20141127 Ver 1_Notes136_MDC_20141107

Employment and program opportunities are offered to all people regardless of race, color, national origin, sex, age, or disability.
North Carolina State University, North Carolina A&T State University, U.S. Department of Agriculture, and local governments cooperating.
Printed on Recycled Paper
NWQEP NOTESNWQEP NOTESNWQEP NOTESNWQEP NOTESNWQEP NOTES
The NCSU Water Quality Group Newsletter
Number 136 December 2011 ISSN 1062-9149
PROJECT SPOTLIGHT
Minimum Detectable Change (MDC)
Dr. Jean Spooner, Department Biological and Agricultural Engineering
Department, North Carolina State University, Raleigh, NC
Definition and Overview
The Minimum Detectable Change (MDC) is the minimum change
in a pollutant concentration (or load) over a given period of time
required to be considered statistically significant.
The calculation of MDC has several practical uses. Data col-
lected in the first several years of a project or from a similar project
can be used to determine how much change must be measured in
the water resource to be considered statistically significant and not
an artifact of system variability. Calculation of MDC provides feed-
back to the project managers as to whether the proposed land
treatment and water quality monitoring designs are sufficient to
accomplish and detect the expected changes in water quality over
a pre-specified length of time. These calculations facilitate realistic
expectations when evaluating watershed studies. Calculation of the
magnitude of the water quality change required can serve as a
useful tool to evaluate water quality monitoring designs for their
effectiveness in detecting changes in water quality. Closely related,
these calculations can also be used to design effective water qual-
ity monitoring networks (Spooner et al., 1987; 1988).
Bernstein and Zalinski (1983) make a valid distinction between
the magnitude of the ‘statistically’ and ‘biologically’ significant
changes. The size of a statistically significant detectable change
depends on the number of samples. For a fixed sample variability,
a large number of samples results in a large number of degrees of
freedom in the statistical trend test, and therefore, a relatively small
value for the MDC. However, a small statistically significant dif-
ference may have no biological or practical significance. In contrast,
with small sample sizes, statistically significant detectable changes
may be much larger than biologically significant changes. A sys-
NWQEP NOTES is available on the World Wide Web at
http://www.ncsu.edu/waterquality/issues/Default.htm.
To subscribe or unsubscribe to the NWQEP NOTES
listserv: Send e-mail to: [e-mail] (leave sub-
ject field blank). In the body of the e-mail, type
subscribe nwqep_notes or unsubscribe
nwqep_notes in the body of the message.
Note: NWQEP NOTES is only provided elec-
tronically effective June 2007.
IN THIS ISSUE
Minimum Detectable Change (MDC)........1
Editor’s Note ..............................................2
Conference Report ..................................12
NRCS Revises Nutrient Standard ..........12
NRCS Initiative on Drainage Water
Management ...........................................13
Ag NPS Regulations ...............................13
EPA Healthy Watersheds Initiative ..........13
Biological Assessment Primer ..............13
USGS Coop Water Program ...................14
USGS Finds No Consistent Nitrate
Declines in Mississippi River Basin ......14
USGS Decision Support System for
Regional Nutrient Assessment ..............14
Gulf Coast Ecosystem Restoration
Task Force ...............................................15
Meetings ..................................................15
2
NWQEP NOTES — December 2011
MDC is a quantity that is calculated using the pre-planned
statistical trend tests on the measured observations, typically
in the pre-BMP project phase. MDC is used as a guide to
calculate the minimum amount of change expected to be de-
tected given the sample variability, number of samples,
monitoring design and statistical trend tests and significance
level.
General Considerations
The following assumptions are made in the calculation of
MDC.
Historical sample measurements are representative of the
temporal and spatial variation of the past and future
conditions.
Variability due to sampling, transport or laboratory error
is negligible compared to variability over time.
Typically, the pollutant concentrations or load values ex-
hibit a log-normal distribution. When this is the case, the MDC
is expressed as a percent change relative to the initial annual
geometric mean concentration. Given a particular monitor-
ing scheme, the water quality observations and their variability
can be used to calculate the MDC required in the geometric
mean pollutant concentration over time.
When the water quality values are log-normal, calcula-
tions for the MDC values are performed on the base 10
logarithmic scale. Analyses on the logarithmic scale have sev-
eral beneficial features:
The log normal distribution generally fits the distribution
of water quality data. One feature of a log normal
distribution is skewed data on the original scale (e.g.,
many lower values with a few higher values).
The logarithmic transformation on the water quality
variables is usually required for the distributional
assumptions of parametric trend analyses to be met.
The results become dimensionless and are independent
of the units of measurements.
MDC can be expressed as a percentage, rather than an
absolute difference, because the calculations are
performed on the logarithmic scale.
EDITOR’S NOTE
This issue of NWQEP NOTES focuses upon Mini-
mum detectable change (MDC). Minimum detectable
change analysis can answer questions like: “How much
change must be measured in a water resource to be con-
sidered statistically significant?” or “Is the proposed
monitoring plan sufficient to detect the change in con-
centration expected from BMP implementation?”
MDC analyses are useful for persons involved in wa-
tershed nonpoint source monitoring and evaluation projects
such as those in the National Nonpoint Source Monitor-
ing Program (NNPSMP) and the Mississippi River Basin
Initiative, where documentation of water quality response
to the implementation of management measures is the
objective. The MDC techniques discussed below are ap-
plicable to water quality monitoring data collected under
a range of monitoring designs including single fixed sta-
tions and paired-watersheds. MDC analysis can be
performed on datasets that include either pre- and post-
implementation data or just the typically limited
pre-implementation data that watershed projects have in
the planning phase. Better datasets, however, provide more
useful and accurate estimates of MDC.
As always, please feel free to contact me with your
ideas, suggestions, and possible contributions to this news-
letter
Jean Spooner
Editor, NWQEP NOTES
NCSU Water Quality Group
Campus Box 7637, NCSU
Raleigh, NC 27695-7637
Email: [e-mail]
MDC analysis must be consistent with and
based on the planned statistical approach to
analyzing project data.
tem may have exhibited a biologically significant change that
cannot be statistically detected because sample sizes are too
small.
MDC is an extension of the Least Significant Difference
(LSD) concept (Snedecor and Cochran, 1967). The MDC for
a system can be estimated from data collected within the same
system or similar systems. A system is defined by the water-
shed size, water resource, monitoring design, pollutants
measured, sampling frequency, length of monitoring time,
hydrology and meteorology.
3
NWQEP NOTES — December 2011
Sampling frequency determination is very closely related
to MDC calculations. Sample size determination is usually per-
formed by fixing a significance level, power of the test, the
minimum change one wants to detect, the duration of moni-
toring, and the type of statistical test. MDC is calculated
similarly except the sample size (i.e., number of samples) is
fixed and the power is set to 50% to solve for the minimum
detectable change. MDC is the amount of change you can
detect given the sample variability. Therefore, many of the
same formulas that are used for confidence limit and sample
size determination are similar to those used to calculate MDC.
Factors Affecting the Magnitude of the MDC
The MDC is a function of pollutant variability, sampling
frequency, length of monitoring time, explanatory variables or
covariates (e.g., season, meteorological, and hydrologic vari-
ables) used in the analyses which ‘adjust’ or ‘explain’ some of
the variability in the measured data, magnitude and structure
of the autocorrelation, and statistical techniques and signifi-
cance level used to analyze the data.
Spatial and Temporal Variability
The basic concept in the calculation of MDC is simple:
variability in water quality measurements is examined to esti-
mate the magnitude of changes in water quality needed to detect
significant differences over time. Hydrologic systems are highly
variable, often resulting in large values for MDC. Variations
occur in both spatial and temporal dimensions. Variations in
water quality measurements are due to several factors includ-
ing:
A change in land treatment resulting in decreased
concentrations and/or loadings to receiving waters
(determining the amount of water quality change is usually
a key objective of a watershed project)
Sampling and analytical error
Monitoring design (e.g., sampling frequency, sampling
location, variables measured)
Changes in meteorological and hydrologic conditions
Seasonality
Changes in input to and exports from the system. For
example, changes in upstream concentrations can affect
the downstream water quality.
MDC is proportionally related to the standard deviation of
the sample estimate of trend (e.g., standard deviation of the
sample estimate of slope for a linear trend or standard devia-
tion of samples in the pre-BMP time period for a step trend).
This standard deviation is a function of the variability in Y that
is not explained by the statistical trend model (i.e., error vari-
ance). As such, any known sources of variation that can be
added to the statistical trend model to minimize the error vari-
ance will also serve to reduce the MDC and increase the ability
to detect a real change in water quality due to land treatment.
For example, adjusting for changes in explanatory variables
such as streamflow or changes in land use (other than the
BMPs) would reduce both the standard error and the MDC.
It should be noted that sample variability may be affected
by sampling frequency. For frequent sampling directed at in-
cluding storm events, variability is usually higher than for
fixed-interval sampling directed at monitoring ambient condi-
tions. In addition, the nature of collection and data aggregation
will directly affect the variability and the autocorrelation. Com-
posite or aggregated samples are generally less variable than
single grab samples and exhibit a lower degree of autocorrelation
as compared to non-aggregated data.
Sampling Frequency and Record Length
The MDC calculation is the change required for a speci-
fied sample frequency and duration. MDC decreases with an
increase in the number of samples and/or duration of sam-
pling.
Increasing sampling frequency and/or record length (e.g.,
increasing the number of years for monitoring) results in an
increase in the number of samples (N), and therefore increases
the degrees of freedom in the statistical trend tests and results
in a smaller MDC value. Increasing the number of samples
results in a decrease in MDC (on the logarithmic scale) ap-
proximately proportional to the increase in the square root of
N. However, increasing N by increasing the sample frequency
may not decrease the MDC by this total proportion due to the
effects of temporal autocorrelation.
Increasing record length has several advantages over in-
creasing sampling frequency. Increasing record length serves
to add degrees of freedom to the statistical trend models. In
addition, increasing the number of years adds extra verifica-
tion that the observed changes are real and not a result of an
unknown or unmeasured variable that also exhibits large year-
to-year variations. Increasing record length also serves to
increase the time base from which extrapolations may be made.
Seasonal, Meteorological and Hydrologic Variability
The standard error of a trend estimate can effectively be
reduced by accounting for seasonality and meteorological and
hydrologic variables in the trend tests. Because these variables
or covariates can help reduce the amount of variability that
Sampling frequency and MDC are closely related
parameters. The planned sampling frequency and
duration strongly influence the MDC, and the MDC
largely dictates the sampling frequency necessary to
measure such change within a specified time period.
4
NWQEP NOTES — December 2011
cannot be ‘explained’ they are commonly called ‘explanatory
variables.’ For example, Hirsch and Gilroy (1985) found that
a model that removes variability in sulfate loading rates due to
precipitation and varying seasonal mean values can reduce the
step trend standard deviation by 32%, and therefore, the mag-
nitude of change needed for statistically detectable change
would be reduced by 32%.
Incorporation of appropriate explanatory variables increases
the probability of detecting significant changes and serves to
produce statistical trend analysis results that better represent
true changes due to BMP implementation rather than changes
due to hydrologic and meteorological variability. Commonly
used explanatory variables for hydrologic and meteorological
variability include streamflow and total precipitation.
Adjustment for seasonal, meteorological and hydrologic
variability is also important to remove bias in trend estimates
due to changes in these factors between sampling times and
years. Interpretations regarding the direction, magnitude and
significance in water quality changes may be incorrect if hy-
drologic and/or meteorological variability is not accounted for
in the statistical trend models.
If significant variation exists between the seasonal means
and/or variances and is not considered in the statistical trend
models, then the assumptions of identical and independent dis-
tribution of the residuals (from the statistical model) will be
violated and the results for the statistical trend analyses (both
parametric and nonparametric) will not be valid. Non-identical
distributions can occur when the seasonal means vary from
the overall mean and/or the variances within seasons are dif-
ferent for each season. Non-independence can occur because
seasons have cyclic patterns, e.g., winters are similar to win-
ters, summers to summers, etc.
Autocorrelation
Temporal autocorrelation exists if an observation is related
or correlated with past observations (not independent).
Autocorrelation in water quality observations taken less fre-
quently than daily is usually positive and follows an
autoregressive structure of order 1, AR(1). More complicated
autocorrelation models (AutoRegressive Integrated Moving-
Average or ARIMA models with more lag terms and moving
average terms) are usually needed for daily or more frequent
sampling designs. Positive autocorrelation usually results in a
reduction of information (e.g., less degrees of freedom than
the actual number of samples) in a data series and affects
statistical trend analyses and their interpretations. Each addi-
tional sample adds information, but not a full degree of freedom
if it’s not independent of the previous sample.
If significant autocorrelation exists and is not considered
in the statistical trend models, then the assumption of inde-
pendence of the residuals will have been violated. The result is
incorrect estimates of the standard deviations on the statistical
parameters (e.g., mean, slope, step trend estimate) which in
turn results in incorrect interpretations regarding the statisti-
cal significance of these statistical parameters. Autocorrelation
must be incorporated into the statistical trend models to obtain
an accurate estimate of MDC (e.g., using time series analy-
ses). Autocorrelation can also be reduce by data aggregation
(e.g., weekly, monthly), but this will decrease the degrees of
freedom.
Statistical Trend Tests
MDC is influenced by the statistical trend test selected.
For the MDC estimate to be valid, the required assumptions
must be met. Independent and identically distributed residuals
are requirements for both parametric and nonparametric trend
tests. Normality is an additional assumption placed on most
parametric trend tests. However, parametric tests for step or
linear trends are fairly robust and therefore do not require ‘ide-
ally’ normal data to provide valid results.
The standard error on the trend estimate, and therefore,
the MDC estimate will be minimized if the form of the ex-
pected water quality trend is correctly represented in the
statistical trend model. For example, if BMP implementation
occurs in a short period of time after a pre-BMP period, a
trend model using a step change would be appropriate. If the
BMPs are implemented over a longer period of time, a linear
or ramp trend would be more appropriate.
A step change can be examined by the use of tests such as
the parametric Student’s t-test or the nonparametric Wilcoxon
rank sum test. The two-sample Student’s t-test and the non-
parametric Wilcoxon rank sum tests for step change are popular
step change tests used in water quality trend analyses because
they are easy to use. Analysis of Covariance (ANCOVA) can
test for step changes after adjusting for variability in explana-
tory variables or covariates (e.g., streamflow). When a sudden
system alternation, such as BMP implementation occurs, the
BMPs can be called an ‘intervention.’ In statistical terms, in-
tervention analysis can be used to extend the two-sample
Student’s t-test to include adjustments for autocorrelation.
The most popular types of statistical models for linear
change include the parametric linear regression and the non-
parametric Kendall’s tau (with the Sen’s Slope Estimator).
Autocorrelation is most easily accounted for by the use of
linear regression models with time series errors. When using a
statistical software package that can adjust for autocorrelation
(e.g., PROC AUTOREG in SAS (SAS, 1999)), it requires no
extra effort to correctly incorporate the needed time series as
MDC is influenced by the statistical trend test selected.
The MDC will be minimized if the correct statistical
trend model (e.g., step vs. linear or ramp) is selected.
5
NWQEP NOTES — December 2011
well as explanatory variables. See USEPA (2011) for an over-
view of other statistical software packages that may be also
be used.
Steps to Calculate the MDC
The calculation MDC or the water quality concentration
change required to detect significant trends requires several
steps. The procedure varies slightly based upon:
Pattern of the expected change and therefore appropriate
statistical model (e.g., step, linear, or ramp trend).
Whether the data used are in the original scale (e.g., mg/l
or kg) or log transformed.
Incorporation of time series to adjust for autocorrelation.
Addition of explanatory variables such as streamflow or
season.
The following steps and examples are adopted from Spooner
et al. (1987 and 1988):
Step 1. Define the Monitoring Goal and Choose the Ap-
propriate Statistical Trend Test Approach. One goal may be
to detect a statistically significant linear trend in the annual
mean (geometric mean if using log transformed data) pollut-
ant concentrations that may be related to land treatment changes.
A linear regression model using log-transformed data would
then be appropriate. An alternative goal to detect a statistically
significant change in the post-BMP period as compared to a
pre-BMP period would require a step change statistical test
such as the t-test or ANCOVA.
For linear trends, an appropriate regression trend model
would be a linear trend either without:
Yt = β0 + β1DATE + eτ
or, with explanatory variables as appropriate:
Yt = β0 + β1DATE + ΣβiXi + eτ
Where: Yt = Water quality variable value at time t. If Y is log
normal, then Yt is the log-transformed water quality
variable value.
Xi = Explanatory variable, i=2,3,... (X2, X3, etc. could
also be log transformed; the DATE variable is con-
sidered Xi)
β0 = Intercept
β1 = Slope or linear trend on DATE
βi = Regression coefficients for explanatory variables
et = Error term (this is denoted at Vt if the error series
has an autocorrelated structure, see Step 4 and Ex-
ample 1)
Note that even though no (zero) trend is expected if this
test uses only the pre-BMP data, it is appropriate to include the
trend (DATE) term in the statistical model when this is the
planned statistical model.
For a step trend, the DATE can have the values of 0 for
pre-BMP or 1 for post-BMP data. When planning or evaluat-
ing a monitoring design, there may not yet be any post-BMP
data and only pre-BMP data would then be used in the MDC
calculations.
Note: the paired watershed study and the above/below-
before/after watershed designs are analyzed using an Analysis
of Covariance (ANCOV) where ‘Date’ is 0 or 1 and the ex-
planatory variable is either the control watershed values
(concentrations/loads) or the upstream values paired with the
treatment or downstream values, respectively.
Step 2. Perform Exploratory Data Analyses. Preliminary
data inspections are performed to determine if the residuals
are distributed with a normal distribution and constant vari-
ance. Normal distribution is required in the parametric analyses;
constant variance is required in both parametric and nonpara-
metric analyses. The water quality monitoring data are usually
not normal, however, and often do not exhibit constant vari-
ance over the data range.
The water quality data sets are examined using univariate
procedures such as those available with the SAS procedure
PROC UNIVARIATE or within JMP (SAS Institute 2010, 2008)
to verify distributional assumptions required for statistical pro-
cedures. Specific attention is given to the statistics on normality,
skewness and kurtosis. Both the original and logarithmic trans-
formed values are tested.
Step 3. Perform Data Transformations. Water quality data
typically follow log-normal distributions and the base 10 loga-
rithmic transformation is typically used to minimize the violation
of the assumptions of normality and constant variance. In this
case, the MDC calculations use the log-transformed data until
the last step of expressing the percent change. The natural log
transformation may alternatively be used.
The logarithmic base 10 transformation applies to the de-
pendent water quality variables used in trend detection (i.e.,
suspended sediment, TP, ortho phosphorus and fecal coliform).
Technically, explanatory variables in statistical trend models
do not have any distributional requirements because it is only
the distribution of the residuals that is crucial. However, if
Exploratory data analysis (US EPA, 2005) is an
important step in determining whether available data
meet the assumptions (e.g., normality, constant
variance) of planned statistical tests.
6
NWQEP NOTES — December 2011
they do exhibit log normal distributions, explanatory variables
are also log transformed which usually helps with the distri-
bution requirements of the residuals. Typical explanatory
variables that are log transformed include upstream concen-
trations and stream flow.
Step 4. Test for Autocorrelation. Tests are performed on
the water quality time series to determine if there is
autocorrelation. An autoregressive, lag 1 (AR(1)) error struc-
ture (i.e., correlation between two sequential observations) in
the water quality trend data is common. The tests usually as-
sume samples are collected with equal time intervals. The
regression trend models used are the same as those planned
for the future trend analyses (See Step 1). The data should be
ordered by collection date.
The Durbin Watson (DW) test for autocorrelation can be
performed on the residuals from the linear regression models
to determine if the concentration measurements are related to
previous measurements. Several software packages will cal-
culate the DW tests. For example, this test can be performed
with the SAS procedure PROC REG or PROC AUTOREG
(SAS Institute, 1999), or within the least squares regression
analysis with JMP (SAS Institute, 2008). The Durbin Watson
test assumes the residuals exhibit an AR(1) autocorrelation
structure. Alternatively, the significance of the first order
autocorrelation coefficient is tested in SAS using a time series
statistical procedure such as PROC AUTOREG or time series
analyses within JMP. It should be noted that PROC AUTOREG
allows for missing Y-values, but equally-spaced date entries
should all be included in the data set.
Alternatively, the assumption of independent residuals can
be tested by passing the residuals from these regression trend
models to the SAS procedure PROC ARIMA (SAS Institute,
1999) or time series analysis within JMP (SAS Institute, 2008).
The autocorrelation structure is examined to determine if the
independence assumption is valid and, if not, to determine the
appropriate autocorrelation structure for the simple trend mod-
els. The chi-square test of white noise supplied by PROC
ARIMA is also used to test whether the residuals are indepen-
dent.
Step 5. Calculate the Estimated Standard Error. The vari-
ability observed in either historic or pre-BMP water quality
monitoring data is used to estimate the MDC. Any available
post-BMP data can also be included in this step. The estimated
standard error is obtained by running the same statistical model
that will be used to detect a trend once BMPs have been in-
stalled (same trend models identified in Step 1).
For a linear trend, an estimate of the standard deviation
on the slope over time is obtained by using the output from
statistical regression analysis with a linear trend, time series
errors (if applicable), and appropriate explanatory variables. If
additional duration of monitoring is planned, and therefore a
longer data set will be the obtained in the future, the standard
deviation of the future trend slope can be estimated from the
existing data by:
Where:
sb = estimate for the standard deviation of the trend for the
total planned duration of monitoring
bs' = standard deviation of the slope for the existing data
n = number of samples in the existing data
C = correction factor equal to the proportional increase in
planned samples. For example, if 4 years of existing data
are available and 8 years of total monitoring is planned,
C=2 (i.e., 8/4). This factor will reduce the standard error
on the slope and, therefore, the amount of change per
year required for statistical significance.
A large sample approximation for the adjustment factor is:
For a step trend, it is necessary to have an estimate of the
standard deviation of the difference between the mean
values of the pre- vs. post- data ( ). In practice,
an estimate is obtained by using the following formula:
Where:= Estimated standard error of the
difference between the mean values of the pre- and the
post- BMP periods.
MSE = sp
2 = Estimate of the pooled Mean Square Error
(MSE) or, equivalently, variance (sp
2) within each period.
The MSE estimate is obtained from the output of a
statistical analysis using a t-test or ANCOVA with
appropriate time series and explanatory variables.
The variance (square of the standard deviation) of pre-
BMP data can be used to estimate MSE or sp
2 for both pre and
post periods if post-BMP data are not available and there are
no explanatory variables or autocorrelation (see Example 2).
For log normal data calculate this value on the log transformed
data.
Missing values are allowed. It is not important at this step
if a trend is present or not because this step obtains the esti-
mate on the standard deviation of the trend statistic.
For both linear and step trends, if autocorrelation is present
a time series statistical procedure such as SAS’s PROC
AUTOREG that uses Generalized Least Squares (GLS) with
Yule Walker methods should be employed because it takes into
2)n(C
2)(nss
*bb −
−='
C
1
bbss'=
postpre
postpre n
MSE
n
MSEs)XX(+=−
)xx(s postpre−
)xx(s postpre−
- -
7
NWQEP NOTES — December 2011
account the autocorrelation structure of the residuals to ob-
tain valid standard deviations (Brocklebank and Dickey, 1986).
The standard error on the trend estimate for simple trend models
(e.g., step, linear, or ramp trends) with AR(1) error terms is
larger than that (incorrectly) calculated by Ordinary Least
Squares (OLS). Matalas (1967) cited theoretical adjustments
that can be used. The true standard deviation has the follow-
ing large sample approximation:
Where: sb = true standard deviation of the trend (slope or
difference between 2 means) estimate (e.g., calculated
using GLS)
bs'= incorrect variance of the trend estimate calculated
without regard to autocorrelation using OLS (e.g., using
a statistical linear regression procedure that does not take
into account autocorrelation)
ρ = autocorrelation coefficient for autoregressive lag 1,
AR(1)
Step 6. Calculate the MDC. MDC is essentially one-half
of the confidence interval for the slope of a linear regression
model or for the difference between the mean values of the
pre- and post BMP periods.
For a linear trend, the MDC is calculated by multiplying
the estimated standard deviation of the slope by the t-sta-
tistic and the total monitoring timeframe:
MDC = (N) * t(n*N-2)df * 365 * sb1
Where: t(n*N-2)df = One-sided Student’s t-statistic (α = .05)
N= Number of monitoring years
n= Number of samples per year
df= degrees of freedom
365= Correction factor to put the slope on an
annual basis when DATE is entered as a Date (day)
variable, e.g., the slope is in units per day. If DATE
values were 1-12 for months and the slope was
expressed ‘per month’ then this value would be “12.”
sb1 = Standard deviation on the slope estimated
for the total expected monitoring duration (from Step
5)
MDC= the MDC on either the original data scale
or the log scale if the data were log-transformed
For a step trend, the MDC is one-half of the confidence
interval for detecting a change between the mean values in the
pre- vs. post- BMP periods.
In practice, an estimate is obtained by using the following
equivalent formula:
Where: = one-sided Student’s t-value with
(npre + npost -2) degrees of freedom.
npre + npost = the combined number of samples in the
pre- and post-BMP periods
= estimated standard error of the
difference between the mean values in the pre- and the
post- BMP periods.
MSE = sp
2 = Estimate of the pooled Mean Square Error
(MSE) or, equivalently, variance (sp
2) within each period.
The MSE estimate is obtained from the output of a
statistical analysis using a t -test or ANCOVA with
appropriate time series and explanatory variables. If post-
BMP data are not available, no autocorrelation is present,
and no explanatory variables are appropriate, MSE or sp
2
can be estimated by the variance (square of the standard
deviation) of pre-BMP data.
The pre- and post-periods can have different sample sizes
but should have the same sampling frequency (e.g., weekly).
The following considerations should be noted:
The choice of one- or two-sided t-statistic is based upon
the question being asked. Typically, the question is whether
there has been a statistically significant decrease in pollutant
loads or concentrations and a one-sided t-statistic would
be appropriate. A two-sided t-statistic would be appropriate
if the question being evaluated is whether a change in
pollutant loads or concentrations has occurred. The value
of the t-statistic for a two-sided test is larger, resulting in
a larger MDC value.
At this stage in the analysis, the MDC is either in the original
data scale (e.g., mg/L) if non-transformed data are used,
or, more typically in the log scale if log-transformed data
are use.
Step 7. Express MDC as a Percent Decrease. If the data
analyzed were not transformed, MDC as a percent change
(MDC%) is simply the MDC from Step 6 divided by the aver-
age value in the pre-BMP period expressed as a percentage
(e.g., MDC% = 100*(MDC/mean of Pre-BMP data).
When calculating MDC as a percent change it is
important to note whether the data analyzed were log
transformed because the formula is different from that
used for data that were not log transformed.
p 1
p 1s s bb −
+='
postpre
postpre
t n
MSE
n
MSE*MDC )2nn(+=−+
)2nn(−+postpre
t
)XX()2nn(sMDC postprepostpre*t −−−=
)xx(s
postpre−
8
NWQEP NOTES — December 2011
If the data were log-transformed, a simple calculation
can be performed to express the MDC as a percent decrease
in the geometric mean concentration relative to the initial geo-
metric mean concentration or load. The calculation is:
MDC% = (1 - 10-MDC) * 100
Where: MDC is on the log scale and MDC% is a
percentage.
For log-transformed data MDC is the difference required
on the logarithmic scale to detect a significant decreasing trend
(calculated in Steps 5 and 6 using log transformed data).
MDC% and MDC are positive numbers if mean concentra-
tions decrease over time. For example, for MDC= 0.1 (10-0.1 =
0.79), the MDC% or percent reduction in water quality re-
quired for statistical significance = 21%; for MDC = 0.2 (10-0.2 = 0.63),
MDC% = 37%. In the cases where detection of a positive
trend is desired (e.g., Secchi depth measurements), the per-
cent decrease would be negative and the input for MDC must
be forced to be negative.
It should be noted that if the natural logarithmic transfor-
mation had been used, then:
MDC% = (1 - exp-MDC) * 100
MDC Calculation Examples
Example 1: A linear trend with autocorrelation and
covariates or explanatory variables; Y values log transformed.
The basic statistical trend model used in this example is linear
regression with time series errors, techniques documented by
Brocklebank and Dickey (1986). Typically, Autoregressive Lag
1 or AR(1) is appropriate and a DATE explanatory variable is
included in the model. The DATE variable is used to estimate
the magnitude of a linear trend and to estimate the variation
not accounted for by the linear trend term observed in the
water quality measurements. The estimate of variation on the
“slope” of DATE is then used to calculate an estimate of Mini-
mum Detectable Change (MDC). The significance of the linear
trend, its magnitude, or its direction is not important in the
calculation of MDC. The important statistical parameter is the
standard deviation on the slope estimate of the linear trend.
The SAS procedure, PROC AUTOREG (SAS Institute,
1999) can be used in the analyses. The linear regression model
estimated at each monitoring location is:
Yt = β0 + β1DATE + Vτ
or, with explanatory variables:
Yt = β0 + β1DATE + ΣβiXi + Vτ
Where:Yt = Log-transformed water quality variable value at
time t,
Vt = Error term assumed to be generated by an
autoregressive process of order 1, AR(1)
β0 = Intercept
β1 = Slope or linear trend on DATE
β1 = Unique regression coefficients for each explana-
tory variables
Xi = Explanatory variable, i=2,3,..,
The standard deviations on the slope over time from linear
regression models are used to calculate the MDCs. A signifi-
cance level of α=.05 and a Type II error of b=0.5 are assumed.
The standard deviation on the slope is a function of the mean
square error (MSE or s2) estimated by the Yule Walker Method
and Generalized Least Squares, degree of autocorrelation, and
the degrees of freedom (d.f.). The d.f. is a function of the
number of monitoring years and sample frequency. If contin-
ued sampling is planned, the estimate of the standard deviation
of the trend slope is adjusted by a correction factor given in
Step 5.
MDC is calculated by:
MDC = (N) * t(n*N-2)df * 365 * sb1
Where: t(n*N-2)df = One-sided Student’s t-statistic (α = .05)
N = Number of monitoring years
n = Number of samples per year
365 = Correction factor to put the slope on an
annual basis because DATE is assumed to be entered
as a Date variable (i.e., the slope is in units per day).
If DATE values were entered as 1-12 for months
causing the slope to be expressed as ‘per month’ then
this value would be “12.”
sb1 = Standard deviation on the slope
MDC = MDC on the log scale in this case
The calculations are illustrated below with the following
assumptions:
N = 5 years existing (10 years planned)
n = 52 weekly samples per year
DATE was entered into the computer program as a DATE,
so the slope is expressed in units per day
t(n*N-2)df = t258 = 1.6513 (one-sided)
sb1 = 0.0000229 (This is the standard deviation on the slope
for the trend, which is log scale for this example because
log transformed data are assumed. It is very important to
carry several significant digits because the number might
be small.)
9
NWQEP NOTES — December 2011
The MDC for the existing 5 years of data can be calcu-
lated. The calculations for MDC and then MDC% for this
example using Y values that are log transformed are as fol-
lows:
MDC = (N) * t(n*N-2)df * 365 * sb1
MDC = 5 * 1.6513 * 365 * 0.0000229
MDC = 0.06901 (units on log scale)
MDC% = (1 - 10-MDC) * 100 (percentage on geometric
mean)
MDC% = (1 - 10-0.06901) * 100
MDC% = 15% (percentage on geometric mean) or an
average of 3% change per year
Note: If a 2-sided t-statistic value was used then t=1.969
and MDC (log scale) is 0.0823 and MDC% is 17%.
The MDC estimate if the sampling duration will be doubled
to a total of 10 years:
= 0.0000229 *0.70574
= 0.00001616
MDC (10 years) = 10 * 1.6513 * 365 * 0.00001616
= 0.0974 (units on log scale)
= 20% over 10 years (or an average of 2% change
per year)
The addition of appropriate explanatory variables and sam-
pling frequency can decrease the magnitude of the calculated
MDC. For example, Spooner et al. (1987) demonstrated that
adding salinity as a covariate in the Tillamook Bay, Oregon
watershed study decreased the MDC% for fecal coliform over
an 11-year period of time (with biweekly samples) from 42%
to 36%. For the same study, the MDC% for fecal coliform
decreased from 55% to 42% when comparing monthly to
biweekly sampling over an 11-year study. Spooner et al. (1987
and 1988) also demonstrated that variability and therefore MDC
is also affected by the pollutant measured, the size of the wa-
tershed, and appropriate selection of explanatory variables.
If a seasonal pattern is present, explanatory variables can
be added to help adjust for this source of variability. Sin/
Cosine terms can be used for such adjustment (US EPA, 2011).
Alternatively, a seasonal pattern can be included by adjusting
for monthly means. In this case, monthly variables (Mi) can
be added as explanatory variables where:
Mi=1 for Month i, else 0
i=1 to 11 for January to November, respectively (De-
cember values have “0” entered for all Mi variables.
Example 2: A step trend, no autocorrelation, and no
covariates or explanatory variables; Y values on original scale
(not transformed). In this example, the plan would be to de-
tect a significant change in the average values between the
pre- and post-BMP periods. The pre- and post-periods can
have different sample sizes but should have the same sampling
frequency (e.g., weekly).
In this simplified situation, the MDC would be equivalent
to the Least Significant Difference (LSD). MDC would be
calculated as:
Where: = one-sided Student’s t-value with
(npre + npost -2) degrees of freedom.
npre + npost = the combined number of samples
in the pre- and post-BMP periods
MSE = Estimate of the pooled Mean Square Error
(MSE) or variance (sp
2) within each period. The vari-
ance (square of the standard deviation) of pre-BMP
data can be used to estimate MSE or sp
2 for both pre
and post periods if post-BMP data are not available
(the usual case when designing monitoring programs).
For log normal data calculate this value on the log
transformed data.
The calculations are illustrated below with the following
assumptions:
npre =52 samples in the pre-BMP period
npost =52 samples in the post-BMP period
Mean X = 36.9 mg/l, mean of the 52 samples in the
pre-BMP period
sp = 21.2 mg/L = standard deviation of the 52 pre-
BMP samples
MSE =sp
2 = 449.44
= t102 = 1.6599
The MDC would be:
The addition of explanatory variables can
decrease the magnitude of the MDC.
2)260(2
2)(2600000229.02)n(C
2)(nss
**
b1(5years) years) (10 b1
−
−=
−
−='
postpre
postpre
t n
MSE
n
MSE*MDC )2nn(+=−−
)2(−−postprennt
)2(−−postprennt
postpre
postpre
t n
MSE
n
MSE*MDC )2nn(+=−−
10
NWQEP NOTES — December 2011
MDC = 6.9 mg/l
Percent change required = MDC%=100*(6.9/36.9)=19%.
Use the equation described under “Step 7” above to calcu-
late percent change for log-transformed data. If the data are
autorcorrelated, use a time series model, or the approximation
given in Step 5 to adjust the standard error of the difference in
the pre- and post-BMP means.
Example 3: Paired watershed study or Above/Below-Be-
fore/After watershed study analyzed using Analysis of
Covariance (ANCOVA); Y values log transformed; no
autocorrelation. The paired watershed approach requires a
minimum of two watersheds, control and treatment, and two
periods of study, calibration and treatment (Clausen and
Spooner, 1993). The control watershed accounts for year-to-
year or seasonal climatic variations. During the calibration
period, the two watersheds are treated identically and paired
water quality data are collected (e.g., event-based, weekly).
During the treatment period, the treatment watershed is treated
with a BMP(s) while the control watershed remains under the
same management employed during the calibration period.
Under the above/below-before/after approach water quality
downstream and upstream of a BMP location is monitored for
time periods before and after BMP implementation.
Data from these two watershed designs can be analyzed
with similar ANCOVA approaches. The Y values in the equa-
tion below are taken from either the treatment watershed in a
paired-watershed study or the downstream site in an above/
below study. The values for the explanatory (X) variable are
taken from the control watershed in a paired-watershed de-
sign or from the upstream site in an above/below design. The
Yt and Xt are the pollutant concentrations or loads measured at
the same time period t. Each monitoring design has another
explanatory variable that is represented by 0 or 1 for the ‘pre-
BMP’ and ‘post- BMP’ periods, respectively.
The ANCOVA model is:
Yt = β0 + β1(Period) + B2Xt + et
Where: Yt = Water quality variable value at time t (from treat-
ment watershed or downstream site). If Y is log
normal, then Yt is the log-transformed water quality
variable value.
Period = ‘0’ for pre-BMP period and ‘1’ for post-
BMP period (alternatively, period can be treated as a
grouping variable and entered as characters).
Xt = Explanatory variable value at time t (water qual-
ity values from control watershed or upstream site).
Values are log transformed if distribution is log-nor-
mal.
β0 = Y intercept
β1, β2 = Regression coefficients
et = Error term
The SAS procedure PROC GLM (SAS Institute, 2010),
JMP (SAS Institute, 2008), or SPSS (IBM, 2011) can be used
for the analysis. Period would be identified as a ‘Class’ vari-
able in PROC GLM or ‘Character’ variable in JMP. The “Fit
Model” dialog box would be used in JMP. Users would select
the Y variable, use the “Add” option to include the X (i.e.,
control) and Period variables, and then choose ‘Run Model.’
It is important to note that because MDCs are generally
calculated prior to the treatment period, this example assumes
that the slopes for the pre- and post- periods will be similar.
The Durbin Watson statistic to check for autocorrelation can
be calculated as an option under both SPSS and either SAS
procedure. If autocorrelation is significant, PROC AUTOREG
can be used for the analysis with Period values set to numeric
‘0’ and ‘1’.
The treatment effect will be the difference in the least square
means (lsmeans) between the pre- and post-BMP periods. The
MDC is the difference that would be statistically significant
and therefore based upon the standard error of the difference
between lsmeans values. The lsmeans are the estimates of the
values of Y for each of the pre- and post- BMP period evalu-
ated at the overall average value of all the X (treatment) values
collected during the entire study period. MDC is calculated
from the standard error on the difference in lsmeans. The
standard error is given by the JMP procedure when users
choose the option for ‘detailed comparisons’.
The MDC on the log values would be:
MDC =
Where:= One-sided Student’s t-value with
(npre + npost -3) degrees of freedom (Note that the t-
statistic given in JMP is the two-sided value).
npre + npost = The combined number of samples
in the pre- and post-BMP periods
= Estimated standard error of the
difference between the least square mean values in
the Pre- and the Post- BMP periods. This is com-
puted by using the following approximation (adapted
from Snedecor and Cochran, 1967, p. 423):
MSE is found in the Analysis of Variance table from
the output of the applied statistical analysis, and n is
the number of samples within each period. The
adjustment “Factor” is 1 or greater and increases
52
449
52
4496599.1MDC +=
)lsmeanlsmean(*)3nn(postprepostprest−−+
)3nn(−+postpret
)lsmeanlsmean(postpre
s −
Factor*n
2*MSE
11
NWQEP NOTES — December 2011
when the difference between the mean of the X
(control watershed or upstream) data in the pre- period
compared to the post-BMP period increase. It is
assumed to be “1” for MDC calculations. This
“Factor” adjustment makes clear the importance of
collecting samples in the pre- and post- periods that
have similar ranges and variability in hydrological
conditions.
To express MDC as a percentage change required in geo-
metric mean value:
MDC% = (1 - 10-MDC) * 100, where MDC is on the log
scale
MDC Summar y
The Minimum Detectable Change (MDC) is the minimum
change in a pollutant concentration (or load) over a given pe-
riod of time required to be considered statistically significant.
MDC calculations can be very helpful in the design of cost-
effective monitoring programs, as well as increasing awareness
regarding the potential a watershed project has for achieving
measurable results. These calculations also illustrate the value
of adjusting for changes in hydrologic and meteorological vari-
ables. Not only is the ability to detect real changes increased,
but valid conclusions regarding the magnitude and direction
of measured change(s) in a water quality variable can be made.
Calculation of MDC can also be used to illustrate the impor-
tance of relatively long monitoring time frames. In addition,
comparison of the actual changes in water quality to the MDC
values can be used to document BMP effectiveness on a
subwatershed basis.
The magnitude of MDC is often larger than expected by
watershed projects and funding agencies, leading to misun-
derstanding regarding the needed level of BMP implementation,
intensity of monitoring, and duration of monitoring. The mag-
nitude of MDC can be reduced by:
Accounting for changes in discharge, precipitation, ground
water table depth or other applicable hydrologic/meteoro-
logical explanatory variable(s).
Accounting for changes in incoming pollutant
concentrations upstream of the BMP implementation sub-
watershed (i.e., upstream concentrations).
Increasing the length of the monitoring period.
Increasing the sample frequency.
Applying the statistical trend technique that best matches
the implementation of BMPs and other land use changes.
References
Bernstein, B.B. and J. Zalinski. 1983. An optimum sampling
design and power tests for environmental biologists. Jour-
nal of Environmental Management 16(1):35-43.
Brocklebank, J.C. and D. Dickey. 1986. SAS system for fore-
casting time series, 1986 Edition. SAS Institute Inc., Cary,
North Carolina. 240 p.
Clausen, J. and J. Spooner. 1993. Paired watershed study de-
sign. USEPA, 841-F-93-009. Office of Water, Washington,
D.C.
Hirsch, R.M. and E.J. Gilroy. 1985. Detectability of step trends
in the rate of atmospheric deposition of sulfate. Water Re-
sources Bulletin 21(5):773-784.
IBM. 2011. IBM SPSS Statistics, IBM, http://www-
01.ibm.com/software/ [Accessed September 23, 2011].
Matalas, N.C. 1967. Time series analysis. Water Resources
Research, 3(3):817-829.
SAS Institute Inc. 2010. SAS Software, Version 9.1. SAS In-
stitute Inc. Cary, North Carolina.
SAS Institute Inc. 1999. SAS/ETS User’s Guide, Version 8.
SAS Institute Inc. Cary, North Carolina.
SAS Institute Inc. 2008. JMP Software: Version 8. SAS Insti-
tute Inc., Cary, North Carolina.
Snedecor, G.W. and W.G. Cochran. 1967. Statistical Meth-
ods, Sixth Edition. The Iowa State University Press, Ames,
Iowa. 593 p.
Spooner, J., R.P. Maas, M.D. Smolen and C.A. Jamieson. 1987.
Increasing the sensitivity of nonpoint source control moni-
toring programs. p. 243-257. In: Symposium On
Monitoring, Modeling, and Mediating Water Quality. Ameri-
can Water Resources Association, Bethesda, Maryland.
710p.
Spooner, J., S.L. Brichford, D.A. Dickey, R.P. Maas, M.D.
Smolen, G.J. Ritter and E.G. Flaig. 1988. Determining the
sensitivity of the water quality monitoring program in the
Taylor Creek-Nubbin Slough, Florida Project. Lake and
Reservoir Management, 4(2):113- 124.
US EPA. 2005. Monitoring data: exploring your data, the first
step. NNPSMP Tech Note 1. Office of Water, Washing-
ton, DC. http://ncsu.edu/waterquality/319monitoring/
TechNotes/technote1_exploring_data.pdf [Accessed 11-17-
2011].
US EPA.. 2011. Statistical analysis for monotonic trends,
NNPSMP Tech Note 6, Office of Water, Washington, DC.
http://ncsu.edu/waterquality/319monitoring/TechNotes/
technote6_monotonic_trends.pdf [Accessed 11-17-2011 ].

12
NWQEP NOTES — December 2011
CONFERENCE REPORT
The 19th Annual National Section 319 Nonpoint Source
Monitoring Program (NNPSMP) Workshop was held on
September 25-28, 2011 in Philadelphia, PA.
This symposium integrated the 19th National 319 Non Point
Source Monitoring Workshop, 2011 Pennsylvania Stormwater
Management Symposium and the 5th International Low Im-
pact Development Conference. The conference showcased
approximately 250 platform presentations, 70 professional
posters, several specialized talks, panels, LID short courses,
vendor booths and tours. Attendance was 722 participants.
The symposium was a joint effort of Villanova University, NC
State University and the University of Maryland along with 13
cooperators. The PowerPoint presentations will be accessible
from http://www.bae.ncsu.edu/stormwater/2011lid/.

INFORMATION
USDA NRCS Revises National 590
Nutrient Management Standard
USDA NRCS Revised the National 590 Nutrient Manage-
ment Standard to achieve maximum agricultural and
environmental benefits. This update stresses increased use of
technology and local information. The following press release
gives details:
Syracuse, N.Y., December 13, 2011 – Agriculture Secre-
tary Tom Vilsack today announced that the U.S. Department
of Agriculture has revised its national conservation practice
standard on nutrient management to help producers better
manage the application of nutrients on agricultural land. Proper
application of nitrogen and phosphorus offers tremendous
benefits to producers and the public, including cost savings to
the producer and the protection or improvement of ground
and surface water, air quality, soil quality and agricultural
sustainability.
“Protecting America’s supply of clean and abundant water
is an important objective for USDA,” Vilsack said. “This pre-
cious resource is the foundation for healthy ecosystems and
sustainable agricultural production. USDA provides voluntary
technical and financial assistance to help producers manage
their nutrients to ensure a clean and abundant water supply
while maintaining viable farm and ranch operations.”
The nutrient management conservation practice is an im-
portant tool in the USDA Natural Resources Conservation
Service conservation toolbox. The agency’s staff uses this
conservation practice to help farmers and ranchers apply their
nutrients more efficiently. Proper management of nitrogen and
phosphorus, including the use of organic sources of nitrogen
such as animal manure, legumes and cover crops, can save
producers money. The nutrient management standard provides
a roadmap for NRCS’s staff and others to help producers
apply available nutrient sources in the right amount, from the
right source, in the right place, at the right time for maximum
agricultural and environmental benefits.
NRCS’s nutrient management experts worked with uni-
versities, non-government organizations, industry and others
to revise the standard to ensure it is scientifically sound. Key
changes in the standard include expanding the use of technol-
ogy to streamline the nutrient management process and allowing
states more flexibility in providing site-specific nutrient man-
agement planning using local information when working with
producers. NRCS staff offices will have until January 1, 2013
to comply with erosion, nitrogen and phosphorus criteria for
their state nutrient management standard.
The revised national standard is being released at a time
when the agency is working with various partners to address
nutrient management concerns identified in three recently re-
leased Conservation Effects Assessment Project (CEAP)
cropland studies. These CEAP studies assessed the effective-
ness of conservation practices in the Upper Mississippi Basin,
the Chesapeake Bay Watershed and the Great Lakes Basin.
One significant resource concern identified in all three studies
is the loss of nitrogen and phosphorus from cropland. Most
nitrogen losses are attributed to nitrate leaching through the
soil to groundwater. Most phosphorus is lost due to erosion
because phosphorus attaches itself to displaced soil particles
that are transported by runoff to nearby waterways. Improved
nutrient management and effective erosion control work to-
gether to reduce the loss of nutrients from agricultural land,
resulting in improved water quality in downstream rural and
urban communities. The revised standard will provide tools
and strategies to help producers address the natural resource
concerns relating to excess nutrients on agricultural land.
NRCS offers voluntary technical and financial assistance
to producers nationwide for planning and implementing on-
farm nutrient management plans. Producers can use this
assistance to help meet Federal, State, Tribal and local envi-
ronmental regulations.
Nutrient Management (590) Standard: http://www.nrcs.
usda.gov/Internet/FSE_DOCUMENTS/stelprdb1046177.pdf
For more information about how nutrient management fits
into NRCS’s conservation work, visit our Nutrient and Pest
Management web page: http://www.nrcs.usda.gov/wps/por-
tal/nrcs/main/national/landuse/crops/npm.
Or contact your local USDA-NRCS office.
13
NWQEP NOTES — December 2011
USDA NRCS Initiative on Drainage Water
Management - Mississippi River Basin
USDA NRCS has put a priority emphasis on drainage wa-
ter management in the Mississippi River Basin. NRCS has a
new initiative to increase the adoption of agricultural drainage
water managment (ADWM) for conservation benefits. The
NRCS Ag Water Management Summit was held in October
2011 to help stimulate increased adoption of ADWM nation-
wide as a part of a conservation system. The Summit
highlighted previous research and adoption experiences, the
current situation, and future opportunities. Specific goals of
the Summit were:
Differentiate ADWM from the traditional practice of
draining land for the production of crops.
Explain the current ADWM technologies and potential
innovations (bioreactors, buffers, wetlands, two-stage
ditches, and companion practices)
Assure sound environmental management and protect
ecosystem values.
Action items and observations from the Summit included:
NRCS is developing a strategy to reach out to the
agriculture and conservation communities across the
country over the next year to promote drainage water
management implementation
A systems approach is the most effective way to address
nitrates in our water resources
There is a need to identify viable economic incentives for
producers to implement drainage water management
More information and data on innovative practices are
needed
There is broad support for NRCS’s effort to promote
drainage water management to protect and improve water
quality
Information and presentations are available at http://
www.nrcs.usda.gov/wps/portal/nrcs/detail/national/water/
manage/?&cid=stelprdb1045304.
Ag NPS Regulations
An excellent compilation of state water regulations can be
found at http://www.nationalaglawcenter.org/readingrooms/
waterlaw/.
In addition, the Mississippi River Collaborative, with the
Environmental Law and Policy Center (Chicago) in the lead,
issued a report in May 2010 on agricultural NPS regulations.
Nonpoint source regulations in California, Delaware, Iowa,
Kentucky, Maryland, Oregon and Wisconsin were highlighted.
In addition, vegetative buffer regulations were summarized
for North Carolina and Minnesota. Land application setback
regulations were summarized for Alabama, Arkansas, Colo-
rado, Georgia, Illinois, Iowa, Maine, Minnesota, New Jersey,
Pennsylvania and Wyoming. Winter manure restrictions for
Colorado, Delaware, Illinois, Indiana, Iowa, Kansas, Maine,
Maryland, Minnesota, Tennessee, Vermont and Wyoming were
summarized. Existing state regulations for livestock exclusion
were summarized for Colorado, Kentucky, Minnesota, New
Jersey and Wisconsin. Fall fertilizer restrictions were summa-
rized for Maryland, Minnesota, Nebraska, West Virginia, and
Wisconsin. Titled, “Cultivating Clean Water,” the full report
can be found at http://elpc.org/elpc-publications?PHPSESSID=
87c19f7a767adfa9c8fd4e9b4a79084b
USEPA Healthy Watersheds Initiative
The U.S. Environmental Protection Agency (EPA) recently
announced the release of the Healthy Watersheds Initiative
(HWI) National Framework and Action Plan. The HWI is
intended to protect the nation’s remaining healthy watersheds,
prevent them from becoming impaired, and accelerate resto-
ration successes. The HWI National Framework and Action
Plan aims to provide a clear, consistent framework for action,
both internally among EPA’s own programs and externally in
working with the Agency’s partners. EPA will work with states
and other partners to identify healthy watersheds at the state
scale and develop and implement comprehensive state healthy
watersheds strategies that set priorities for protection and in-
form priorities for restoration.
Healthy watersheds provide many ecological services as
well as economic benefits. If successfully implemented, the
HWI promises to greatly enhance our ability to meet the Clean
Water Act Section 101(a) objective of restoring and maintain-
ing the chemical, physical, and biological integrity of the
nation’s waters. The HWI National Framework and Action
Plan is available at http://www.epa.gov/healthywatersheds.
Biological Assessment Primer
The EPA has published A Primer on Using Biological As-
sessment to Support Water Quality Management. This technical
document serves as a primer on the role of biological assess-
ments in a variety of water quality management program
applications including reporting on the condition of aquatic
biota, developing biological criteria, and assessing environmen-
tal results of management actions. The Primer provides
information on new technical tools and approaches for devel-
oping strong biological assessment programs and includes
examples of the application of biological assessment informa-
tion by states and tribes. More information is available at http:/
/water.epa.gov/scitech/swguidance/standards/criteria/aqlife/
biocriteria/uses_index.cfm.
14
NWQEP NOTES — December 2011
USGS Cooperative Water Program
The USGS Cooperative Water Program monitors and as-
sesses water in every state, protectorate, and territory of the
U.S. in partnership with nearly 1,600 local, state and Tribal
agencies.
The CWP is the foundation for USGS’ robust water moni-
toring networks (quantity and quality), including, for example,
the collection of data at more than 75 percent of the nation’s
7,700 stream gauges. The CWP also supports about 700 in-
terpretative studies annually, covering a wide range of issues
that are important to the USGS water mission and that inform
local, state, and tribal water decisions. Learn more from the
CWP’s revised website at: http://water.usgs.gov/coop/
No Consistent Declines in
Nitrate Levels in Large Rivers in the
Mississippi River Basin
The following USGS findings were released in the journal
of Environmental Science & Technology, accessible at: http:/
/water.usgs.gov/nawqa/pubs/nitrate_trends.
Despite efforts to reduce nitrate in the Mississippi River
basin, nitrate concentrations and transport at eight mainstem
or large tributary sites have not declined from 1980-2008. The
results are based on a new statistical method developed by the
USGS that takes into account variation in river flows, in order
to paint an accurate picture of long-term trends. These results
reflect the cumulative changes over time in nitrate sources
and conservation practices throughout the Mississippi River
basin, and highlight the need for comprehensive nutrient man-
agement strategies that will reduce nutrient levels in both
streams and groundwater.
Key findings:
Nitrate concentrations increased considerably at two sites
(Mississippi River at Clinton, IA and at Missouri River at
Hermann, MO)
Nitrate concentrations remained very nearly the same or
increased at the other six sites (Iowa River at Wapello,
IA; Illinois River at Valley City, IL; Ohio River near Grand
Chain, IL; and along the Mississippi River at Grafton, IL;
at Thebes, IL; and near the Old River Outflow Channel)
Nitrate transport to the Gulf of Mexico increased from
1980 to 2008
Increases in nitrate concentrations in groundwater are
having a substantial effect on nitrate concentrations in rivers
and transport to the Gulf of Mexico. The evidence for
this comes from the fact that increases in nitrate
concentrations are particularly large during low flow
conditions.
Contacts:
Lori Sprague, [e-mail], 303-236-6921
Robert Hirsch, [e-mail], 703-648-5888
New Online Management Tool to Help
Guide Action on Excessive Nutrients in
Rivers and Estuaries
The USGS recently released a decision support system
using SPARROW (Spatially Referenced Regressions On Wa-
tershed attributes) that provides access to six newly-developed
regional models that describe how rivers receive and trans-
port nutrients from natural and human sources to nutrient
sensitive waters such as those along the Atlantic Coast, Gulf
of Mexico, and Pacific Northwest Coast.
Each region and locality has a unique set of nutrient sources
and characteristics that determine how those nutrients are trans-
ported to streams.
“Using the decision support system, users can evaluate
combinations of source reduction scenarios that target one or
multiple sources of nutrients and see the change in the amount
of nutrients transported to downstream waters - a capability
that has not been widely available in the past,” said Stephen
Preston, USGS hydrologist and coordinator for these regional
models. For example, the decision support system indicates
that reducing wastewater discharges throughout the Neuse
River Basin in North Carolina by 25 percent will reduce the
amount of nitrogen transported to the Pamlico Sound from
the Neuse River Basin by 3 percent; whereas a 25 percent
reduction in agricultural sources, such as fertilizer and ma-
nure, will reduce the amount of nitrogen by 12 percent.
Based on the six regional modeling results, wastewater ef-
fluent and urban runoff are significant sources of nutrients in
the Northeast and mid-Atlantic, while agricultural sources such
as farm fertilizers and animal manure contribute heavily to
nutrient concentrations in the Midwest and Central regions of
the nation. Atmospheric deposition is the largest contributor
of nitrogen in many streams in the eastern United States, and
naturally occurring geologic sources are a major source of
phosphorus in many areas.
Additionally, the six models used in the decision support
system show that the amount of nutrients transported varies
greatly among the regions, because nutrients can be removed
in reservoirs or used by plants before they reach downstream
waters. Temperature and precipitation variation across the
country also affect the rates of nutrient movement and loss on
the land and in streams and reservoirs.
15
NWQEP NOTES — December 2011
The USGS developed the SPARROW water-quality mod-
els to assist with the interpretation of available water-resource
data and provide predictions of water quality in unmonitored
streams. The regional SPARROW models incorporate geospatial
data on geology, soils, land use, fertilizer, manure, wastewater
treatment facilities, temperature, precipitation and other wa-
tershed characteristics, from USGS, NOAA, USDA, and
USEPA. These data are then linked to measurements of stream
flow from USGS streamgages and water-quality monitoring
data from approximately 2,700 sites operated by 73 monitor-
ing agencies.
Access the new USGS online decision support system on
nutrients at http://water.usgs.gov/nawqa/sparrow/mrb.
Gulf of Mexico Regional Ecosystem
Restoration Strategy
The Gulf Coast Ecosystem Restoration Task Force released
the Gulf of Mexico Regional Ecosystem Restoration Strategy
in December 2011. Four overarching goals were stated:
Restore and conserve habitat
Restore water quality
Replenish and protect living coastal and marine resources
Enhance community resilience
Specific goals and strategies are detailed in the report: http:/
/www.epa.gov/gulfcoasttaskforce/pdfs/GulfCoast
Report_Full_12-04_508-1.pdf

MEETINGS
Meeting Announcements — 2012
January/Februar y
5th International Perspective on Water Resources & the
Environment (IPWE 2012). Marrakech, Morocco. January
5-8, 2012. http://www.asce.org/events/EventDetail.aspx?id=
12884905950
2012 National Association of Conservation Districts
(NACD) Annual Meeting. Las Vegas, NV. January 29-Feb-
ruary 1, 2012. http://www.nacdnet.org/events/annualmeeting/
March
AWWA 2012 Sustainable Water Management Conference
& Exposition. Marriott Portland Waterfront Hotel, Portland,
Oregon. March 18 - 21, 2012. http://www.awwa.org/Con-
ferences/SustainableManagement.cfm?ItemNumber=
56511&navItemNumber=56514
AWRA Spring Specialty Conference: GIS & Water Re-
sources VII. Sheraton New Orleans Hotel, New Orleans, LA.
March 26-28, 2012. http://awra.org/meetings/Spring2012/
AridLID 2012 Conference: Green Infrastructure and Low
Impact Development in Arid Environments. Tucson, Ari-
zona. March 27-29, 2012. http://AridLID.org
April/May
2012 Student Water Conference. Oklahoma State Uni-
versity, Stillwater, OK. April 4-5, 2012. Only graduate and
undergraduate students are invited to submit abstracts for oral
and poster. http://agwater.okstate.edu/news-events/student-wa-
ter-research-conference (Abstracts are due by 1/6/2012)
8th National Water Monitoring Conference - Water: One
Resource - Shared Effort - Common Future. Portland, OR.
Apr 30 - May 4, 2012. http://acwi.gov/monitoring/conference/
2012/
New England Interstate Water Pollution Control Com-
mission (NEIWPCC) and the New Hampshire Department
of Environmental Services 23rd Annual Nonpoint Source
Pollution Conference for the New England Region.
Sheraton Harborside Hotel, Portsmouth, NH. May 15-16 , 2012.
Information and previous years’ presentations: http://
www.neiwpcc.org/npsconference.
2012 Land Grant and Sea Grant National Water Confer-
ence. Marriott Waterfront Hotel, Portland, OR. May 20-24,
2012. http://www.usawaterquality.org/conferences/2012/ (Ab-
stracts are due by 1/2/2012)
June
Society of Wetland Scientists and the Greater Everglades
Ecosystem Restoration Conference with 9th INTECOL In-
ternational Wetlands Conference. Orlando, FL June 3-8,
2012. http://www.conference.ifas.ufl.edu/INTECOL/
2012 Ohio Stormwater Conference. SeaGate Convention
Center. Toledo, OH. June 7-8, 2012. http://
www.ohioswa.com/calendar-events/annual-conference/
Urban Environmental Pollution 2012: Creating Healthy,
Liveable Cities. Amsterdam, Netherlands. Jun 17-20, 2012.
http://www.uepconference.com/ (Abstracts due 1/16/2012)
16
NWQEP NOTES — December 2011
July/August
11 th International Conference on Precision Agriculture.
Hyatt Regency, Indianapolis, IN. July 15-18, 2012. https://
www.ispag.org/ (Abstracts still being accepted)
WEF Stormwater Symposium 2012. Sheraton Baltimore City
Center, Baltimore, Maryland, July 18 – 20, 2012. http://
www.wef.org/stormwater2012/
67th Annual International Conference for the Soil and
Water Conservation Society (SWCS): Choosing Conser-
vation: Considering Ecology, Economics and Ethics. Ft.
Worth, TX. July 22-25, 2012. http://www.swcs.org/12ac
2012 American Society of Agricultural & Biological Engi-
neers (ASABE) Annual International Meeting. Dallas, TX.
July 29-August 1, 2012. http://www.asabemeetings
.org/
October
2012 Stream Restoration Conference. Wilmington, NC.
October 15-18, 2012. http://www.bae.ncsu.edu/training_and_
credit/workshops.php
20th Annual Nonpoint Source (NPS) Monitoring Work-
shop. Double Tree Hilton at Warren Place in Tulsa, OK. October
14-17, 2012. http://npsmonitoring.tetratech-ffx.com/
November
The 32nd International Symposium of the North Ameri-
can Lake Management Society. Madison, WI. November 7
– 9, 2012. http://www.nalms.org/
2012 Annual Water Resources Conference. Hyatt Regency
Jacksonville Riverfront, Jacksonville, FL. Nov. 12-15, 2012.
http://awra.org/
December
ACES and Ecosystem Markets 2012. Ft. Lauderdale, FL.
December 10-13, 2012. http://www.conference.ifas.ufl.edu/
aces (Abstracts still being accepted)

Production of NWQEP NOTES is funded through U.S.
Environmental Protection Agency (USEPA) contract
#EP-C-08-004. Task Order Manager: Paul Thomas,
Water Division, EPA Region 5. 77 W. Jackson St.,
Chicago, IL 60604.
20th Annual Nonpoint Source
Monitoring Workshop
The Secrets of Success Making the Most
of Available Resources
Tulsa, Oklahoma | October 14–17, 2012
The Annual Nonpoint Source (NPS) Monitoring
Workshop is an important forum for sharing informa-
tion and improving communication on ways to control
and track NPS pollution at its source and in receiving
waterbodies. The focus of the 20th National Workshop
is cost-efficiency across a range of topics including plan-
ning and implementation of land treatment to solve NPS
problems, water quality monitoring for NPS problem
assessment and project effectiveness, data sharing for
multiple purposes, and communication of NPS water
quality issues and findings to the general public.
This event will bring together NPS monitoring and
management personnel from state, federal, Tribal and
municipal governments, the private sector, academia, en-
vironmental groups and local watershed organizations to
provide examples of lessons learned from completed NPS
projects, demonstrations of new technologies and moni-
toring approaches, and documentation of successful
application of NPS control practices and measures.
A number of technical workshops and interactive
learning sessions will be offered to build knowledge and
skills, transfer technology, and promote innovative moni-
toring and evaluation techniques. Field tours will be
offered in both agricultural and urban settings.
Please mark your calendar and check the conference
website (http://npsmonitoring.tetratech-ffx.com) for de-
tails on abstract topics, the workshop agenda, and
registration information. The deadline for abstract sub-
mittal is May 1, 2012. Abstracts and short bios should
be emailed to [e-mail].
17
NWQEP NOTES — December 2011
NCSU Water Quality Group
Campus Box 7637
North Carolina State University
Raleigh, NC 27695-7637
Telephone: [phone]
Fax: [phone]
Website: http://www.ncsu.edu/waterquality/
Personnel
Jean SpoonerKaren R. Hall
Robert O. EvansWilliam F. Hunt
Kristopher BassGregory D. Jennings
Jamie BlackwellBonnie Kurth
Michael R. Burchell IIDaniel E. Line
Barbara A. DollCatherine S. Smith
Garry L. GrabowLaura Lombardo Szpir
NC STATE UNIVERSITY