1

1

Analyzing Data to Characterize Your Watershed

Thomas Davenportdavenport.thomas@epa.gov

2

Steps in the Watershed

Planning and Implementation

Process

3

Incorporation of the nine minimum elements

4

5

Water quality info & analysis• Water quality goals

– Designated uses, WQ criteria– Restoration and protection goals– Flooding, aesthetics, others???

• Monitoring and assessment results– Desktop data mining, local monitoring results– ID impaired & threatened waters

• Key pollutants / stressors– Check 303(d); local monitoring/assessment

• Pollutant sources– From 303(d) or other assessment

• Current loads– Estimate, model, or otherwise quantify

6

Types of Data for Watershed Characterization

• Physical and Natural Features – Watershed boundaries– Hydrology– Topography– Soils– Climate– Habitat– Wildlife

• Land Use and Population Characteristics– Land use and land cover– Existing management

practices– Demographics

• Waterbody Conditions– Water quality standards– 305(b) report– 303(d) list– TMDL reports– Source Water Protection

Areas• Pollutant Sources

– Point sources– Nonpoint sources

• Waterbody Monitoring Data– Water quality data– Flow data– Biological data

2

7

Flow data is available from the US Geological

Survey web site at http://waterdata.usgs.gov

/nwis/rt

8

T.C. Stiles, 2001; B.Cleland, 2002

9 10

How can we estimate loads?

11

What is a “load?”• Maybe measured by weight .

. . – Kilograms per day– Pounds per week– Tons per month

• Maybe not . . . – Concentration-based expression of the

“load” (e.g., milligrams per liter)• mg/L x L/day = mg/day [C = m/v]

– # of miles of streambank needing stabilization or vegetation

– # of AFOs requiring nutrient plans– % of urban area to be ‘perforated’

12

Existing loads come from lots of places . . .

3

13

Existing loads come from:

• Point-source discharges (NPDES facilities)– Info is available on the discharges (DMRs, etc.)– Some are steady-flow, others are precip-driven

• Nonpoint sources (polluted runoff)– All are (mostly) precip-driven– Calculating the “wash-off, runoff” load is tough– Literature values can be used to estimate– Modeling gets you closer . . . . do you need it?

• Air / atmospheric deposition– Can be significant in some locations

14

Steady-load: sewage treatment plant discharge via

infrared photography

15

Identification of causes & sources• What “pollutants” are you dealing with?

– Chemical or other stressors or causes of impairment

• How big is the problem for each?• How do you know?

– Did you “measure” them?– Did you estimate? How?

• Where are they coming from?– Can you put the info on a map?

• Can you estimate the % from each source?

16

Data-driven Approaches• Estimate source loads using:

– Monitoring data• Periodic water quality concentrations and flow

gauging data• Facility discharge monitoring reports

– Literature• Loading rates, often by landuse (e.g.,

lbs/acre/year)• Typical facility concentrations and flow

17

Is a Data-driven Approach Appropriate?• Monitoring data

– Does it represent most conditions that occur (low flow, storms, etc.)?

– Are spatial and source variability well-represented?

– Have all parameters of interest been monitored?– Is there a clear path to a management strategy?

18

Load Estimates – Monitoring Data• In simplest terms…

load = flow x concentration• Load duration curves

– Flow-based presentation• Statistical techniques

– Relationships between flow and concentration to “fill in the blanks” when data aren’t available

– Examples include:• Regression approach• FLUX

4

19

Load Duration Curves• Rank daily flow and generate flow duration curve• Multiply water quality concentrations by corresponding flow

values• Flow “curve” represents water quality target

5104

1

10

100

1000

10000

100000

1000000

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Observed Flow Exceedence at KAIN18

Tota

l Sus

pend

ed S

olid

s Lo

ad (k

g/da

y)

Allowable Total Suspended Solids Load (kg/day) Observed Total Suspended Solids Load (kg/day)

Observed (Surface Flow > 50%)

20

Load Estimates – Literature• Landuse-specific loading rates (typically annual)

• Multiply loading rate by area:loadall = (arealu1 x loading ratelu1)+ (arealu2 x loading ratelu2) +…

• Generally for landuse or watershed-wide analysis

• Many sources: Lin (2004); Beaulac and Reckhow(1982), etc.

• Use with caution (need correct representation for your local watershed)– Pollution sources

– Climate

– Soils

21

Example Load Estimation Based on Literature Values

22

23

Limitations of Data-driven Approaches

• Monitoring data– Reflect current/historical conditions (limited use

for future predictions)– Insight limited by extent of data (usually water

quality data) • Often not source-specific• May reflect a small range of flow conditions

• Literature– Not reflective of local conditions– Wide variation among literature– Often a “static” value (e.g., annual)

24

Key Points:

• Many tools are available to quantify pollutant loads

• Approach depends on intended use of predictions

• Simplest approaches are data-driven• Watershed modeling is more complex and time-

consuming– provides more insight into spatial and temporal

characteristics– useful for future predictions and evaluation of

management options• One size doesn’t fit all!

5

25

Load assessment is is on ongoing learning process – iterative & creative!

26

Identify candidate practices

27

Identify Opportunities for Implementation• Impervious

analysis

• Political constraints and priorities

• Physical constraints

• Environmental constraints

28

Quantifying Load Reductions to Meet Objectives• Sources of load quantification data

– Watershed modeling/load quantification results (previously described), TMDL reports, etc.

– Source and spatial targets for implementation

Table 5-4. Nutrient and BOD baseline and allocation loads by contributing subwatershed for impaired waters of the Cedar Creek watershed (daily averages based on the 2001-2003 meteorological regime).

BOD TN TP Reach ID/Subwatershed Baseline

(lb/day)TMDL

(lb/day)Percent

ReductionBaseline (lb/day)

TMDL (lb/day)

Percent Reduction

Baseline (lb/day)

TMDL (lb/day)

Percent Reduction

1, Lower Cedar Creek 24.74 14.12 43% 31.49 17.51 44% 1.54 0.85 44% 2, Slaughter Creek 4.44 2.74 38% 4.74 2.74 42% 0.20 0.12 41% 3, Slaughter Creek 15.16 9.03 40% 23.76 13.31 44% 0.87 0.49 43% 4, Slaughter Creek 61.40 35.49 42% 97.83 54.47 44% 3.65 2.05 44% 5, Slaughter Creek 17.54 9.98 43% 30.81 17.09 45% 1.12 0.62 44% 6, Slaughter Creek 34.70 23.42 33% 34.38 20.68 40% 1.28 0.76 40% 7, Slaughter Creek 31.19 18.67 40% 47.92 26.91 44% 1.87 1.04 44% 8, Slaughter Creek 52.48 32.26 39% 77.65 43.68 44% 2.99 1.69 43%

29

9%

11%

13%

15%

17%

19%

21%

23%

$0 $200,000 $400,000 $600,000 $800,000 $1,000,000 $1,200,000Cost ($)

Run

off V

olum

e R

educ

tion

(% fr

om E

xist

ing

Con

ditio

n)

Trad

e-O

ff Cu

rve

Optimal Solution

Initial Run

Identify Optimal SolutionExample of BMP-DSS Multiple Run Output

30

Quantify BMP Options at a Subwatershed or Watershed Scale• Goals

– Quantify selected BMP strategies (i.e., individual BMPs or BMP pairings)• Watershed scale• Local scale

– Compare potential load reductions to target– Determine optimal strategy considering

environmental benefits and $$• Available Tools

– Spreadsheet tools– Watershed/site-scale models

6

31

Prioritizing/targeting BMPs

• Importance of waterbody– Drinking water source, recreational resource

• Magnitude of impairment(s)– Level of effort needed; public interest/attention

• Existing loads (stressors & sources)– Magnitude, spatial variation, clustering

• Ability of BMPs to reduce loads– Sure thing, or a shot in the dark?

• Feasibility of implementation– Willing partners? Public support?

• Additional benefits– Recreational enhancements, demonstration

32

What Tool Should I Select?• Has a model already been used for load

quantification?• What scale is important? • Is an annual load reduction estimate sufficient? • Should individual storms be evaluated?

• Spreadsheet tools– Normally good for annual/overall reductions– Usually at a watershed scale – sometimes at the site

scale• Watershed models

– Allow for continuous/long-term simulation– Often can be used for storm evaluation– Ability to function at all scales – site and watershed

33

Flashiness Index

Even without pollutant load data, can be used to quantify hydrologic impacts of watershed change on hydrologic regime and to evaluate efforts to restore natural flow regimes

34

Examples• Minimize the total volume of surface water runoff that flows

from any specific site during and following development, in order to replicate pre-development hydrology to the maximum extent practicable

• Achieve average annual 85% Total Suspended Solids (TSS) removal for the developed area of a site. Areas designated as open space that are not developed do not require stormwater treatment. All sites must employ Low Impact Development (LID) practices to control and treat runoff from the first inch of rainfall

0

5

10

15

20

25

11:00 AM 12:00 PM 1:00 PM 2:00 PM

Time (hours)St

ream

Flo

w (f

f3 /s

) Developed site hydrograph

BMP influence on hydrograph

35

Clinton River Watershed Site Evaluation Tool

• Evaluate potential benefits of BMPs at the site development scale

• Inputs– Site characteristics– BMP characteristics

• Outputs– Peak discharge– Annual runoff– Pollutant Loads

Nitrogen Rate

Target

0.002.004.006.008.00

10.0012.0014.0016.0018.00

1-yr 24-hr storm

01234567

11:00 AM 12:00 PM 1:00 P

cfs

Post, no BM PsExistingPost, with BM Ps

36

Why a Site Evaluation Tool?• Evaluate flow and water quality impact

of proposed residential and commercial development

• Identify most cost-effective suite of BMPs

• Support decision-making activities– Tool used in combination with other

data/information to make final management decisions at the site scale

– Promote consistency• Help with Phase 2 reporting

requirements

7

37

Example

38

Areal Loading Rates ExistingLanduse

Designwithout BMPs

Designwith BMPs Target

Meets Goal?

Total Nitrogen (lb/ac/yr) 0.66 9.56 7.17 6.00 NO!Total Phosphorus (lb/ac/yr) 0.11 1.53 0.92 1.33 YesSediment (ton/ac/yr) 0.011 0.120 0.018

Site is located in Urban Residential Nutrient ZoneTN loading rate is higher than the buy-down range of 3.6 to 6 lb/ac/yr

Nitrogen Rate

Target

0.00

2.00

4.00

6.00

8.00

10.00

12.00

Phosphorus Rate

Target

0.00

0.50

1.00

1.50

2.00

Sediment Rate

0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

39 40

Areal Loading Rates ExistingLanduse

Designwithout BMPs

Designwith BMPs Target

Meets Goal?

Total Nitrogen (lb/ac/yr) 0.66 9.56 3.26 6.00 YesTotal Phosphorus (lb/ac/yr) 0.11 1.53 0.30 1.33 YesSediment (ton/ac/yr) 0.011 0.120 0.001

Site is located in Urban Residential Nutrient ZoneTN loading rate is below the buy-down range of 3.6 to 6 lb/ac/yr

Nitrogen Rate

Target

0.00

2.00

4.00

6.00

8.00

10.00

12.00

Phosphorus Rate

Target

0.00

0.50

1.00

1.50

2.00

Sediment Rate

0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

41

Key Points in Quantifying BMP Performance:

• Quantifying potential impacts from BMPs is critical to watershed planning– Provides a guide toward achieving load reduction goal– Informs selection of a management strategy

• Spreadsheet and modeling tools are available– Spreadsheet tools

• Most useful for watershed-scale analysis• Operate on a large time step

– Watershed/site-scale models• Useful for local scale, as well as watershed-scale• Can operate on a short time-step (including individual storms)• Provide a key first step for engineering design

• Again, one size doesn’t fit all!

42

Examples of Different Scenarios to Meet the Same Load Target

8

43

phosphorus load

0

50

100

150

200

250

300

1 2 3 4 5 6 7 8 9 10 11 12

months

kg/d

ay

Typical monitoring program measures water quality at a relatively few points in time, then “connects the dots”.

Note: Total annual load is assumed to be the area under the curve. 44

However:

If storm events aren’t measured, the actual load may be much larger.

phosphorus load

0

50

100

150

200

250

300

1 2 3 4 5 6 7 8 9 10 11 12

months

kg/d

ay

45

dissolved oxygen

0

2

4

68

10

12

14

16

1 2 3 4 5 6 7 8

days

mg

/ lite

r

Many monitoring schemes miss daily variation as well.

Blue dots are typical late afternoon measurements of dissolved oxygen. In many productive systems, however, DO drops to very low values at night.

46

All Samples

0

20

40

60

80

100

4926

35

4926

29

4926

39

4926

38

4926

28

4926

36

4926

26

4926

37

Perc

ent

Total PhosphorusTotal Suspended Solids

Pollution Indicator Exceedence(Total P and TSS)

020406080

100

4926

35

4926

29

4926

39

4926

38

4926

28

4926

36

4926

26

4926

37

Perc

ent

Total PhosphorusTotal Suspended Solids

All Samples

Non Runoff Samples

Comparison of storm event data with non-storm data at 8 different urban stations in Utah

Note difference in percent exceedence of pollutant criteria.

Data from Rieke, Mesner and Gilles, 2005

47

Upstream/downstream monitoring demonstrates impact of BMP very effectively.

Rees Creek TSS load

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

50000

1 2 3 4 5 6 7 8 9

weeks

kg /

day upstream

downstream

48

But… will upstream/downstream monitoring capture all impacts?

9

49

Bear River phosphorus load

0

50

100

150

200

250

300

350

400

1 2 3 4 5 6 7 8 9

weeks

load

(kg/

day)

upstream

downstream

Loading from large feeding operation swamped by the large loads carried by the Bear River.

50

NPS issues: livestock, grazing, feedlots

NPS issues: excessive pasturing along stream

51

Findings: BMP implementations improved overall stream physical habitat conditions.

However:

Spring Creek (with relatively large amounts of BMPs in both upland and riparian) :

improved habitat, temps and fisheries

Joos-Eagle (with very few BMPs installed)

No change in fish or thermal regime

habitat only improved in localized areas.

52

RESULTSCooler water

temperatures in pools and deeper runs

Reduced width-depth ratios compared to unrestored reaches

Rainbow trout numbers increased in restored reaches, while constant or decreasing in unrestoredand control reaches

Trout Counted Each Visit During Study Period

Mean Mean±SE Mean±SD DARKL MCCL1 MCCL2 MCCLR MCCR

SITE

-20

0

20

40

60

80

100

120

140

Num

ber o

f Tro

ut

control before after

53

Findings:

• Fisheries improvement thru riparian bmps alone will only work if the watershed is already in good shape, or if upland bmps are installed.

• A minimum of 30-50% bmp implementation at a watershed scale may be necessary

54

10

55

Cross-Section Locations - Chalk Creek

5000

5050

5100

4825 4875 4925 4975 5025 5075Easting (m)

Nor

thin

g (m

)

left bankfulls 98right bankfulls 98left bankfulls 95right bankfulls 95X-Section G0X-Section G1X-Section G2X-Section G3X-Section G4X-Section G5X-Section G6

G0

G3

G4

G5

Transect Locations at Curtis Property

G2

56

Cross Section - G0

98.5

99.5

100.5

75.000 85.000 95.000

G0 98 G0 95 G0 94qbr est 98 qbr est 95 qbl est 94G0 03 qbr est 03

Cross Section - G1

98.5

99.5

100.5

35.000 45.000 55.000 65.000 75.000

G1 98 G1 95 G1 94qbr est 98 qbr est 95 qbr est 94G1 03 qbr est 03

Surveying demonstrates response of stream width, depth, lateral migration.

In these cross sections, stream has deepened and narrowed at BMP site.

57

1978-1989y = 0.037x + 0.0813R2 = 0.5294, n=75

1990-2003y = 0.0327x + 0.0385R2 = 0.513, n=123

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0 5 10 15 20 25 30

Discharge (m3 sec -1 )

Tota

l Pho

spho

rus (

mg

L-1)

.

1978-19891990-1999

The reports also utilize monitoring data collected by DWQ. In this example, response was measured as a change in the slope of the TP/discharge relationship.

58

Urbanization

59

OUTCOMESINPUTS OUTPUTS

Activities Participation Short Medium Long

term

Programmaticinvestments

i

EVALUATION

PLANNING

60

25

• Statistics is a vital part of any study design as it isusually the tool used to answer the question.

• The type of statistical method to be used is guided by yourquestion and the design of your study.

• Statistics is a separate course in itself.

• It is the data assumptions inherent in statistical methodsthat drives the development of so many sampling strategies.

• Knowledge of statistics is required to address sample sizeestimates and determine the cost of science-related programs.

Statistics:

11

61

29

• The issue of scale has been referred to as the centralproblem in ecology.

• There is no single correct scale at which to study somethingand the results you get are dependent on the scale you choose.

• A linear change in scale does not necessarily mean a linearchange in the measured value. Scale dependent patterns canbe complex and non-linear.

• How can you select the appropriate scale(s) for a study?

Scale:

62

37

Power Analysis,Effect Size &Sample SizeEstimation:

Power analysis quantifies the relationships between the following:

• Effect Size: The smallest detectable change.

• Variation: The amount of “noise” in the data.

• Confidence: Probability of not making a “false alarm” (Type 1 error).

• Power: Probability of not missing a significant change when it happens(Type 2 error).

• Sample Size: The amount of data you have.

Using Power Analysis, if you have estimates for any 4 of these things,you can solve for the 5th thing.

63

38

ThePowerCurve:

Sam

ple

Pow

er

Sample Size

high

low

small large

A

B

C

not effectiveor efficient

effective butnot efficient

effective andefficient

64

39

Preliminary& Sequential

Sampling:

• Where do you get estimates for power analysis parameters?

• Preliminary Sampling (i.e., pilot projects) occurs before the formalstart of a study. This stage allows for testing of methods, evaluatingeffort, and for collecting power analysis parameters to pin downthe strength of the proposed study.

• Sequential Sampling (i.e., ongoing monitoring) occurs when data areanalyzed during the data collection phase of a project. Sequentialsampling, analysis, and evaluation makes the manager more familiarwith the study and its data and promotes Adaptive Management.

Data collected from preliminary & sequential sampling can often beused in the overall study. People who skip this step often end up wasting money, not saving it.

65

44

Money isAlways

Tight:

• Study design should be carefully considered beforeany investment in monitoring is made.

• No amount of statistical hocus pocus will turn data collected from abad design into useful information.

• Not all designs are complex and not all topics covered today willalways be relevant, but some always will be (e.g., scale, disturbingvariables - bias).

• Most study design problems groups face are common acrossthe country. If you have a problem chances are someone has alreadythought about it.

66

MonitoringSeveral major watershed monitoring projects have reported little or no improvement in water quality after extensive implementation of best management practices (BMPs) in the watershed:

• Uncooperative weather• Improper selection of BMPs• Mistakes in understanding of pollution sources • Poor experimental design• Lag time

12

67

VT NMP Project 1993 - 2001Evaluate effectiveness of livestock exclusion, streambank

protection, and riparian restoration in reducing runoff of nutrients, sediment, and bacteria from agricultural land to surface waters

68

Paired watershed designContinuous dischargeFlow-proportional automated

composite sampling (weekly)Total Phosphorus (TP)Total Kjeldahl Nitrogen (TKN)Total Suspended Solids (TSS)

Bi-weekly grab samplingIndicator bacteria Temp., conductivity, D.O.

Annual biomonitoringMacroinvertebratesHabitatFish

Annual land use/management

69

[TP] -15%[TKN] -12%[TSS] -34%E. coli -29%Temperature -6%TP load -49% -800

kg/yrTKN load -38% -2200

kg/yrTSS load -28% -115,000

kg/yr

RESULTS

Macroinvertebrate IBI improved to meet biocriteriaNo significant change in fish community

70

OR Upper Grande Ronde NMP Project 1995 - 2003

Improve salmonid community through restoration of habitat and stream temperature regime

Document effectiveness of channel restoration on water temperature and salmonid community

71

½ mile

Reintroduced W et Meadow Channel 1997

Diversion 1997

N

McIntyre Road

McCoy Creek Channelized 1968

Meadow Creek Temperature Monitoring Site

Direction of stream flow

New Bridge and Culvert

2001

Reconstructed W et Meadow Channel 2002

MCCM

MCCR

MCCLR

MEADL

MCCL1&2

Before/after channel restoration, with controlContinuous air & water temperature, periodic habitat

assessment, snorkel surveys for fish monitoring

72

Is a Model Necessary? It depends what you want to know…

• What are the loads associated with individual sources?

• Where and when does impairment occur? • Is a particular source or multiple sources

generally causing the problem?• Will management actions result in meeting

water quality standards?• Which combination of management actions will

most effectively meet load targets?• Will future conditions make impairments

worse?• How can future growth be managed to

minimize adverse impacts?Models are used in many areas…TMDLs, stormwater evaluation and design, permitting, hazardous waste remediation,

dredging, coastal planning, watershed management and planning, air studies

Probably Not

Probably

13

73

To model, or not to model . . .

•As these things increase:– Number of pollutants– Complexity of loads/stressors– Uncertainty regarding existing information– Expense involved in addressing problems

• The need for more sophisticated modeling also increases