Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141...
Transcript of Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141...
![Page 1: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/1.jpg)
![Page 2: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/2.jpg)
Presented atPresented at
Great Rivers Reference Condition Great Rivers Reference Condition WorkshopWorkshop
January 10January 10--11, Cincinnati, OH11, Cincinnati, OHSponsored by Sponsored by
The U.S. Environmental Protection Agency and The Council of StatThe U.S. Environmental Protection Agency and The Council of State Governmentse Governments
![Page 3: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/3.jpg)
The EMAPThe EMAP--GRE approach to GRE approach to reference conditionsreference conditions
Ted Ted AngradiAngradi and David and David BolgrienBolgrienUS EPA Office of Research and Development US EPA Office of Research and Development MidMid--Continent Ecology DivisionContinent Ecology DivisionNational Health and Environmental Effects Research LaboratoryNational Health and Environmental Effects Research LaboratoryDuluth, MNDuluth, MN
GIS programmers: Tatiana Nawrocki, Roger Meyer, Matt Starry, and James QuinnComputer Science Corporation
![Page 4: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/4.jpg)
Assumptions• No reach of any of the 3 Great River of the CB
(MO,MS,OH) is in pristine condition, but there is variation in condition along each river that provides the current scope for empirical bioassessment based on internal referencecondition.
• EMAP-GRE assessment is based on least disturbed (least impaired) conditions (LDC) by default. Other thresholds based on minimally-disturbed condition will be incorporated as available.
![Page 5: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/5.jpg)
Internal Reference Condition• Advantages
– Representative of the assessment unit– All indicators are available because same methods– Can extract reference sites from the probability
sample– All reaches and rivers can be treated the same
• Challenges– Perception that least disturbed condition sets a too-
low bar for Great River assessment– Sites are not independent (all sites but 1 on each river
are downriver from other sites)
![Page 6: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/6.jpg)
I. Identification and sampling of internal reacheslikely to be in LDC using “local proximity” model
II. Use abiotic filters to find sites actually in LDCfrom among all sampled sites
III. Verify reference set using biotic indicators
3-phased approach to reference
![Page 7: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/7.jpg)
EMAP-GRE GIS “Local Proximity” Model
• Scores large number of potential sites based on proximity to human disturbances
• Complete small-watershed-scale landscape analysis deemed not cost-efficient
• We treat all small tributaries as if they were NPDES permits (i.e., bad)
• So its conservative.
• Lots of subjective decisions were made (scoring thresholds, weights) during the build. The result is a really GIS-model-assisted BPJ approach to identifying potential reference reaches.
![Page 8: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/8.jpg)
Rationale for “local proximity”
• Most EMAP-GRE biotic metrics are measured in shallow nearshore habitat (the exception is plankton).
• Human disturbance: tributaries, NPDES permitted outfalls, crossings, etc. most strongly influence the immediate downriver near-shore condition (sediment contamination, water chemistry, sediment particle size, temperature).
• Cumulative effects of thalweg (non-local) WQ and habitat ignored in our model.
• This gradient is dealt with during filtering (Phase II).
![Page 9: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/9.jpg)
Cumulative upstream effects
Local effects
NearshoreEMAP-GRE
sample locations
Goal is to find and sample this location
![Page 10: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/10.jpg)
Points were generated for each river at 500-m intervals starting from the downriver end using the National Hydrograpic Database linework. These points serve as a base layer for all proximity analyses.
RiverNo of Sites
OH 3129MS 2785MO1 702MO2 322MO3 2569TOTAL 9507
Base points
![Page 11: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/11.jpg)
Population/distance ratios for protected area polygonsPLAREADIST6B
Population/distance ratios for urban polygonsUBPODIST5
Protected land in a site neighborhood (% in 5-km radius)PROTPER6A
Route distance to nearest upriver primary tributary weighted proportionally to runoff from developed and cultivated watershedarea
PTRBAD7
Forest and wetland in site neighborhood (% in 5-km radius)FORWETPER11
Density of upriver NPDES permits (number/10km)NPDESDEN13
Impervious surface in site neighborhood (% in 5-km radius)IMPERVPER14
Density of upriver secondary tributaries (number/10 km)STRIBDEN12
Cultivated land in site neighborhood (% in 5-km radius)CULPER10
Route distance to nearest upstream secondary tributarySTRIBUP9
Route distance to nearest upriver primary tributary weighted proportionally to runoff from undeveloped watershed area
PTRGOOD8
Route distance to nearest upriver permitted discharge (NPDES)NPDESUP4
Route distance to nearest upriver road or railroad crossingRDRLUP3
Route distance to nearest downriver damDMDN2
Route distance to nearest upriver damDMUP1
DescriptionVariableNo
14 variables calculated in the GIS Proximity Model
![Page 12: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/12.jpg)
Railroad and road crossings
![Page 13: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/13.jpg)
#
##
#
#
#
##
#
#
#
#
#
#
#
#
##
#
#
#
#
#
#
##
##
#
##
# #
##
#
Attributing points with land use (forest, cultivated land, impervious surface). 5 km radius.
![Page 14: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/14.jpg)
!(!(!(
!(!(!( !(!(
!( !(!(
!(!(!(
!(
!( !(
!( !( !(
!(
!(
!(!( !(
!(
!(!(!(
!(
!( !( !(
!( !(
!( !( !(
!(!(
!(!( !(
!( !( !(!(!( !(!( !(
!(
!( Primary Tributary Intersections
Primary Tributary Watersheds
Primary tributaries (drains at least one 8-digit HUC)
![Page 15: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/15.jpg)
0.010.031310056333930168967410.5Upper Iowa
0.030.041918787484697721087010.7Cuivre
0.010.04944683354959587598220.6Cannon
0.000.05584323545258911898820.6Zumbro
0.040.062655706525643376376710.8South Fabius
0.020.071849917736193202877610.8Root
0.030.082379371526772057407220.9Turkey
0.030.111963762328946028328021.1Maquoketa
0.120.1568088472011109171746021.8Big Muddy
0.080.1547990735011197838176911.6Salt
0.030.1620932205111861582908321.4Wapsipinicon
0.050.2734369214118043837398132.1Skunk
0.100.4659509181129054482528123.5Kaskaskia
0.070.5244589507732698972318623.7Minnesota
0.140.7681553290346213531138235.4Des Moines
0.140.8880153164253640963788436.2Lower Iowa
0.130.9074498472554632213128536.2Rock
0.000.90611844348273750013747537198.7Missouri
0.000.9027193241801540950368480518.1Illinois
0.020.001391837182369884946300.4Trempealeau
0.150.068607579255981538124011.5Black
0.350.0918904024757351565182532.6Meramec
0.600.25326529014316821191643314.9St. Croix
0.790.37427187699824029308123516.7Chippewa
0.840.48452550306130170020413827.5Upper Mississippi
0.900.54482934832333559878174018.2Wisconsin
PTRGOOD weight
PTRBADweightOther m3DEVAG m3% AG% DevelopedQ km3Tributary
Model refinement: distance from primary tributaries was split into 2 variables:
PTRBAD distance weighted by a normalizedproportion of runoff from AG+DEV LULC
PTRGOOD distance weighted by a normalized proportion of runoff from other LULC
![Page 16: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/16.jpg)
St Croix RiverSt Croix River•Only 34% of catchment is Ag + Developed
•PTRBAD (close is bad) W= 0.25 •PTRGOOD (close is good) W= 0.6
For a site 10 km below confluence:
PTRBAD = distance * (1-W)= 10*0.75 = 7.5 km
= makes site score a little worse
PTRGOOD = 10 *0.4 = 4 km= makes site score a lot better
Net effect is that sites below St Croix Riverscore better than they would if there were
no St. Croix.
![Page 17: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/17.jpg)
Iowa RiverIowa River•87% of catchment is Ag + Developed
•PTRBAD (close is bad) W= 0.88 •PTRGOOD (close is good) W= 0.14
For a site 10 km below confluence:
PTRBAD = distance * (1-W)= 10*0.12 = 1.2 km
= makes site score a lot worse
PTRGOOD = 10 *0.86 = 8.6 km= makes site score a little better
Net effect is that sites below Iowa Riverscore worse than they would if there were
no Iowa River
![Page 18: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/18.jpg)
Yellowstone River:Turbid but “clean”
![Page 19: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/19.jpg)
Subjectivity: variable weights and scoring thresholds
0.5
1.0
1.0
0.5
0.5
1.0
2.0
2.0
NA
0.5
1.0
2.0
0.5
0.5
1.0
MississippiWeight Threshold
NA
2
2
NA
NA
5 km
NA
40 km
NA
NA
NA
2 km
2 km
5 km
10 km
0.5
1.0
1.0
0.5
0.5
1.0
2.0
2.0
0.5
NA
1.0
2.0
0.5
0.5
1.0
OhioWeight Threshold
NA
2
2
NA
NA
5 km
NA
40 km
NA
NA
NA
2 km
2 km
5 km
10 km
0.5
1.0
1.0
0.5
0.5
1.0
2.0
2.0
0.5
NA
1.0
2.0
0.5
0.5
2
MissouriWeight Threshold
NA
2
2
NA
NA
5 km
NA20–40 km
NA
NA
NA
2 km
2 km
5 km
40 km
PLAREADIST6B
UBPODIST5
PROTPER6A
PTRBAD7
FORWETPER11
NPDESDEN13
IMPERVPER14
STRIBDEN12
CULPER10
STRIBUP9
PTRGOOD8
NPDESUP4
RDRLUP3
DMDN2
DMUP1
VariableNo
Emphasizes the local effect because all of thevariation in scoring is forced into the first 2 km below an outfall
Score doesn’t get any better beyond 2 km downriver from permit location
![Page 20: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/20.jpg)
What the model actually does for each river:
• Scores all points 1-6 for each of 14 metrics• Scores all points based on an additive index
based on summed and weighted metric scores• Each site gets a normalized score of 1-10 where
1s and 2s are least likely to be in LDC and 10s and 9s are most likely to be in LDC:
1 2 3 5 6 7 89
10
4
![Page 21: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/21.jpg)
Model output – raw weighted sums of metric scores for every point
![Page 22: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/22.jpg)
Model output – normalized scores
![Page 23: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/23.jpg)
Model output – reference reaches
![Page 24: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/24.jpg)
Reference reach
Medium score reach
Low score reach
Goal of model is not to definitively find the verybest sites out there but to increase the probabilitythat our 2006-2007 sample will include a higherproportion of well dispersed LDC sites than astraight probability sample would.
![Page 25: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/25.jpg)
12%
12%
41%
27%
9%
22%
8%
37%
9%
19%
36%
0%
10%
21%
26%
Percent of length scored as reference by the model for arbitrary map units
![Page 26: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/26.jpg)
Next steps• Review reference reaches with our partners
– Identify counterintuitive results– Add back in known good reaches– Clip known impaired reaches– Adjust and re-run model
• Build a new probability design based on these reaches and draw sites.
• Sample sites in July-September 2006-07. • Use BPJ to pick some additional “off-frame”
LDC, tributary and “trashed” sites?
All GIS input data are public domain. Model is available to anybody who wants it.
![Page 27: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/27.jpg)
I. Identification of sites (internal reaches) likely to be in LDC using “local proximity” model
II. Use abiotic filters to find sites actually in LDCfrom among all sampled sites
III. Verify reference set using biotic indicators
EMAP-GRE’s 3-phased approach to reference
![Page 28: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/28.jpg)
Example of site-scale screening for wadeablestreams
For EMAP/REMAP wadeable stream datasets, the reference sites are generally screened on various chemistry and physical habitat variables with criteria and screens varying by region.
It works for wadeable streams but its hard to apply to GRE data
![Page 29: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/29.jpg)
EMAP-GRE Variation: “multi-metric natural gradient approach”
• Assign a score to each sampled site while explicitly considering a natural gradient for each river.
• Allows different expectations for different parts of the gradient.
• Basic idea is from Thom Whittier, Dynamac, Corvallis, OR and co-authors.
![Page 30: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/30.jpg)
The concept
Natural gradient
Abi
otic
met
ric jFor some abiotic filtering metricsthere is a strong natural gradient
![Page 31: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/31.jpg)
Natural gradient
Abi
otic
met
ric j
Traditional hard criteria approach would excludesites partly because of the natural gradient
LDC
HDC
![Page 32: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/32.jpg)
Natural gradient
Indi
cato
rjInstead base the expectation on
the gradient itself
LDCj
HDCj
![Page 33: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/33.jpg)
Assignments• LDC (green) or HDC (red) assigned to sites based on
distribution of sites relative to natural gradient (graphical approach in this example)
• Working rule of thumb: try to get 10-20% of total sites into LDC for each metric
• But you don’t need the same % for each metric
• If there is no natural gradient for a metric, use river-wide or strata-wide criteria to assign
• Gets easier with more sites because total variation increases
![Page 34: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/34.jpg)
River mile above confluenceis an adequate (and free)surrogate for watershed areafor every sampled point
Provides a single gradient for all three rivers
Actual 2004 sites
Gaps are reservoirs
![Page 35: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/35.jpg)
• Total P• Total N• NH4• SO4• Chloride • DO• Large woody debris density• Riparian development score• Human disturbance score• % Canopy density at river’s edge• Riparian vegetation coverage score• Sediment toxicity (% survival of H. azteca)• Distance to upriver NPDES permit (GIS model output)• Weighted distance to upriver primary tributary (GIS model output)• Distance to upriver secondary tributary (GIS model output)• Will add dissolved metals (data not available yet)
Filtering metrics in this example
![Page 36: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/36.jpg)
2004 sample data: DO on Missouri River
Scored as LDC
Scored as HDC
![Page 37: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/37.jpg)
2004 sample data: TN on Missouri River
![Page 38: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/38.jpg)
2004 sample data: NH4 on Missouri River
![Page 39: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/39.jpg)
2004 sample data: TP on Missouri River
![Page 40: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/40.jpg)
2004 sample data: Chloride on Missouri River
![Page 41: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/41.jpg)
2004 sample data: Sulfate on Missouri River
![Page 42: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/42.jpg)
2004 sample data: Riparian development on Missouri River
![Page 43: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/43.jpg)
2004 sample data: Human disturbance on Missouri River
![Page 44: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/44.jpg)
2004 sample data: LWD on Missouri River
![Page 45: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/45.jpg)
2004 sample data: Sediment toxicity on Missouri River
![Page 46: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/46.jpg)
2004 sample data: Riparian canopy on Missouri River
![Page 47: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/47.jpg)
2004 sample data: Riparian vegetation score on Missouri River
![Page 48: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/48.jpg)
2004 sample data: Distance from upriver NPDES Permit
![Page 49: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/49.jpg)
2004 sample data: Weighted (PTRBAD) distance to upriver primary tributary
![Page 50: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/50.jpg)
2004 sample data: Distance to upriver secondary tributary
![Page 51: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/51.jpg)
For each site, take the difference in number of metrics in LDC and the number in HDC to get
a net score ranging from 15 to -15
![Page 52: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/52.jpg)
Weighting “robust”metrics spreads
data out more
These sites are usedto validate approach
![Page 53: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/53.jpg)
2004 sample data: TP on Missouri River
Q event?
![Page 54: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/54.jpg)
Discharge and WQ
![Page 55: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/55.jpg)
Water column (flow-integrated)chemistry is problematic for screening
dilution dilution
Concentration Concentration
![Page 56: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/56.jpg)
River mile
Gradients strongest on Missouri, but occur in all 3 rivers
![Page 57: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/57.jpg)
Multi-metric natural gradient approach:• Reveals human disturbance masked by natural
variation along river• Essential for some abiotic variables besides
WQ.• Avoids arbitrary “hard” criteria for single metrics.• Can be used in conjunction with other methods
(e.g., pass/fall).• Its integrative because it includes, WQ, habitat,
stressor, and landscape metrics.
![Page 58: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/58.jpg)
Multi-metric natural gradient approach:• Its robust because its “multimetric” A few
event-driven, bad, or missing classifications won’t effect results much.
• But, water chemistry may not be very useful for filtering
• Avoids “bipolar” sites in reference set • Forces reference sites down the gradient• Intuitive but untested on large rivers.
![Page 59: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/59.jpg)
Alternative approach
The same data can be explicitly stratified into upper, middle and lower MO and filtered using a series of hard filters (fail 1 filter = not LDC; pass all = LDC).
We will do both and figure out which one gives us the best set of reference sites.
Its research.
![Page 60: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/60.jpg)
Least Most Least Most Least Most
% In
divi
dual
s
0
20
40
60
80
100 Xeric Mountains Plains
Least Most Least Most Least Most
# Ta
xa0
10
20
30
40 Xeric Mountains Plains
Multimetric indexTaxa richness
Num
ber o
f tax
a
MM
I sco
re
Ohio Mississippi Missouri Ohio Mississippi Missouri
Validation of approach: least vs. mostdisturbed for a variety of filter options
Not real data
![Page 61: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/61.jpg)
Geographic stratification?
Lower MO
Middle MO
Upper MO
![Page 62: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/62.jpg)
Geographic stratification: different reference expectations for different reaches of the same river With stratification:• Expectations more realistic for each strata
(=reach)• Fewer political problems: “don’t assess my state
with your reference conditions”• Requires well dispersed reference sites.Without stratification:• Less complicated (only 1 LDC for each river)• Much higher variation in condition between states• Reduced credibility of assessment given realities
of BAC?
![Page 63: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/63.jpg)
Un-natural revetted shoreline Natural shoreline
Should we treat revetment as a substrate-coarsening stress or as a habitat Strata for this assessment so that we can detect other stress?
If we treat it as a human disturbance causing stress:• Most of the Lower Missouri will automatically be impaired • Revetment effects will probably swamp out other stressors for benthos at least• Sets higher but maybe unattainable bar for LDC
If we treat it as a habitat strata:• Will allow correction for substrate to detect other stressor effects.• Sets low bar for current system, but we can report the extent of rip rap and, if we restore a shoreline, we can switch to the other reference expectation
• Requires more sites (3 geog strata * 2 hab strata = 6 expectations for Missouri)
Revetment vs natural shoreline
![Page 64: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/64.jpg)
Some pending issues• Should we stratify habitat for revetted vs.
natural shorelines? (yes)
• Which variables from GIS model should we add to the filter? (NPDES, tribs, dams)
• Should we downweight or eliminate WQ metrics from filter because of the thalwegbias and discharge effect? (eliminate most)
• What is sufficient sample size? (lots more)
![Page 65: Angradi EMAP-GRE approach final [Read-Only] | US EPA ... · Skunk 2.1 3 81 1804383739 343692141 0.27 0.05 Kaskaskia 3.5 2 81 2905448252 595091811 0.46 0.10 Minnesota 3.7 2 86 3269897231](https://reader034.fdocuments.in/reader034/viewer/2022043009/5f9c33eb147a97271551d28d/html5/thumbnails/65.jpg)
• We are going to try GIS-based proximity model plus multi-metric filtering to find least disturbed sites.
• Least disturbed condition from internal reference sites will give us an assessment starting point for all our indicators on all 3 rivers, but we will use whatever other thresholds are out there.
• Priority for 06-07 is sampling as many likely LDC sites as possible!
Conclusions
Missouri River at Mandan, ND