a stepwise approach to fidelity analysis.
Fidelity is the term that describes the likelihood that an organism or guild is found in one
class, versus another class. The more faithful that organism or guild is to a class, the more
EFSAB Meeting Summary ‐ 2/19/2012 Page 5 of 37
fidelity it shows. EDF wanted to work with RTI for a few reasons. One is the WaterFALL
program. EDF worked with RTI to come up with ideas and worked with DWR to see what
thos ide eas might look like. We looked for data sources and found three:
originally rare species, 1.Natural Heritage data which incorporates initially a
2.DWQ bug and fish data for water quality analysis, an
3. Fish data from the Wildlife Resource Commission.
nd
d
Jennifer Phelan of RTI presented the proposal for moving forward to the EF SAB.
Michelle Cutrofello and Phillip Jones were also present to help answer questions specifically
in relation to the hydrology and statistical analysis components, respectively. Sam Pearsall
also helped answer questions. The slideshow can be found on the project website.
Project Objective: To adopt a stream classification system that represents the distribution of
aquatic biota in North Carolina. Establish a connection between stream classes and aquatic
biot usa ing 3 components:
1.Evaluate the biological fidelity to the 7 stream classes developed by the EFS
2.Compare fidelities of aquatic biota to other stream classification systems
3. If necessary, modify the EFS stream classes to more accurately describe the
distribution of biota
Component 1 has 2 parts: 1 and 1a.
Steps 2 and 3 are not currently funded and will only be briefly reviewed in this presentation.
Step 1
Examine the biological fidelity of aquatic biota to the 7 EFS stream classes. To examine them
in their current form and see how well they describe the distribution. To do that we will pair
stream classes with biology and then conduct statistical analysis to find out probabilities of
species occurrence and biological fidelity to stream classes. We anticipate pairing USGS gages
(185 gages) with biological monitoring stations and the aquatic biota at these stations. These
EFSAB Meeting Summary ‐ 2/19/2012 Page 6 of 37
185 gages were selected because of their stable (have not shown changes in hydrology over
time) flows. We will assign classes to the monitoring stations. Criteria include the same
NHD+ catchment or they could be a maximum 1 km distance between a station and a USGS
o. gage, but there cannot be any sources of major instream flow alteration between the tw
Question (Q): That doesn’t seem reasonable to remove bio data if there are tributaries.
Response (R): We want to make sure the stream class with the biota is correct. If there is a
USGS gage associated with a stream class and there is a major source of flow alteration, it may
s that sound reasonable? change the functional stream class. Doe
R: I have to think about it some more.
Presentation continued
We will use 4 databases for this work:
• DWQ) Benthic macroinvertebrate ‐ NC Division of Water Quality (
• ram
• Stream fish community ‐ NC DWQ
Natural Heritage Inventory ‐ NC Natural Heritage Prog
• Trout database ‐ N.C. Wildlife Resources Commission
We expect to conduct fidelity analysis differently for each of these databases. To maximize
number of pairings between stream classes and biology the plan is to create 500 additional
s. virtual gages using WaterFALL data and pair these with biological monitoring station
ates ‐ did you use qualitative or quantitative data? Q: About benthic macroinvertebr
R: We used presence or absence.
Comment (C): It’s very important to clearly state the assumptions built into this, and the
inherent bias built into this. For example, the benthic macroinvertebrate database is 30 yr
old, very extensive, semi‐quantitative and built upon wadeable streams. The issue of the
gages is also biased; depending on funding source the gage was located in a certain place.
None of the data is random or developed for ecoflow classification. So if we understand the
bias, we can better understand the results. In reference to whether or not to use the semi‐
quantitative info, presence/absence data might provide some difficult bias to cope with.
Often during heavy runoff events, species will be transported downstream so if they are found
it doesn’t necessarily mean they are representative there. This is a good effort, just need to
clearly capture and state the assumptions and bias. I completely support it, it is the majority
of data that is out there, it is well worth exploring. Let’s be sure not to over‐interpret the
results.
ological flows. R: Yes, the data was collected for a variety of reasons, none of which was ec
Q: Question rephrase, will it be more useful to include relative abundance?
R: Best we can do today is promise to evaluate data we have, see if possible to use the relative
abundance info. We’re driven to least capable dataset when using multiple sets. But we’ll see
if we can.
Addressing the issue of not dropping biological data because of distance of gage ‐ the virtual
gages may help this.
C: We've been talking about biological bias but there are also hydrologic biases. We haven’t
talked about the representativeness of the hydrology data ‐ the gages are not located
randomly, in fact many are clumped. There needs to be discussion of biases that were
EFSAB Meeting Summary ‐ 2/19/2012 Page 7 of 37
brought into gage locations. If we don’t take them into account now then we propagate them
in the WaterFALL model. I am assuming RTI has looked at this already when developing
WaterFALL.
R: 2 responses.
1. We will generate 500 virtual gages, not based on the location of the current gages, but
according to a different set of prejudices. They’ll be classified and we’ll see out the
classification system looks after that.
2. We’ve looked closely at regional classification by McManamay. He used different
variables and arrived at similar classes. The likelihood that our classes are more than
trivially biased by gage location is small, but the 500 virtual gages will resolve that for
us.
C: Isn't there a bias in WaterFALL because it comes from gage locations?
R: WaterFALL does not use any gage information to create flows. It is a rainfall model so it
runs on landuse and soils. Gages are used to calibrate but it is done in a relative manner and
that is something RTI is looking into, to see if it introduces any bias. The calibrations are
largely focused on 44 headwater watersheds in NC with little to no human influence.
C: May be useful to do another WaterFALL presentation in the future to demonstrate how it
can be used to simulate gages.
C: I don't need to know how, just what is the variability. It is 10 or 10,000?
ties. C: All science is imperfect. As we get information we need to state the biases or difficul
C: The comments don’t mean we’re not supportive, we just want to be able to state the
inherent biases.
Presentation continued
There was a high level of correspondence between NC classes (7) and McManamay classes
(8). Criteria for locating the 500 virtual gage stations:
stations distributed evenly across the state. Monitoring
Eliminate:
• catchments with impaired water quality (as determined by NC Division of Water
Quality – 303d listings)
• scharges and/or catchments with major in‐stream flow alterations (impoundments, di
intake points)
catchments with “poor” or “questionable” biological monitoring data •
Select:
• catchments that contain biological monitoring stations from multiple aquatic biota
datasets
• tions biological monitoring stations sampled during years with average climate condi
biological monitoring stations with most recent biological data
• biological monitoring stations with multiple biological measurement dates and
presence/absence that doesn’t change by > 10%
•
EFSAB Meeting Summary ‐ 2/19/2012 Page 8 of 37
• biological monitoring stations upstream from USGS reference gage
Q: Does this get you around the issue that DWQ data is for wadeable streams? Are we limited
to wadeable streams for output?
R: Yes, for initial output we are limited by the data available.
C: The vast majority of NHP data sites are not monitoring stations but from an inventory
approach where somebody hit a site once in 25 years, with no repeat sampling. I’m
encouraged we have DWQ data of fish and macroinvertebrates and it will fit nicely with what
you are describing and the somewhat standardized methods over time and relative
abundance data. The NHP data and NCWRC data won’t necessarily fit this description very
well. Can you comment on this?
R: We can try to address these concerns as a first filter by trying to get catchments with
multiple monitoring stations so we can include both NHP data and the trout data. We
appreciate that is a concern.
Q: Jim: I know DWQ benthos data is from wadeable streams, are the other limited that way?
R: Yes to fish data, mostly to NHP data, and trout data. Coastal plain streams and swamps are
particular problems.
C: Even with 500 virtual gages and best intentions in world to use all data we can find, as well
as possible, in the end our analyses will be inadequate in the coastal plain. If any of you are in
a position to launch significant bio monitoring activities in the coastal plain that would be
great. We’re facing significant challenges there. It’s not that the water issues are less
. significant, we’re just not going to have the data to do a great fidelity analysis in coastal plain
C: Fish community database was definitely low in coastal plain, but still 15‐20% of 180 fish
community sites were in coastal plain.
C: Data has been collected in coastal plain, but I don’t know that they are in data set formats
that are quickly available for use, beside the trout data. May not even be done in a year so
won’t be helpful. Questions may be asked of other specific entities to fill the gap.
C: I agree with coastal plain remarks. DWQ has not been able to do a significant amount
there. We don’t have sufficient methods to analyze the data to make evaluations for water
quality. We’ve revised our methods /criteria for other systems like swamp systems. It may
be 10 years before we have sufficient data for good coastal plain analyses. We’ll be able to get
there sometime in future.
C: We’re taking the best available data and doing what we hope is useful;results will
determine if it is.
C: There may be NCSU shad work and anadromous fish studies that could be plugged in but
the data isn't currently in an appropriate database.
Q: RTI wants to use biological information from years with average climate conditions and
therefore average flow conditions, yet the interest of the larger analyses means it is important
to know the extreme conditions and how the biology responds to extreme flows. How do you
reconcile that?
R: Good point. It is important to consider the variety of flow when you are looking at
flow/biology relationships, but for this purpose we want to try to avoid extreme or dramatic
events where the species may not be found during those times at those locations. The purpose
EFSAB Meeting Summary ‐ 2/19/2012 Page 9 of 37
behind this analysis is to determine what biota is typically found in this stream class, so we
don’t want to include years where extreme events cause them not to be there.
extreme years, then that may be critical for the analysis. R: But, if they are there in only
R: Hadn’t thought about that.
C: The classification system includes classes that have extreme conditions.
R: That is a factor that should be included in determining which stations to include.
Q: When a rerun of the classification is conducted, do you include the virtual 500 gages? Does
that give us a deeper or better grasp of what the classification represents.
R: Maybe. I imagine that after classifying the 500 virtual and 185 existing gages, if we found
classes consisted of tight clusters with small diameters then we would not rewrite the
classifications, but if we end up with soup, then we would rerun the classification and maybe
have a new set of classes. I'm not sure if this is a good use of money.
C: Comment about extreme flow issue: There are variables in the software that use very high
and low flows to determine if it is in a specific class. That is over 80 years of record. For
biologic data, in terms of filtering that, is eliminating a biological monitoring data point if it
happened to be collected on that year when there was a significant event, but those other data
points that are left in, will still have a history of high and low flows where streams still subject
of the extreme events.
R: But dramatic events may lead to a species presence because of the event.
C: Trying to synthesize some of what has been said. If we incorporate extreme flows into
modeling and different types of stream classes, if we leave out the bio data in those extreme
points, are we only talking about leaving out a small data set, are we significantly reducing
biological data because of the decision of what is an extreme flow event ?
R: If a class supports organism X most of the time, but during extreme flow events, organism
X is washed downstream or killed by drought, in both cases what we don’t want to do is
determine that this class doesn’t support organism X, since the vast majority of time it does.
So class that includes extreme conditions as part of determining variables is one thing, but we
want to sample what is in the class normally, not in transient event.
R: Then we need to include a lag function. I think you’ll find that the amount of
discriminatory ability in the class system won’t be significantly changed by many degrees by
ust. including the periods of extreme flows. If you don’t, you need to adj
R: We can look into that, to see if extreme years change the results.
C: You can do both, look at narrow more typical range, then look at all the data, run a delta
analysis to see difference.
Presentation continued:
Third component of the step 1 methods: The statistical analysis we will use to determine the
biological fidelity. We are proposing the Random Forest non‐parametric analyses to
determine probability of species occurrence and biological fidelity to stream classes. We
would run the analysis on the 185 gages alone, the 500 virtual gages alone, compare the
results, and then perhaps combine them.
EFSAB Meeting Summary ‐ 2/19/2012 Page 10 of 37
Using this yes/no model we can develop a table of fidelity.
This shows that species 2 and 9 have high fidelity, but species 5 for example does not show
fidelity and maybe doesn't have high specificity for a specific niche. But a table like this would
be indicative that there is no fidelity to the classification system.
EFSAB Meeting Summary ‐ 2/19/2012 Page 11 of 37
Q: Why use a random forest analysis on individual species? You could do discriminatory
analysis, or analysis of similarities, and incorporate entire communities into this, and come up
quickly with an answer showing whether you are getting discrimination. Here you don't
know whether you are getting discrimination, you have no tool to tell you. With these other
methods you could actually get a probably that your classes are actually being classified
correctly.
R: My understanding is that this will produce probabilities of occurrence.
C: But that isn't what you want to know, you want to know if the classification system is
working correctly, so you need to know what percentage of your sites are being misclassified
‐ this isn’t going to get you that information. I think the first step to do is a discrimination
analysis or analysis of similarity incorporating the entire assemblage information and come
up with an overall assessment of whether these communities are following the classifications,
before you jump into a very lengthy and detailed analysis using random forest of each
individual species. There’s an overall question to ask upfront.
R: This may sound complicated but it is easy to run and very efficient. Even with thousands
of rows and columns it can run in a few minutes. It is also easy to set up and to change. In
terms of accuracy measurements, the results can be used to correctly classify, or for
specificity or sensitivity, or kappa statistics. Also, it is not parametric, so there are no
distributional assumptions about the sites. To a certain extent, randomized sampling is less
of an issue.
Q: Looking at the SLIDE 12 table showing no fidelity, would your conclusion be that there is
no difference between the communities in each of those streams based on that diagram?
R: Yes, the streams don’t capture distinct distributions of the biology across the spatial
landscape.
C: I think there is a difference at the hierarchical level of species about whether you find them
across streams and communities. I could argue stream D has a different community than
stream A based on that diagram. Are we interested in stream communities, species, both,
neither, and is that analysis going to get you that community level.
R: Looking at SLIDE 4 ‐ The strategy here uses a working hypothesis‐ we will show low levels
of fidelity after matching up existent biological data with existing classes. At step 3, we’ll
subset existing classes using other variables, topoadaphic variables (those associated with
substrate and shape of land, like slope and velocity and also maybe climate) and ask
ourselves, do the classes subset into new classes to which we do show fidelity. My working
hypotheses is that they do and we will be able to subset these classes using non‐flow variables
so we begin to see significant fidelity. The reason we’re working on a species approach to
fidelity is that’s the only one I know of. If we attempt a community based approach, we first
have to come up with community classes for streams statewide.
C: We already have the classes, you’re trying to test if the communities will break out the way
the classes do. So if you run an analysis of similarity which is a multivariate way that uses all
the species information simultaneously, and you have certain classes defined and it will tell
you whether those classes are divided up biologically the same way or not, it will even give
you a significance level.
R: I follow your intent but I don’t know your method well enough. Will you send me and Jim
a 1‐page description of statistical strategy you’d like us to consider? And we will incorporate
it if possible.
EFSAB Meeting Summary ‐ 2/19/2012 Page 12 of 37
C: If you send me the data I can do the analysis of similarity for you. I’d start at that level
before doing individual species because there is a possibility of compounding errors
whenever we do the same thing over and over. We, as a group need to consider: Do we need
to consider the assemblages or do we consider each individual species, or do we consider the
e things. guilds that are being used to define the habitat parameters or do we do all 3 of thes
R: The guilds are pretty much defined by some combination of flow and substrate.
C: I mean in terms of analysis. There are 3 things going on ‐ the individual species, the
assemblages, and the metrics, which is where we pull all the information and combine it
together. Those are 3 different endpoints. Which one are we going to consider?
R: All we are doing at this point is attempting to ask the question, are the classes real, or is
there a way to make them real using universally available data, such as soils, topography, and
climate. And we define reality by whether or not they show consistency based on their
biological composition.
ysis. 1. I suggest RTI send the biological data to Tom who can do the discriminate anal
2. We continue with fidelity work and in the end we will hopefully have a set of
classes we think are real and have consistent biological characteristics so we can have
ecological flows baselines to comply with the law
C: My concern is what if we do all this and we wind up changing everything and then later we
find that if we actually run the guilds, the classes would be fine? We should analyze the
ore changing anything. species, the assemblages and guilds bef
’t we analyzing the guilds now? Q: Aren
R: Yes.
out? Q: Which guilds are we talking ab
R: For the fish, the fast, slow, etc.
R: That ain’t part of this analysis.
C: My point is if we use the guilds as the end point and we find out the guilds actually end up
matching our current classification system but the species didn’t, then we have spent a lot of
money revamping everything when we already have a guild endpoint that matches
everything. We should analyze species, classification and guilds before we spend too much
time.
pproach? Q: How would we run guilds through this a
C: The guilds would essentially be species.
C: But some of those guilds, those curves that were used in habitat were as simple as
shallow/fast, so I’m not sure how you use that as a biological datapoint in this analysis.
C: I think it could be a percentage of the total community, where a certain percentage might
be more or less.
time. C: But that would mean we’d have to assign species to guilds which would take a long
C: The point is good, I understand where you’re going I’m just trying to figure out the
mechanics of it.
C: We don't have the fish data available to really get these answers we need for the fish guild
approach.
EFSAB Meeting Summary ‐ 2/19/2012 Page 13 of 37
C: I agree it would be important to do both a species level analysis and a community level
analyses and I think those could be done relatively easily. Also, parallel to the fidelity
analysis, an indicator species analysis should be done to see if they are specific indicator
species to see if there are specific indicator species that are representative of the stream
classes. This would be along the lines of a fidelity analysis but we might have a couple key
species that jump out as better indicators than others. Those 3 things could pretty easily be
done together with same data set.
ation continued: Present
Step 1a
Objective: To test biological fidelity to other stream classification systems and compare with
EFS
To look at a variety of classifications systems that have been developed for the southeast and
to look at the biological fidelity for these systems. We looked at (references available on the
website):
• ams McManamay et al. (2011) – regional classification of unregulated stre
• Konrad (in review) – hydrological classification in southeastern U.S.
Both systems are developed for southeast US and based on minimally altered/unaltered
streams and USGS gages and also based upon hydrologic methods. Both produce 8 distinct
stream classes, where as EFS has 7. The analyses will see how similar they are with reference
to classification of streams and compare the fidelities of this different stream classes, by
repeating the random forest or other statistical method that we decide to use. We would
repeat that analysis for each for the stream classes and then compare them. Perhaps one of
the classification systems works better than others. We anticipate Step 1 will give us a good
idea of which data sets are the best to use in this fidelity analyses, so maybe for step 1a we
may restrict the analysis to only the biofidelity bases which were useful.
yet. The following is a brief review of Steps 2 and 3 which are not currently slated to happen
Step 2: Assess the biological fidelity of aquatic biota to stream classes that only include
streams that are not altered (i.e., minimal instream flow alterations). We are using natural
streams in order to eliminate any other stressors that might be influencing the biology. The
EFS system based its classification on 18 years of 185 gages with stable flows, but that doesn't
mean they are unaltered streams. For example something outside of stable flows could have
occurred over 18 years ago. To do this we would:
• re‐classify streams (at 185 gage locations) using WaterFALL hydrologic data
(unaltered condition) and EFS software
• streams that change classes with the reclassification are considered “altered”
eliminate “altered” streams from dataset
• repeat Random Forest non‐parametric (or other statistical analysis we have decided
upon) to determine if biological fidelity to stream classes is improved with dataset
•
restricted to non‐altered streams
Depending on what we find, we may proceed to Step 3 which would further refine the EFS
classification system.
EFSAB Meeting Summary ‐ 2/19/2012 Page 14 of 37
Step 3: Evaluate the ability to improve biological fidelity to stream classes by sub‐dividing
and/or aggregating the 7 EFS stream classes.
• Repeat Random Forest non‐parametric statistical analyses to determine aquatic biota
associations to stream classes and include additional parameters such as
physiographic/eco regions, which may help us take into account spatial separation
that might also influence biota.
The current EFS system doesn’t use geographical data in classifications, so in some cases
there’s a stream class that only occurs in the lower coastal plain also occurring in the
mountains. So, we test to see if the biotic associations could be divided or have ecoregions
that better represents the distribution of biota.
We would use the following 5 physiographic / ecoregions. Larger maps with legends are
available in the presentation on the website.
Classification System Reference
1 Ecoregions of the
Conterminous United
States
Omernik (1987)
2 Bailey’s Ecoregions and
Subregions of the United
States
http://www.nationalatlas.gov/mld/ecoregp.html,
http://na.fs.fed.us/sustainability/ecomap/section_descriptions.pdf
3 Physiographic Regions
of the Conterminous
United States
Fenneman and Johnson (1964)
4 TNC Ecological Drainage
Units
http://www.2c1forest.org/atlas/metadata/edu_metadata.htm
5 Hydrologic Landscapes Wolock (2003)
1. Omernick ecoregions Level 3:
Ecoregions denote areas of general similarity in ecosystems and in the type, quality, and
quantity of environmental resources. The approach used to compile this map is based on the
premise that ecological regions can be identified through the analysis of patterns of biotic and
abiotic phenomena, including geology, physiography, vegetation, climate, soils, land use,
wildlife, and hydrology. The continental United States contains 104 regions. Four ecoregions
in NC.
EFSAB Meeting Summary ‐ 2/19/2012 Page 15 of 37
2. Bailey:
This map layer is commonly called Bailey's ecoregions and is based on four levels of detail
(hierarchy) that include the influences of climate (precip and temp), vegetation (natural land
cover),terrain features, and elevation. Five ecoregions in NC.
3. Feneman:
This is a polygon coverage of Physiographic Divisions in the conterminous United States. It
was automated from Fenneman's 1:7,000,000‐scale map, and is based on topography, rock
types and structure, and geologic and geomorphic history. Four ecoregions in NC.
4. TNC Ecological Drainage Units:
Ecological Drainage Units (EDUs) group watersheds that share a common zoogeographic
history, physiographic and climatic characteristics, and therefore likely have a distinct set of
EFSAB Meeting Summary ‐ 2/19/2012 Page 16 of 37
freshwater assemblages and habitats. Twelve ecoregions in NC
5. Wolock:
Hydrologic landscape regions based on similarities in land‐surface form, geologic texture, and
climate characteristics. Twelve ecoregions in NC.
Other parameters that might be included in this analysis is the actual flow metric that might
determine stream classes. There are 22 flow metrics that were used by EFS. Conceivably
there might be clusters of biology occurring within the same stream class that have greater
affinity for different flow metrics. These “clusters” of biota may indicate the ability to divide
stream classes, and biota occurring in multiple stream classes may offer opportunity to
combine classes
Questions for RTI:
C: A couple of points: 1) as you look at including relatively unaltered streams, please look at
DWQ databases on point sources where there are high concentrations of waste on 7Q10
conditions. As well as the DWQ bentho database, there is also a habitat database. Eric Fleek
would be the contact for that. 2)DWQ has developed and funded Omernik level 4. The
current Omernik level 3 is probably not worthwhile.
C: Much of this data is collected on a 5‐ year cycle, so for example we don’t necessarily have
data from Neuse and French Broad in same year.
R: Yes, but most of the sites that will make it under consideration have been evaluated
multiple times over a long period of time, certainly not annually, but say, over a 30 year time
frame. I’d guess that with additional analysis we’ll see a large number of animals repeated
over time. You’re right it’s a basin‐wide rotational approach due to limited resources. Most
EFSAB Meeting Summary ‐ 2/19/2012 Page 17 of 37
but not all sites are repeated over the long term.
C: There is a habitat database describing the habitat for where the biological monitoring sites
are located. It is semi‐quantitative. Erik Fleek is the contact for that data.
V. Small Groups Discussion on Fidelity Testing
The EFSAB and alternates divided up into 5 groups to have a discussion and report out
nswers to the following question: What additional questions and/or concerns do you have
bout the methodology used in the fidelity testing?
a
a
Gro oupne:
1. Is th an: ere such a thing as
• unaltered stream?
land use patterns •
• irrigation withdrawals
2. Con ta cern: for two/three of the classes, we do not have biological da
• use data from other sources if available and easily assessable
3.How do you account for sampling biases? (for example during a low flow year?)
4.Can we look at Konrad’s classification? Is his paper ready for review?
5. What is the timeline for this process? (for Fidelity or the EFSAB‐ if latter, reference
Tom Reeder’s presentation)
Gro 2up
ne 1.If classification relies on output from a particular tool (e.g., WaterFall) to determi
class, will DWR have access to that tool?
r
2.Large river data lacking‐ may be available from FERC relicensing projects but fo
altered streams (Group 4 also posed the same solution for larger river systems)
3. Other (non‐hydrologic) variables for classification need to be widely and easily
applicable and available. For example, gradient could be important.
Gro 3up
1. Avoid circular thinking: where do we stop – biology, hydrology, biology, etc.
sses on biology, or then hydrologic ‐ what is the stopping criteria • assemblage cla
2. Need definitions:
a. unaltered flow – how many water sources in areas with unaltered flow
at constitutes an b.altered flow vs. unaltered (hydro) – buffer for defining wh
erence sites for bio
alteration.
c. reference sites for hydro vs. ref
distance d. tributary effects with
3. Criteria for site selection
EFSAB Meeting Summary ‐ 2/19/2012 Page 18 of 37
a. changes in biota over time
b. changes in conditions over time
ding conditions of biology; need info on lag
o site selection.
4.A prior flow condition, effects of prece
rate this int
effects and how to incorpo
5.Time frame of analysis and reporting
6.Exploratory data analysis
7.Reference sites – definitions for hydrology and biology (fish and bugs)
8. Need to add additional parallel analysis that includes altered sites (hydrology) is/in
existing hydro classification (e.g. Neuse above and below dam)
Gro 4up
1.Define community based assemblage vs. individual species response
2.Concern that smaller wadeable stream communities may show more
sensitivity/difference relationships than larger rivers (*size factor)
3. For larger rivers not wadeable – will we have enough bio data to show fidelity?
(*hydropower/utilities companies may be willing to share data from larger rivers –
tap into FERC data)
4.Native vs. non‐native species fish (whether to include non‐natives in the analysis?)
5.Importance of physiography: hydrology – typology. If no fidelity does it mean it is
incorrect or that they are not related?
iew from 6.Support approach of virtual gage data. It would help to have some input rev
hydrologists
7. How does WaterFall address groundwater inflow and lag behind rainfall?
G
roup 5
1. Presence or absence of data species may not reflect communities and guilds
appropriately. Density information would help.
2.Is what we are doing defensible regarding ecological integrity? This will be critical
when legal issues come into play.
3. What are the effects of species abundancy refraction (?) curves on amount of sampling
used for plus/minus information? (change in # of species with # of samples with
4.
different amounts of samples in different streams)
Potential data sets beyond the five introduced earlier in the Fidelity presentation?
mic data sets may be valuable – eg‐ long‐term data from 1 or 2 a. academic? Acade
streams to get ideas of how things may work across other streams
b.other agencies?
c. value of long‐term or intense for 1 or few systems to evaluate uncertainty of larger
data set of data used.
5. What is the end game? If all works with this, how does it affect what Jim is doing? How
does it contribute to what Jim is doing?
EFSAB Meeting Summary ‐ 2/19/2012 Page 19 of 37
VI. The Nature Conservancy Classification System for NE
Webinar Presentation on the Northeast Aquatic Habitat Classification and Mapping
by Mark Anderson, Mary Davis
Cat Burns introduced Mark Anderson with The Nature Conservancy (TNC). Mark is the
director of conservation science for the NC Eastern Region and got the NE Classification
project going. The approach is different from many other approaches we’ve talked about.
Mary Davis is also online so she can explain how the work from the NE is being expanded to
the southeast, how the project is evolving, timeframe, and what may be available to the
EFSAB.
Mark explained that his colleague Arlene Olivero is not on the call but did a lot of the GIS work
on the project. You can follow up with either her or Mark for questions. Project was funded
by NE Association of Fish and Wildlife Agencies‐ 13 state FW agencies pooled their money.
Project Objective: create a standardized, 13‐state aquatic habitat classification and mapping
system to provide a foundation for state and regional conservation in Northeast and Mid‐
Atlantic. Products were: NE Aquatic Habitat Classification System (NAHCS), a standardized
aquatic classification system; GIS dataset of aquatic habitat using the AHSCS.
Process (slide3) ‐ We put together a large steering committee, a couple federal partners and
the 13 states (30 people). Began by compiling existing aquatic classification systems
developed by each state, opened them up and looked at them, held monthly topic‐focused
worksgroup calls to review different variables, discuss whether we should use them, think
about thresholds, and reach consensus.
After looking at the state classifications, compared them, and had a long discussion. We found
that (Slides 5‐6) the 13 states in NE and mid‐Atlantic, there was little consistency with how
states classified aquatic systems. Depending on the state they used different variables such as
size, temperature, gradient, chemistry, indicator fish communities, etc. We couldn’t find a lot
of consistency across the states, they classified their lakes differently also. After discussing,
the team agreed strongly they wanted an unambiguous biophysically based standard
taxonomy that we would develop for this project and that would work across the states. We
decided right at the start we wanted a geophysical classification.
Classification approach (slide 7) involved taking a stream system like the network seen here,
and look at a variety of variables such as size, hydrology chemistry, looked at it stream reach
by stream reach and come up with a set of variables to characterize the stream. It was an
important discussion that we were clear the product was not intended to override state
classification, it was intended as having a way to look at stream patterns across the states to
link them together so they would be unambiguous. It’s important in the NE and mid‐Atlantic
where states were various sizes.
Starting with that group, first thing we did was brainstorm out a lot of different variables that
have been used to classify various streams (Initial Classification variables slide 8). We argued
and discussed which variables were important, and agreed after a month or two upon 4
variables that we thought were important in this region.
(slide 9)Results: Four Key habitat variables: Size, gradient geology, temperature
The next part of the project, a lot of discussion went into each of these 4 variables and figuring
out how to parse it up to reflect biodiversity. I’ll show one example of how we got there,
which might give you ideas for your process. (Slide 10). One result ‐ we divided one system
EFSAB Meeting Summary ‐ 2/19/2012 Page 20 of 37
into 7 size classes. Size classes were based on square miles of upstream drainage area. We
have headwaters and creeks (we called streams), then for rivers, which included small river,
medium tributaries, to great. Getting to these 7 classes we had a lot of tests, work, and
discussion.
Pennsylvania tested different stream class combinations. We started with 5 classes, then what
you see here is how well the 5 classes differentiate the fish fauna in PA in the Atlantic Basin
and the Ohio Basin. The 5 classes did a pretty good job of differentiating the fish fauna in
those 2 basins. We tried some other class combinations, some worked well in Ohio‐Great
Lakes (Slide 11). We ended up using the 5 classes that worked well in both basins, plus a few
more larger classes that worked well in the Ohio Basin. This was an example of how we
worked our way through size classes.
Second one was gradient (slide 12), was feeling that rise/run for several miles was a key
variable. Here we classified gradient into 6 classes. Very low and low‐ coastal plain, high and
very high are very steep gradient stream systems in the mountains. We sorted through how
many classes we needed, using known rare fish, mussels, a little bit of aquatic insects, then did
cluster analysis based on biota to see if we got clusters of biota in the different gradient
stream classes and how different they really were. The lower classes were more similar, until
you got into the high gradients.
Geology was the 3rd gradient that we talked about a lot (slide 12). What we were after was
different stream pHs with different buffering capacities, we tested if we could approximate
that based on bedrock geology. Had a couple thousand stream pH samples. (Slide 15) This
slide shows average stream pH by underlying geology type. Characterized streams on acidic
(on left) with more neutral, and highly calcarious limestone (on right). Had interesting
biodiversity that coordinated with these 3 pH classes.
Final class was temperature. This is the piece of model we feel least great about. It’s a
straight forward model ‐ 4 temp classes, cold transitional cool, transitional warm, and warm.
We joked about it because most of the states have what they call “warm” and cold streams but
when you compare a warm stream in Maine with a cold stream in NJ they are similar. It took
us a while to get it sorted out right. What you see is a temperature model based on an air
temp models divided across different stream sizes. In lower temperature class in largest river
you call it transitional cool, then in smaller river you call it cold. In different sizes
(headwaters) we brought in slightly different model with base flow.
So in the end we have we have a mapped stream system with each stream types as some
combination of these 4 variables ‐ 259 stream types occur (slide 18). They range from an
extreme of very high gradient, acidic, cold headwater stream; to the other end, very low
gradient, calcareous, warm great river. Not all combinations occur but here’s what we’ve been
doing with this. Now it got fun. You can look at a map of all high gradient, acidic, coldwater
creeks (slide 19). You can look across states and see the same geophysical setting across all
the states. Here is a picture of highly buffered, calcareous, headwater creeks (slide 20).
It’s a stream GIS network that has the classifications built into it, but it also contains another
100+ habitat descriptors that we compiled in the process (Slide 21). If you don’t like the
classification system or want to use something slightly different you can use the raw
variables. Here is a query somebody did to see which streams are on the edge of transitional
cool vs transitional warm, they were interested in climate change. Here are the places that
flip over easily (slide 22).
EFSAB Meeting Summary ‐ 2/19/2012 Page 21 of 37
Also included are a bunch of large scale descriptors like what freshwater ecoregion the
stream is in, etc.
Lastly (slide 24), we discovered that 256 stream types was more than we could fathom so we
simplified a system using the key 92 types. We developed “trumping rules” using size as most
important , then gradient, geology, temperature. It’s a simpler structure. You can use either,
they are both in the data set. I won’t go through how.
The dataset is now downloadable on the NE Association of Fish and Wildlife website. It is
getting used in many states, and we’ve just finished a measures report on state of streams.
We used this habitat system to look at streams, quality and quantity across the region. Now
we’re focusing on linking biota directly to classification. That’s tricky, because different biota
link to same stream system depending on where you are in the region‐ basin, history of basin,
species pool in that basin. We’re just beginning to work that out.
Mary will talk about how we’re expanding, customizing it to the SE states.
Mary Davis’ presentation: We looked at the NE classification when we were working with
Southeast Aquatic Resource Partnership (SARP), on our instream flow project as we were
addressing need to classify systems for a framework for same purpose Mark outlined, to
create a classification that enables communication across broader political boundaries, states,
regions.
We took advantage of what NE had done‐ felt it was thorough, they had an interesting process
to develop the class system. We in the South Atlantic region (pink area on map‐ slide 23)
extending a little further to eastern AL and FL, in past few months. SARP put Nature
Conservancy under contract to extend the classification system south. Mark and Arlene have
been working with us, with the instream flow work funding we’ve put a regional committee of
experts together, we’ve been having monthly calls. We’ve talked about size, gradient, stream
temperature. We’re adding to what they did by putting a hydrologic classification on top of
the geologic and biological classes they developed.
Where we are: We’ve gone through the different topics and have settled on a framework for
size and stream gradient, available at website. We’re still considering them draft until sent to
states for further review. We have experts on these classifications we’re talking to monthly.
We want technically a comfort level extending this to SE. Size class ‐ are we comfortable using
basin area for determining size class versus mean annual flow. Both had good arguments, so
in SE we have size classes based on both. Gradient ‐ we’re using the same as NE.
NC has agreed to do some modeling for us to determine pH and stream temperature. This is
turning out to be a little problematic, not translating as cleanly. pH ‐ we have very acidic
systems dominated by organic matter, black water streams, we’re trying to pick them up. NC
is developing a data set with different soil attributes including organic soils to model pH
accurately.
Temperature ‐ NE system ended with warm streams (warm in southern VA). In Florida that
warm category isn’t going to apply well to SE. We’re working on fish, temperature data in
similar way as NE did that will extend to classes in NE down.
We’ve got the geology worked out and available for your area. That leaves last big cahona‐
the hydrologic classification. We’re not trying to come up with the “right” classification
system, not trying to replace the state’s system. VA and NC were the only states that had
hydrologic classification. Jim Henrickson’s work, and a VA Tech student used different
approach but same variables. Ryan McManamay did a classification system ‐ NC and VA is
based on this.
EFSAB Meeting Summary ‐ 2/19/2012 Page 22 of 37
A third hydrologic classification system we were going to look at is Chris Konrad’s‐ it’s not
ready for prime time. We’ll use McManamay’s and Henricksons regional classifications, we’ll
give those to Arlene. She’ll use gaged locations that have been assigned hydrological
classifications. She’ll try to build a model based on the parameters that Mark just went
through ‐ size, gradient, pH, temperature, geology, soil depth, some other things important in
determining a stream’s hydrological regime. She’ll be using that to try to extrapolate to
ungaged river segments. The hydrological classes will be developed for 2 different systems.
We hope to have McManamay’s classification done by mid‐March. Arlene is going on
maternity leave so we should have that soon for NC.
We also have completed compiling ecological aquatic data for SE. We have the tools to begin
testing the classifications more closely, particularly the biological fidelity testing, we’re
watching your project closely for lessons learned. We may incorporate what you’ve done
aybe add on and improve. m
Discussion/Questions/Comments for speaker
Question (Q): Mary, the work you’re doing, is for the South Atlantic LCC? (Yes) Is any of that
portion going to be covering the western part of NC, which is in the Appalachian LCC?
Response (R): Yes the South Atlantic LCC goes to boundary of piedmont and mountains. In
our classification system we’ve been working more along state lines. We will have the info for
entire state of NC.
Facilitator: Does anyone have any reactions or implications?
Q: What is situation with getting biological details into classification of SE? (biota)
R: Fish community sampling data is getting compiled, sampling locations are being linked to
NHD data set, so we can merge that information with the classification, then it is a statistical
exercise to find models that I.D. fish assemblages and whether any of the classification
attributes inform where fish assemblages are, or on other hand do fish assemblages help us
find classes of the stream. NC has given their data, but I’d like to get more ecological types,
macro‐invertebrate data to strengthen that, would like to get the sensitive species data (NHP
may have some data). I thought it interesting they were using rare species data to test
attributes in NE.
This is an unfunded mandate‐ I have the data and a statistician. I want to see what RTI comes
out with, then maybe we can do more extensive analysis based on their approach.
Q: Are any states using these 92 types or ones in their boundaries, to define instream flow
requirement for ecological integrity? To define instream flow baselines?
R: Some of states are trying to define instream flow‐ MA and MD. I don’t know their success
in that.
R: PA is using size class and gradient, not successful in hydrology classification. VA is using
size and eco‐region‐ they are having some success in testing sign of flow‐ecology
relationships. When they split them into coastal plain and Piedmont the relationships
improve. They do have McManamay’s hydrologic classes but I haven’t seen them used in the
Potomac River project.
The EFSAB took a break, then came back and continued questions and answers with Mary and
Mark.
EFSAB Meeting Summary ‐ 2/19/2012 Page 23 of 37
Q: 92 classes is a lot, there is no way we’ll make recommendations specific to 92; how do you
recommend that we refine that to a manageable approach if wanted to use a similar approach
so it’s relevant to defining instream flow requirements?
R: What states have done is look at their own state and determine how many of the
subdivisions of the 4 variables are relevant to flow (4 size classes and 2 for gradient classes
for example then collapse from there). The 92 is for all the whole region, there are about 30
in a state.
Some states use the classification attributes that will be altered by change in flow. Geology for
example won’t change if you take water out of the stream. MI is a good example. They had 12
types of streams, 3 temps and 4 sizes. They knew if they took water out of a small cool
transitional stream, it would likely turn it to a warm stream if that water slowed down and
had time to warm up.
You can classify as broadly or not as like‐ some are not classifying at all. If you proceed with
classification, consider in NC what is likely to change if you alter attributes.
Q: I find the classification for the NE satisfying in that you can communicate to a wide
audience about what a particular stream is. A particular community could understand what
the system their withdrawal would be part of, and implications of withdrawal on the system.
Hydrologic class would just be one more factor that would be considered along with size,
geology for example. It would be another factor in writing a prescription for environmental
flows.
R: Also important to consider what kinds of streams are being impacted. Where are your
withdrawals happening? Headwaters? Piedmont may be more stable streams so you could
eliminate many.
Comment (C): It doesn’t matter how many variables you have if they don’t change when the
flow changes, then can’t use to address what happens with alteration with flow. If you have
92 classes and your assignment is to determine the necessary seasonal pattern of flows
(including variance) to maintain ecological integrity, what strategies can you use? In NC,
we’re trying to develop hydrologic classes then overlay with other factors like size, gradient,
temperature. Sounds like what you may be doing in SE is coming up with a set of classes
where hydrology is a 5th or 6th variable that you subset into classes. Seems that the problem
is we need to come up with classes that respond with changes in flow.
C: We need to be careful whether setting up classes by variables that respond to changes in
flow. That works if you are seeing if this change in flow results in a class change. That’s only
one way to try to determine eco‐flow guideline. For example gradient is not going to change
much with change in flow, whereas gradient may be a big effect in how a stream changes in
change in flow. Eliminating classes that don’t change much in response to change in flow
could have pitfalls. In looking at the 4 factors used in the NE: Size (drainage area) is highly
correlated to hydrology. You may not need a separate drainage area attribute if you’ve
already classified with hydrology statistics. Geology in NC for example‐ don’t know that it’s
that important. Gradient and temperature could be important for single state classification
here.
C: I’m not arguing otherwise. We care whether biology reacts to changes in flow. A number
of these variables (like temperature, gradient, mean annual flow which is likely a good
surrogate for stream size) are good variables that need to be imbedded in classification, but
I’m worried about being able to tie variations to flow to biological responses. If you can get
there by tying biology to the flow, then have a pathway.
EFSAB Meeting Summary ‐ 2/19/2012 Page 24 of 37
Q: In NC, salinity is important because we have salinity so far inland, and will be affected by
changes in flow. Did Mark consider salinity? Was there a salinity threshold/boundary in your
study?
R: We did not really. We’ve been trying to map for years the threshold for brackish, fresh,
and higher salinity. Very difficult because so many variations in coast and shoreline patterns.
It’s important, but did not include—too difficult.
Q: In my opinion it is hard for me to get from hydrology to biology. I don’t think we are going
to see high fidelity between info we have and hydrology. I do think it will relate to the
variables included in this classification (size, pH, gradient). If that is true, do we need to
tighten the fidelity of natural habitat classification and hydrologic classification? A 2 step
process?
C: To me it’s what you do first. In NE they did the landscape classification first, then
hydrology. Here we have the hydrologic done first and are trying to add the other stuff. It’s
going to be some of both. Coming at it from both ends of the spectrum.
R: This discussion reminds me of another discussion about whether define size by drainage
area or mean annual flow. They’re highly correlated so get you the same place. If climate
continues to change and you base it on mean annual flow, on the ground the streams change
categories. If you base it on drainage area, as climate changes and mean annual flow changes,
the stream is classified the same way at same place, but your descriptors change for how that
stream system works. Do the classes change or do the descriptors change?
Mary’s last comment to group: Classification for ecological flows in a number of case studies
around world, classification is not showing up as an important factor in determining the flow
requirements of a stream, in expecting the classification to improve the ecological and
instream flow relationships. This can be adaptive. Pick a classification and run with that. Put
language in recommendation that you are going on best info available and recommendations
can be improved as more info becomes available. There is a lot of work going into the
biological fidelity testing of classifications, if you can come back to it in the future, we will
ave a lot of information to offer you. h
VII. River basin hydrologic model update and visual
representation of stream classifications
Steve Reed provided a brief update of the status of river basin hydrologic model development,
and introduced Michele Cutrofello of RTI, to share a demonstration of color coding to
illustrate stream classifications at work. Steve’s 4 slides are posted on the EFSAB website.
EFSAB Meeting Summary ‐ 2/19/2012 Page 25 of 37
Steve’s presentation: The
legislation tasks DWR with
developing river basin
hydrologic models for each of
sins. the 17 major river ba
Status of River basin
hydrologic model
development: Cape Fear and
Neuse are being updated to
current info and can be run
together. Broad is light green
in west. Kerr are the 2
models that have been
completed. That’s only part
of the story. The product is a river basin water resources plan, really. The model is the tool
that helps us make evaluations in time and space across a river basin. These eco‐flows are
being used for planning, it is going into a plan. When we determine places in the future where
there will be an impact on ecological integrity or a water shortage, our plan will lay out
different alternatives for people to look at. That’s what your work is building towards.
Roanoke River basin was developed 20 years ago in conflicts between NC and VA, it is being
updated and will be practically a new model. We expect that model to be under contract in
next few months. Catawba basin‐ we’ll partner with Catawba water management group to
update the model and water resource plan there. Later in summer hopefully will move into
west, Hiwassee and Little TN, we’ll develop a model in partnership with TVA. We’ll have
models for all major basins done by 2018. This is why classification is important for a
statewide perspective, think about that geographic area and the area we need an evaluation of
eco‐flows.
Tar‐ Pam basin‐ 4th largest river basin, in NC covers 6,000 sq miles. The color codes on this
map are 8 digit HUCs (DWQ calls them subbasins). Blue circles are surface water
withdrawals…our models do not function in tidal areas. We don’t find surface water
withdrawals in salt waters. So we’re not able to model ecological integrity below a certain
part‐ this model ends downstream of Greenville. Anywhere there is 100,000 gpd withdrawn
or wastewater being discharged, there will be a node in the model, where we can make a
calculation/evaluation on eco flows. This is our Tar River basin hydro model schematic
(shown above). Yellow nodes‐ only some are where calculations can be made. It’s less
complex than the Neuse. If you combine that with all the other models and hundreds of nodes
and different types of habitats, that’s why classification is important. Appreciate the work on
trying to incorporate the biology. If it comes out that eco integrity has one method for finding
it, we may not need it. We’ll assume there will be differences in how systems respond to
changes in flow.
I’ll answer questions or turn over to Michele. She’ll explain this handout that shows Swift
Creek and how WaterFALL is being used.
Q: In lower Tar‐Pamlico, they have a regulatory model they use for withdrawals. Since OASIS
is a planning tool, a facility that uses a customized model, will they continue to use that
model?
R: An instream flow study was done there, that is a good example of a coastal stream, down
near Greenville. We began it thinking it was similar to upstream. We did the study, which
EFSAB Meeting Summary ‐ 2/19/2012 Page 26 of 37
used a 2 dimensional salinity and DO parameter. The habitat is based on salinity, since that is
where you’re finding creatures. There is a gage in Greenville that you can have higher flows at
lower levels, not the normal discharge relationship because it has tidal influences. They are
looking at it in another step, our flows is an input into their model, then they look 30‐50 years
ahead‐ will it be reducing or changing the patterns? The 2 will work together. That is one of
the more complex areas.
Michele’s presentation:
OASIS does good job of explaining flows. WaterFALL was used to describe the stream
classification where there are no flow gages. In the original OASIS node diagram purple
arrows indicate places where OASIS requires time series of inflows for the model. They have
created a dataset of what is going in at that point. WaterFALL uses both land use and climate
inputs and can provide a different set of inflows than are currently being used so that the
issues of climate change and land use change can be examined with OASIS using a more
hysically‐based representation of the hydrology. p
EFSAB Meeting Summary ‐ 2/19/2012 Page 27 of 37
This is same OASIS schematic but represented with NHDPlus catchments for WaterFALL. The
most downstream gage is the Tar River at Greenville. WaterFALL determines what should
naturally be going into the inflow points based on rainfall, land use, future land use or climate
change. All the Grey lines are NHD Plus catchments. There are 4,556 catchments in the Tar
River watershed within the WaterFALL model. Therefore, there are 4,556 segments which
S software to determine the stream classification. can be run through EF
Example on handout
(shown on p 27)‐ Swift
Creek (green) is classified
as a reference watershed
by the USGS (We use their
GAGES II dataset for
guidance on reference
watersheds). We’ve been
using Swift Creek to do
some of our calibrations
and see how it extends to
the whole basin. I selected
a few random stream
segments upstream of the
gage and ran WaterFALL’s
output for these segments
through the EFS
classification software.
The classifications show
that (1) Mainstem is small
flashy (class D/pink), (2)
some tributaries are small
stable(class B/green), and
(3) some headwater
segments are classified as
small stable. Pink dots on
top are DENR monitoring
locations where an Index
of Biological Integrity was
recorded (with supporting
biological data). So we can
go in and classify each of
the sites.
In the Little TN basin,
Cartoogechaye Creek was
classified as small stable
with the USGS gaged flows
and confirmed with the
WaterFALL modeled flows.
Segments further
upstream of the gage were
also classified as small
stable (class B) using
EFSAB Meeting Summary ‐ 2/19/2012 Page 28 of 37
WaterFALL. In Cullasaja Creek, which didn’t have a gage, WaterFALL modeling also classified
it as small stable. Continuing downstream the Little TN gets to be a pretty big river but it’s
still classified as small stable. The quick example shows what happens when you can look
segment by segment along a river rather than solely at places with USGS gages. RTI will be
exploring questions about variability in estimates, validity of comparing to gaged flows, and
xtending calibrations past the gaged sites over the next couple of months with WaterFALL. e
Discussion/Questions/Comments for speaker
Q: How do you determine how good a fit is to a classification?
R: We’ve done several things ‐ looked at gage flows and flow duration curves, comparison to
daily flows. ..then we look at EFS classification and the individual metrics. So far we’ve found
we can do pretty well with everything but reversals, and a couple of the variability metrics
…other than that, the low flow metrics, high flow metrics match pretty well.
Jim: We’ve found similar things looking at OASIS data, there are some metrics that appear
different depending on whether using USGS or OASIS data. There are limits as to how well
you can simulate real natural variability on such a short time scale. On the Swift Creek
portion of the map it struck as counter intuitive that upstream was small stable, and
downstream was small flashy. Some thoughts about that‐ remember that small flashy and
small stable were labels that some of us put on Class B and Class D. EFS thought we shouldn’t
do that, they thought it may create expectations of what classes should be. To tweak to line
up with biological reality, we may want to tweak those that classify as small flashy in the east.
If small flashy comes out that way hydrologically as you go east, bigger rivers are still called
small flashy, as you head west it means something else.
C: Olivero’s classification in the east would describe it whether it were small or not. It’s
possible that flashy and stable are indicative of hydrology, there may be other characteristics
of the classes to draw from that may be more accurate, if we need to describe it rather than
just use alphabetical labels. Swift Creek headwaters being stable; maybe that is the
groundwater contribution. The higher order streams may have overland and runoff flow
override the groundwater and contribute to flashiness (high flows). You can call headwaters
stable, but they’ve been dry at periods of time (hyporheic flows).
Jim: I’ve thought that too. And in the slatebelt stream, the flashiness may be the stream
bottoming out.
Q: How long did this take you to do? It would help me to see the entire stream‐ if there are all
different colors I don’t know what that means. Is this a big deal?
R: This is the first time we’ve done this, we’re working out some post‐processing bottlenecks.
Running WaterFALL is easy, it is the post‐processing that gets you. This took me over a week
of working all day to do this. It’s a manual process but can be automated. A batch processor
has been ordered, it will run much faster. Once it’s automated, filling in Swift Creek would be
day’s work. a
EFSAB Meeting Summary ‐ 2/19/2012 Page 29 of 37
VIII. Habitat modeling for evaluating ecological flows Update
Jim Mead presented methods he is using to illustrate the flow scenarios, which include
changes based on EFSAB feedback. Since the October/November meeting, the following has
been done
• Added 5 more sites, so 7 sites now: Eno River SP, Buckhorn Creek, West Fork Eno River,
have 3 First Broad Upper, First Broad Middle, First Broad Lower, Buffalo Creek. We now
small flashy and 4 small stable streams.
• e Revised output to separate the count of guilds that are < 80% and > 120% of th
unregulated index value for each flow scenario
for the same 19 guilds or species • Revised output so that all sites are evaluated
Jim asked the EFSAB for the following feedback:
• The output for comparison of all sites is in a different graph format. What does the SAB
think about this?
• Based on Mary Freeman’s comments about transferability of habitat preferences between
different sites, Jim did an analysis using the same 11 guilds and species instead of 19.
k about These 11 focus on the shallow habitat / riffle organisms. What does the SAB thin
this?
• One potential concern about focusing on 11 or 19 species for all sites is that the
anadromous (e.g. shad) species and rare/threatened/endangered species (e.g. Cape Fear
Shiner) will not be found at all sites ‐ and therefore drop out of the analysis when focusing
on the 11 or 19 that are common to all sites. What does the SAB think about this?
Jim’s presentation:
I don’t want to get into details or numbers or results today. I have a sampling of scenarios to
ing. show formats, to let you know what I’m working on, and to set the stage for our next meet
Jim reminded the SAB how he had previously presented data, using some example slides.
(Please see EFSAB Meeting Summary October 18 2011 for his last presentation.) For a very
brief reminder:
Slide 1 shows each color as a different flow scenario ranging from 10‐30% of ambient flow
being withdrawn. The numbers in white boxes at the bottom are the habitat index to show
the % difference. We are looking for the gross number of habitat units, so for percent
difference s. you need to also think about the magnitude
Slide 2 shows when we started lumping by seasons.
lated. Initially we lumped numbers or percent guilds < 80 or > 120% unregu
Now they are split out separately. Both analyze the 15 flow scenarios
Slide 3 at. shows percentage of guilds with < 80% of unregulated habit
lide 4 shows percentage of guilds > 120% of unregulated habitat. S
EFSAB Meeting Summary ‐ 2/19/2012 Page 30 of 37
lide 1
Slide 2
lide 3
Slide 4
S
S
EFSAB Meeting Summary ‐ 2/19/2012 Page 31 of 37
Another change was to make all the sites comparable, by having a common set of guilds and species
analyzed for all the sites (this was a good observation by the EFSAB). Originally we did the sites for
specific purposes; we may have run a particular set for a particular reason. To be able to compare
these sites and lump together to see if there are trends, we want to run the same suite of guilds and
species at each site.
There are 7 sites now. Previously we had only 2. I’m trying to maintain a summary page of each
poses and documentationsite for our pur . Appendix 1 has the list of 7 sites.
Broad River sites are small stable streams.
The following slide shows a summary of the information for the 15 flow scenarios for the 7 sites.
EFSAB Meeting Summary ‐ 2/19/2012 Page 32 of 37
dot for each of the 7 sites and a mean of the 7 sites.
how eco‐flows are
phs, I can see how that
Slide 5
There is a
A previous speaker said classification doesn’t make much difference with
determined. I didn’t necessarily agree at the time, but looking at these gra
might be possible.
An example of how to read these graphs, from Slide 5 (19 Guilds): Look at the First Broad, Upper
which is the blue square in the 70% of inflow as flow‐by (5th flow regime from the right). Roughly
50% of the 19 guilds and species have less than 80% of the natural habitat metric when you
withdraw 30% of the ambient flow (leaving 70%).
Any comments on these graphs? Is this a meaningful way to visualize results?
Comment (C): You did have both flashy and stable. It would be nice to set up a scheme where ea
h
EFSAB Meeting Summary ‐ 2/19/2012 Page 33 of 37
ch
es.
class of stream has a particular symbol, then use different colors from different streams or reac
So then you can compare the flashy and small stable.
Response (R): Good. Trying to have a common convention for class would be helpful.
ld Q: There are 2 different classes represented here. Would it be prudent to run a statistic that wou
identify whether or not the classifications actually help describe the variation, or whether lumping
them together is better?
C: If you did a simple box whisker plot for each one of them, comparing the Bs and Ds you would
get a very interesting thing. Bs and Ds overlap pretty substantially at 10% but after that, they are
very different. This may be a meaningful summary statistic.
R: I think it will help us understand if there is consistent response.
Slide 6
Jim’s question for the EFSAB‐ What about using 11 vs 19 guilds/species?
Jim explained these graphs are similar except for the # of guilds. Mary Freeman mentioned
transferability and habitat preference from one location to the next. She indicated the preference
values tend to be more transferable for shallow riffle type guilds than the deep guilds. One
hypothesis is that deep water species have different responses, hiding here or there. In riffles there
are fewer options and they tend to deal in the same way. Her research indicated if you transfer
from one site to the other it’s more robust for shallower environments like riffles.
en redhorse
EFSAB Meeting Summary ‐ 2/19/2012 Page 34
decision as a science advisory board as to how we interpret the information. One way
weighting it, we want to look at it that every one of those is important, to lose x% of th
of 37
edhorse
e
19 guilds: includes 8 shallow, 6 deep, 3 benthos families (EPT), 2 life stages of gold
11 guilds: includes 8 shallow, 3 benthos families (EPT), (Dropped the 6 deep and 2 golden r
stages.)
On the plus side, they are more likely to have transferable habitat preferences. The minus, we’ve
lost a broader range of things. It’s still being done as a site by site evaluation for specific projects. I
don’t see a way to include them when lumping them in a pot, to see consistent response across
locations. For listed and anadromous species they are not regular across sites. If you want to have
comparable, may be better to have less. I don’t want to do them both going forward.
Q: One problem I have about using a % of the 11 or 19 guilds is I feel like you have to look at all the
shallow guilds together, and then the deep guilds together to see if you get responses within the
deep and the shallow. If you have 6 shallow guilds, that is 6 units within the percentage that are
going to respond. If you only have 3 deep, it could overwhelm the deep responses. I’m trying to
think this out. You want to look at shallow responses to the sites…do you want to make them
unique analyses rather than lumping them together. Will it obscure information?
R: You have a good point. The red x and black + on the graph are both low gradient deep habitat
sites. There are just no shallow areas at that site. We picked the only spot we could wade, and it
was still deep. There was a difference at those sites‐ if you focus just on the shallow ones, they fall
out. What you’re saying, instead of having a choice of 11 or 19, could you have 11 shallow and 8 not
shallow and have 2 plots.
C: If you just include shallow you’re biasing towards higher velocity and higher flows, so my gut
reaction is to include gamut of the species or those guilds.
C: When you overlay them you look at the means, you would think that with the shallow guilds,
that the % would be higher, but they are not. I think splitting them out would help us understand
what is going on.
Q: In terms of how these are represented‐ seems like your guilds are discrete‐ they are affected at
80% or reduced or not. (It is binary.) But the % habitat represented by preferred habitat is
proportional… Is there a way to incorporate a weighting factor for the % of habitat that is
represented in the stream site?
R: You struggle to condense this, but have we lost nuance, given everything the same weight. (That
is a problem with proportional data when presented with binary responses). I need to think about
. it. We want it to be a digestible amount and not have the flow scenarios for every guild separately
C: If there is a critical species that only lives in the marginal habitat we need to know that.
Q: I’d include them. In the context of planning and purpose of end product, could you conceive of
giving a presentation where it would be important to separate the 11 from the 8, as opposed to the
broad brush of the 19 and the inclusive community? As opposed to the 2 subsets of the
community?
R: I can’t see it being important for presentation purposes; it may be important for our evaluation
for the data. The deep group response changes the way you interpret the data. Segregating deep
and shallow for our purposes may have some value. I can’t see other folks being as interested as w
are.
C: This is a discussion we’ll need to have. The point is good about how do we want to make a
is instead of
at number at
some point is unacceptable from an ecological perspective. If you weight it to say it’s ok to lose
some of these since there is not a whole lot of them anyway ‐ that is a philosophical debate we will
EFSAB Meeting Summary ‐ 2/19/2012 Page 35 of 37
could look at the data and see if I can put something
: Something concerns me. Let’s look at the 85% flow by of the11 guilds. I don’t know that the
r if
m
ng
. Is
nd
Q: I
lines as
R: Not
t
If
you
R: Whe l falls apart.
need to have. Our charge is not to say how much of an impact is too much of an impact.
R: Yes, we are counting on the SAB to provide ‐ what is an acceptable difference?
C: I thought another group after us would tackle those kind of judgmental policy decisions
following our efforts.
water. But
unacceptable. We
e can say wh
R: To some degree, in determining what is acceptable in context of other demands on
we have decided preliminarily to make the limits below 80% and above 120% as
ay what is allowable in broad perspective, but from ecological integrity wwon’t s at
keeps a percentage of guilds within a natural metric. What is that percentage? Is it that fewer than
I 20% of 19 guilds are outside of 80‐120 range? Or fewer than 50%? These are the questions
thought we’d answer.
Q: Can the problem with proportional difference be addressed with some kind of normalizing ‐ by
flow or amount of habitat?
R: One possible way would be to further normalize them all first, then go with percentage change. I
need to think about that more.
: Could you do a similarity index approach? Look at similarity at baseline, incorporate all the
ifferent guilds at the same time and then say we’ll accept similarity down to 80% or some number.
ook at what 80% means ‐ if you lose a species that is different than reducing all species…will
lify s well. I
Q
d
L
s
t
Q
th
imp and normalize things a
ogether that makes sense.
10% gu
ey are
ilds influenced at that point are the same ones being affected at 75% plus more guilds? O
completely unique set of guilds that were being impacted? Are they unique, or are they
additiv 10% of guilds the same for one streae as you increase the withdrawal? Basically, are the
0% is 1 out of 11.
as for another?
R: No it is just a count. The 1
C: I’m conc
hammered,
erned we’re lumping the numbers, we don’t know it’s the “shallow fast” that are getti
t. That’s the consistency that I’m most interested in
the habita
streams. May e
for example. We don’t know tha
R: It may
t response similar across the classifications?
up with cl
be that these results also feed into our thinking about how to subdivide
are a grou
osed loop, but how they respond to changes in flows may mean that how these respond
p, or a class.
C: Th ta would help us respond to your question of what is acceptable. If we find that this guild is
always the one being hit, it would help to inform us.
se noticed you used index B, is that between 10‐90% habitat? Was that agreed upon? If you u
the entire range of habitat, at the low end do you have thoughts about how that would affect the
far as bottom habitat values?
off the top of my head. Those tables have medium, index a, b, c. When we’ve looked at the
sites and all 3 indices, they’re within 10‐15% in terms of % unregulated. People have argued
against us ing lower flow numbers, but i
mpact to species.
using index C because they thought it meant recommend
doesn’t. Looking at specific project or permit we look at multiple indices.
C: If you w ack off the lower range of those it seems like you might have more i
use the whole range of habitat values.
n y u simulate habitat and you get into really high flows the mode
h
o
EFSAB Meeting Summary ‐ 2/19/2012 Page 36 of 37
dback. Thank you. I’m leaning towards 11 and 8 both deep and shallow.
iscriminates by stream class. I’ll think about getting around
amount. Highlight or
ro that would be good
pecifics if needed.
Jim: I got some good fee
Box plots, a symbol convention that d
the proportional habitat issue.
C: Perhaps separately point out those guilds that are affected by more than x
an a threshold (if it goes to zeput a star on a guild that is affected more th
to point out)
Jim: Do you still want to see the stacked bar charts? How much stuff do you want to wade through?
I’ve generated them for all 7 sites so far. Should we focus on what I’ve presented today?
: Can you make it available on the website to review? I would like to look at it.
im: I’ll think about how to library it into our website and let everyone know where it is. As a
roup we’ll focus on the more condensed stuff, but can get back to s
Q
J
g
Jim: Finally, the other handout ‐ Appendix 2 ‐ is a list by basin where we have habitat sites.
Because of where we have operating OASIS models, I’ve run one Cape Fear and 2 Neuse basin small
flashy streams, and 4 Broad basin small stable streams. The operating OASIS models we have right
now are the Neuse and Tar. Next ones I’ll run will be the Tar, and if we get Cape Fear for Rocky
River site (near Siler City)that modeled really well, and Neuse near Tear Quarry. I went with Broad
because the Oasis model was ready to run and they were recent sites easy to bring up to speed. If
you think of anything else let me know.
IX. April 24 Agenda
DRAFT AGENDA for April 24:
• Fidelity testing update (response by EDF, RTI to questions and suggestions from 02/12
meeting, and a progress update)
• Habitat Modeling scenarios presented by Jim Mead
• Proposal from DWR
Ideas for agenda that were raised at the February meeting:
• DWR proposes a trial balloon with things they might be looking for, EFSAB reaction
• Use small groups for discussion
• Give EFSAB a chance to discuss the data in small groups
• We need to look at different approaches to analyze the biological data
• I'm pretty comfortable with different frameworks about modeling habitat differences,
but we need to discuss and determine how the biology relates and changes
• Tom: if RTI has a chance to look at dataset then I can do some work on. It would be nice
to determine how we can work with the data ahead of time to determine things.
• There is a host of knowledge and information that we already have that we are
excluding from our database. I’m concerned about dropping out the ecological data
available and just looking at presence/absence data. but, I'm OK with waiting to see
what comes up.
• We shouldn't drop to presence absence if you have at least semi qualitative data. I’d
prefer to have both those analyses, and not just the presence absence data.
• Can RTI present any more data via the waterfall model
• RTI‐ we may be able to have things like abundance measures
EFSAB Meeting Summary ‐ 2/19/2012 Page 37 of 37
X. 9BDirections
On April 24, 2012 we will meet from 9:45 – 4:15 at the:
Stanford M. Adams Training Facility at Jordan Lake Educational State Forest
2832 Big Woods Road, Chapel Hill, NC 27517
Map link: HUhttp://g.co/maps/7zx5dU
From Rt 64 and Big Woods Road, it will be the first Forest Service sign on the right. Pass the office
building and continue on through the gate to the education center.
1
Following is a listing of all of the Division of Water Resources’ existing aquatic habitat
model study sites, sorted by hydrologic stream classification. This is also shown on a
google earth map through DWR’s Ecological Flows Science Advisory Board web site at:
http://www.ncwater.org/Data_and_Modeling/eflows/sab/presentations/20110517/
The listing constitutes our best judgment on stream classification. However, it is
important to note that many sites do not have a nearby USGS gage with an unaltered flow
record. If these sites are also in a river basin for which a hydrologic model is not yet
available, then in most cases there is no flow data to analyze with the stream
classification software. These sites are marked with an asterisk and the classification is
not yet confirmed (as of 2/1/2012).
Existing Habitat Models for Small Flashy (D) Stream Classification
Neuse River Basin
1. Eno River State Park – 99.4 square miles
2. Eno River, Hillsborough – 66 square miles – (convert from mainframe model)
3. West Fork Eno River – 11 square miles
4. Swift Creek – 74.1, 80.7, & 116.4 square miles – (consultant 2D model) *
5. Middle Creek – 38.2 & 63.8 square miles – (convert from mainframe model)
Cape Fear River Basin
6. Buckhorn Creek, below Harris Lake – 76.3 square miles
7. Rocky River, Siler City Reservoir – 55 square miles *
Tar River Basin
8. Tar River, Louisburg – 437 square miles
* small flashy classification to be confirmed
2
Existing Habitat Models for Small Seasonal (G) Stream Classification
Neuse River Basin
Cape Fear River Basin
Tar River Basin
Broad River Basin
* small seasonal classification to be confirmed
Existing Habitat Models for Coastal (A) Stream Classification
Neuse River Basin
1. Little River, upper site, 55 mi² (consultant, study in progress)
2. Little River, middle site, 90 mi² (consultant, study in progress)
3. Little River, lower site, 191 mi² (consultant, study in progress)
4. Neuse River, near quarry, 787 mi² *
Cape Fear River Basin
5. Nicks Creek, 26 mi²; *
Tar River Basin
6. Tar River, lower site, 2620 mi² (consultant, study in progress) *
* coastal classification to be confirmed
3
Existing Habitat Models for Large Piedmont (E) Stream Classification
Roanoke River Basin
1. Roanoke River, bypassed reach, 8371 mi² *
2. Roanoke River, upper reach, 8384 mi² *
3. Roanoke River, north channel, 8402 mi² *
4. Roanoke River, south channel, 8407 mi² *
5. Roanoke River, below Weldon, 8432 mi² *
Catawba River Basin
6. Catawba River, below Hickory Reservoir (Oxford Dam), 1314 mi² *
Yadkin – Pee Dee River Basin
7. Pee Dee River, below Tillery Dam, 6051 mi² *
8. Pee Dee River, below Rocky River confluence, 6303 mi² *
9. Pee Dee River, above Blewett Falls Dam, 6694 mi² *
10. Pee Dee River, below Blewett Falls Dam, 6863 mi² *
11. Pee Dee River, lower N.C. site, 7175 mi² *
* large Piedmont classification to be confirmed
Existing Habitat Models for Large Stable (C) Stream Classification
French Broad River Basin
1. French Broad R.; below Craggy Dam, 966 mi²
Little Tennessee River Basin
2. Tuckasegee River, lower middle site, 598 mi² *
3. Tuckasegee River, lower site, 655 mi² *
* large stable classification to be confirmed
4
Existing Habitat Models for Small Stable (B) Stream Classification
Broad River Basin
1. Buffalo Creek, below Kings Mtn. Reservoir – 127 mi²
2. Roberson Creek, 15 mi² *
3. 1st Broad R.; upper, upstream of Knob Crk., 145 mi²
4. 1st Broad River; middle, downstream of Shoal Rock Crk., 202 mi²
5. 1st Broad River; lower, downstream of Hwy. 74, 230 mi²
Catawba River Basin
6. Catawba River, Muddy Creek confluence below Catawba Dam, 98 mi² *
7. Catawba River, Linville Dam tailrace, 383 mi² *
French Broad River Basin
8. Davidson River, 13 mi² *
9. Hominy Creek, upper site, 91 mi² *
10. Hominy Creek, lower site, 95 mi² *
11. Jonathan Creek, 13 mi² *
12. Mills River, 71 mi² *
13. North Fork Mills River, 10 mi² *
14. Bradley Creek, 10 mi² *
15. Ivy “River” Creek, 112 mi² *
16. Cedar Rock (“Grogan”) Creek, 2 mi² *
Hiwassee River Basin
17. Nottely River, 245 mi² *
5
Existing Habitat Models for Small Stable (B) Stream Classification cont’d
Little Tennessee River Basin
18. Cartoogechaye Creek, 51 mi²
19. Dicks Creek, 3 mi² *
20. Nantahala River, bypassed reach between dam & Dicks Ck., 91 mi² *
21. Nantahala River, bypassed reach between Dicks & Whiteoak Cks., 103 mi² *
22. Nantahala River, bypassed reach below Whiteoak Ck., 128 mi² *
23. Queens Creek, 4 mi² *
24. Tuckasegee “East Fork” River, lower site, 81 mi² *
25. West Fork Tuckasegee River, upper site, 39 mi² *
26. West Fork Tuckasegee River, middle site, 53 mi² *
27. Whiteoak Creek, 13 mi² *
28. Wolf Creek, 15 mi² *
Lumber River Basin
29. Drowning Creek, 186 mi²
New River Basin
30. Middle Fork South Fork New River, 12 mi² *
Savannah River Basin
31. Toxaway River, 16.9 mi² *
32. Horsepasture River, upper site, 24 mi² *
33. Horsepasture River, lower site, 24 mi² *
Yadkin River Basin
34. South Yadkin River, 117 mi²
* small stable classification to be confirmed
6
Existing Habitat Models for Medium Stable (F) Stream Classification
Broad River Basin
Roanoke River Basin
1. Dan River, at Joyce Mill, 72 mi²
Catawba River Basin
2. Catawba River, near Morganton, 584 mi² *
French Broad River Basin
3. Pigeon River, upper site, 459 mi² *
4. Pigeon River, middle site, 479 mi² *
5. Pigeon River, lower site, 484 mi² *
Little Tennessee River Basin
6. Cheoah River, upper site, 178 mi² *
7. Cheoah River, upper middle site, 186 mi² *
8. Cheoah River, lower middle site, 200 mi² *
9. Cheoah River, lower site, 206 mi² *
10. Nantahala River, lower mainstem, 144 mi² *
11. Tuckasegee River, upper middle site, 360 mi² *
12. Tuckasegee River, upper site, 347 mi² *
Hiwassee River Basin
13. Hiwassee River, upper site , 198 mi² *
14. Hiwassee River, middle site, 252 mi²
* medium stable classification to be confirmed