TCAA Water Quality Data Review and Information-Sharing Program: Final Report

TCAA Water Quality Data Review
and Information-Sharing Program
FINAL REPORT
University of Florida Water Institute & UF/IFAS Extension
Mark Clark1, Wendy Graham2, Kathleen McKee3 and Jeff Ullman4
1
Wetlands and Water Quality Extension Specialist, Soil and Water Science Department
2
Carl S. Swisher Chair in Water Resources, Director UF Water Institute
3
Research Coordinator, UF Water Institute
4
Assistant Professor, Department of Agricultural and Biological Engineering
June, 2012
Executive summary
In response to a request by growers in the Tri-County Agricultural Area (TCAA), and with cooperation
from the Florida Department of Agriculture and Consumer Services (FDACS), St. Johns River Water
Management District (SJRWMD), and Florida Department of Environmental Protection (FDEP), the
University of Florida Water Institute and Institute of Food and Agricultural Sciences Extension Service
reviewed information resources (Appendix A) and had discussions with SJRWMD and FDEP to better
understand the how impairment of the Lower St. Johns River was determined, how nutrient loadings
were estimated and how nutrient load reduction allocations associated with the Lower St. Johns River
TMDL were calculated. This effort aims to help improve grower understanding of the TMDL/BMAP
process and address several specific questions identified by growers.
Specific questions identified by growers and therefore addressed by the reviewing authors were:
1. What assumptions, methods, and data were used to:
a. Determine that the LSJR is impaired?
b. Determine nutrient loads to the LSJR?
i. What are the estimated relative load contributions for various land uses in the
basin?
ii. Is there information available to show how these loads have changed over time?
c. Develop the TMDLs for the LSJR?
d. Develop the load reduction allocations for the LSJR BMAP and for the TCAA in particular?
i. Is the required load reduction aggregated by total land use in the basin, or is it a
required reduction per unit acre of a specific land use?
ii. Was the feasibility of reducing nutrient loads and effectiveness of implementing
agricultural and urban BMPs considered when allocating load reductions?
iii. What is the performance of the two existing RSTs?
2. Were the assumptions, data, and models used to develop the TMDL and BMAP for the LSJR
consistent with best professional practices for development of other TMDLs? Did the process make
use of the best information available? If not, how could the assumptions, data, models, and
development process be improved?
It was determined that data used to verify that the LSJR was impaired were based on monitoring data
collected principally between 1996 and 2003. This data indicated that chlorophyll-a concentrations
and Trophic State Index (TSI) values were above levels that the State of Florida’s Impaired Water Rule
(IWR) indicate are likely causing impairment to the river as a result of excess nutrients, and although
not specifically used in the verification process, dissolved oxygen levels in the estuarine reaches of the
river (measured during the TMDL development process) did not meet the State of Florida’s dissolved
oxygen standard for Class III marine waters.
Development of the Total Maximum Daily Load (TMDL) for nitrogen and phosphorus in the river was
based on determining the nutrient levels in the water column that would keep algae (as measured by
University of Florida Water Institute and UF/IFAS Extension, June 2012
2
chlorphyll-a concentration) at a level not to exceed 40 µg/L chlorophyll-a for more than 10% of the
time. This threshold was appropriately developed for site-specific conditions in the freshwater portion
of the LSJR that would 1) maintain the diversity of the plankton community, 2) allow for upward
transfer of primary production to higher trophic levels while maintaining zooplankton diversity and 3)
minimize the potential dominance of detrimental algal species and production of algal toxins. The
TMDL was also based on those levels of nutrients that would allow for the Site Specific Alternative
Criteria (SSAC) for oxygen – that was developed for the marine portion of the river – to be met.
The integration of a SSAC for oxygen levels in the marine portion of the river and determination of a
chlorophyll-a threshold of 40 µg/L for less than 10% of the time in the freshwater portion were
prudent deviations from state wide water quality standards as they better represent actual conditions
occurring in the LSJR and more appropriately establish protective levels of nutrients and
biogeochemical response characteristics for this waterbody.
To determine the actual TMDL of nutrients that could be assimilated by the river without going above
the chlorophyll-a response threshold, or below the dissolved oxygen threshold, three models were
used in combination. The PSLM (Pollution Load Screening Model) model was used to estimate
seasonal and sub-basin loading to the river, the EFDC (Environmental Fluid Dynamics Code) model was
used to simulate hydrologic dynamics in the river, and the CE-QUAL-ICM (Corps of Engineers Water
Quality Integrated Compartment Model) model was used to simulate biogeochemical processes and
interactions within the river and to ultimately determine the relationship between various refractory
vs. labile nutrient concentration and chlorophyll-a or dissolved oxygen levels. Use of these models is
generally consistent with best practices for TMDL and BMAP development. The process-based EFDC
and CE-QUAL-ICM models are among the most sophisticated physically-based river models in use, and
probably the best models suited to simulate the complexities of the Lower St. Johns River.
The PLSM is a statistical model that has been used to estimate watershed runoff and nutrient loadings
at various locations in Florida. The strength of the PLSM model used in the LSJR TMDL development is
that it is an empirical model constrained by observed rainfall, streamflow and water quality data taken
within the LSJR. Nevertheless, the procedure used to develop the model and estimate its parameters
required many assumptions. While these assumptions are generally physically-reasonable and
observation-based, future work should formally assess the sensitivity and uncertainty of PLSM
predictions to model structure and parameter uncertainty. In addition, evaluation of the model’s
predictive value should be conducted with additional streamflow and water quality data that have
been collected since the model was originally developed. Conducting these activities in close
collaboration with agricultural stakeholders should help build confidence that the PLSM model
developed for the LSJB is as accurate as possible and suitable for the purpose intended.
While there is uncertainty over the precise values of nutrient loading estimates provided by the PLSM
for the LSJRB, available measured data and model estimates indicate that agricultural practices have
contributed to increased nutrient levels in the LSJR.
Performance of Regional Treatment Systems (RSTs) to assist in reduction of agricultural nutrient loads
not met through implementation of BMPs appears very effective relative to load reduction targets.
University of Florida Water Institute and UF/IFAS Extension, June 2012
3
Long-term efficacy of these systems may change and will need to be monitored, but based on
performance data from the Deep Creek RST (operational since 2006), nitrogen and phosphorous
removal rates are presently significantly greater than target levels.
The reviewing authors found it difficult to navigate the myriad of documents and often complex, and
sometimes conflicting, wording within documents to determine what data and information was used
to support models and assumptions in the TMDL and BMAP. For instance, the specific year that was
used to estimate land use for the PLSM urban and agricultural nonpoint source load estimates used in
the TMDL/BMAP process is not clearly specified in the agency documents1 and was only clarified during
a follow up meeting with agencies. The authors understand that information used in support of this
TMDL and BMAP spanned more than a decade and these documents are not necessarily intended for
understanding by the general public. However clarifying information used in the TMDL/BMAP,
producing a synthesis document targeted at the general public within the basin, and evaluating the
clarity of documents to all basin stakeholders would likely reduce questions and improve stakeholder
understanding and acceptance of the TMDL/BMAP process.
During this review, the authors found a discrepancy between the considerable efforts made by
agencies to facilitate stakeholder awareness and elicit input during the LSJR TMDL/BMAP process; and
the actual awareness by agricultural stakeholders of the process, their understanding of the technical
details and their perception that their concerns were actually being heard. Since the greatest
expectation for success in addressing nutrient loading is a partnership among all stakeholders,
resolving this discrepancy in future BMAP processes should be a high priority. Jordan et al (2011), NRC
(2001, 2008), and Pahl-Wostl (2009) provide interesting insights on strengths and weaknesses of
various methods for engaging the public in environmental decision making, and provide strategies for
bringing multi-stakeholder groups together to improve the capacity, salience, legitimacy, and
credibility of environmental governance and decision-making for issues ranging from TMDL
development to global climate change. During the next round of TMDL/BMAP evaluation as well as
implementation of TMDL/BMAP programs in other watersheds, alternative strategies for engaging
agricultural stakeholders in the process should be explored.
1
For example Hendrickson and Konwinski (1998) p. 6 indicates that 1990 land use was used to
estimate nonpoint source loads. Hendrickson et al (2002) p. 30 indicates that land use information
compiled from 1989-90 was used. The BMAP document (2008) p. 12 indicates that 2000 land use
was used to estimate agricultural nonpoint loads, whereas p. 29 in the same BMAP document
indicates that for nonpoint sources 1997-1998 land use was used to set the initial load and
projected 2008 land uses were used to estimate starting loads. Figure 3 in the BMAP document
shows 2004 land use. At a meeting on 11/3/2011, SJRWMD personnel indicated that 1995 land use
for the LSJR basin was utilized in PLSM to estimate the starting loads for the BMAP, except within
the TCAA where agricultural acreage was updated using results from a 2000 field survey.
University of Florida Water Institute and UF/IFAS Extension, June 2012
4
Table of Contents
Executive summary ....................................................................................................................................................2
Introduction ................................................................................................................................................................6
1.a. What assumptions, methods, and data were used to determine that the LSJR is impaired? ...........................7
1.b. What assumptions, methods, and data were used to determine nutrient loads to the Lower St. Johns River
(LSJR)? ......................................................................................................................................................... 13
1.b.i. What are the estimated relative load contributions for various land uses in the basin? ............................. 29
1.b.ii. Is there information available to show how these loads have changed over time? ..................................... 34
1.c. What assumptions, methods, and data were used to develop the TMDLs for the LSJR BMAP and for the
TCAA in particular? ..................................................................................................................................... 39
1.d. What assumptions, methods, and data were used to develop the load reduction allocations for the LSJR
BMAP and for the TCAA in particular? ....................................................................................................... 40
1.d.i. Is the required load reduction aggregated by total land use in the basin, or is it a required reduction per
unit acre of a specific land use? ................................................................................................................. 44
1.d.ii. Was the feasibility of reducing nutrient loads and effectiveness of implementing agricultural and urban
BMPs considered when allocating load reductions?.................................................................................. 45
1.d.iii. What is the performance of the two existing RSTs? ................................................................................... 45
2. Were the assumptions, data, and models used to develop the loads for the LSJR consistent with best
professional practices for development of other TMDLs........................................................................... 49
References ............................................................................................................................................................... 51
Appendix A - Documents Reviewed ........................................................................................................................ 55
Appendix B - Grower Questions from November 1, 2011 Listening Session .......................................................... 56
Introduction
In response to a request by growers in the Tri-County Agricultural Area (TCAA), and with cooperation
from the Florida Department of Agriculture and Consumer Services (FDACS), St. Johns River Water
Management District (SJRWMD), and Florida Department of Environmental Protection (FDEP), the
University of Florida Water Institute and Institute of Food and Agricultural Sciences Extension Service
(the reviewers) developed a scope of work (SOW) which provides for a review of information
resources and methods used by the SJRWMD and FDEP to estimate nutrient loadings and establish
nutrient load reduction allocations associated with the Lower St. Johns River (LSJR) total maximum
daily load (TMDL). In the scope of work, reviewers were charged with evaluating specific reports and
documents related to the development of the LSJR TMDL and Basin Management Action Plan (BMAP)
in an effort to answer the following grower questions.
1. What assumptions, methods, and data were used to:
a. Determine that the LSJR is impaired?
b. Determine nutrient loads to the LSJR?
i. What are the estimated relative load contributions for various land uses in the
basin?
ii. Is there information available to show how these loads have changed over time?
c. Develop the TMDLs for the LSJR?
d. Develop the load reduction allocations for the LSJR BMAP and for the TCAA in particular?
i. Is the required load reduction aggregated by total land use in the basin, or is it a
required reduction per unit acre of a specific land use?
ii. Was the feasibility of reducing nutrient loads and effectiveness of implementing
agricultural and urban BMPs considered when allocating load reductions?
iii. What is the performance of the two existing RSTs?
2. Were the assumptions, data, and models used to develop the TMDL and BMAP for the LSJR
consistent with best professional practices for development of other TMDLs? Did the process make use
of the best information available? If not, how could the assumptions, data, models, and development
process be improved?
The body of the report that follows focuses on answers and recommendations related to these original
questions posed in the SOW. Appendix B lists additional questions posed by growers during a listening
session held by reviewers on November 1, 2011. Questions in Appendix B are addressed in a separate
document titled “Grower Questions from November 1, 2011 Listening Session.”
University of Florida Water Institute and UF/IFAS Extension, June 2012
6
1.a. What assumptions, methods, and data were used to determine that the LSJR is impaired?
Legal Basis
The process by which a water body is determined to be impaired and the course of action to be taken
once a water body is declared impaired are rooted in Section 303(d) of the federal Clean Water Act
(CWA). Under this act, states are required to submit to the USEPA lists of waters that are not fully
meeting their designated uses based on applicable water quality standards determined to be
protective of the designated use. In 1999, the Florida Watershed Restoration Act (FWRA) (Section
403.067, Florida Statutes [F.S.]) directed the Florida Department of Environmental Protection (FDEP) to
develop, and adopt by rule, a new science-based methodology to identify impaired waters. The new
methodology (adopted as Rule 62-303, F.A.C., in April 2001, and revised in 2006 and 2007) is titled
“The Impaired Waters Rule” (IWR).
Included in the IWR are methods to: a) identify surface waters that will go on the state’s planning list,
which means the surface water is potentially impaired based on data available at the time, and b)
verify that a surfacewater on the planning list is indeed impaired based on additional data collected
from the surface water. Once a water body is verified impaired, it is included on a list of waters for
which FDEP will calculate Total Maximum Daily Loads (TMDLs) for pollutants.
When verifying whether a surface water is impaired, data collected in the surface water for the 5-year
period prior to the assessment are compared to water quality standards established to protect the
designated use of the water. The state-defined designated use for the main stem of the Lower St.
Johns River is that of “Class III” water, which means that it is expected to provide for “fish
consumption, recreation, and propagation and maintenance of a healthy, well-balanced population of
fish and wildlife.” In some instances the standards that protect these uses are numeric and easily
compared to monitoring data; in the case of nutrients, however, the standards at the time were based
on a narrative criteria that reads “In no case shall nutrient concentrations of a body of water be altered
so as to cause an imbalance in natural population of flora or fauna” (62-302.530, F.A.C). Translating
this narrative standard to a numeric standard for nitrogen and phosphorus is a critical and necessary
step in the development of the TMDL. Guidance for this determination is outlined in several sections of
the IWR, including Evaluation of Aquatic Life Use Support (62-303.310), Biological Assessment (62303.330), and Interpretation of Narrative Nutrient Criteria (62-303.350). How IWR guidance was
applied in verifying impairment of the LSJR is outlined next.
Verifying Impairment
Verification of impairment of the main stem of the LSJR was based on several lines of evidence and a
general preponderance of research findings conducted in the area. A more detailed elaboration on
each of the lines of evidence used by FDEP is outlined below. In the LSJR, the nutrients nitrogen and
phosphorus were identified as increasing the Chlorophyll-a concentration (a proxy for algal biomass in
the water column) and Trophic State Index (which is calculated based on the chlorophyll-a, total
nitrogen, and total phosphorus concentrations in the water) above acceptable levels.
University of Florida Water Institute and UF/IFAS Extension, June 2012
7
When applying the various water quality standards to a surface water body a distinction between
freshwater and marine waters is appropriate, since each system may have different sensitivities to, and
natural ranges of, a particular parameter or contaminant. The LSJR is divided into three ecological
zones based on salinity (Figure 1): 1) a predominantly freshwater, tidal, lake-like zone that extends
from south of the city of Palatka north to the mouth of Black Creek; 2) an oligohaline zone extending
from Black Creek northward to the Fuller Warren Bridge (I-95) in Jacksonville; and 3) a predominantly
marine zone (meso-haline) near the mouth.
Figure 1. Differentiation between freshwater and estuarine (oligohaline and meso-polyhaline)
sections of the river (Magley and Joyner, 2008).
University of Florida Water Institute and UF/IFAS Extension, June 2012
8
Figure 2. Water body Identification numbers (WBIDs) for segments along the main stem of the Lower
St. Johns River (Magley and Joyner, 2008).
In addition to partitioning between marine and freshwater zones for the purpose of applying water
quality standards, linear surface water bodies such as the LSJR are also typically partitioned into
smaller segments or reaches, identified as Water Body Identification Units, or WBIDs for short. The
LSJR has 15 WBIDs, starting with 2213A at the mouth of the river and alphabetically going upstream to
2213O (Figure 2).
University of Florida Water Institute and UF/IFAS Extension, June 2012
9
The Marine Zone: Dissolved Oxygen Standards
In the case of Class III waters in marine zones (Figure 1), the water quality standard indicates that
minimum dissolved oxygen (DO) levels can be no less than 4 mg/L at any point in time and that the
minimum daily average cannot be less than 5 mg/L. However, in some instances where natural
fluctuations in DO below these standard criteria are known to occur, Florida Water Quality Standards
(62-302, F.A.C.) indicate that natural conditions should not be diminished. This means that if DO levels
below the 4 mg/L minimum or daily average of 5 mg/L are known to occur naturally for a waterbody,
then a Site Specific Alternative Criterion (SSAC) that more appropriately characterizes DO conditions
for that surface water should be considered. Although dissolved oxygen levels in the marine portion
of the river were not used to verify impairment, monitoring of marine segments of the river during the
TMDL process indicated that minimum dissolved oxygen levels for Class III marine waters were not
being met. To address this issue a SSAC was established for DO in the marine portion of the river that
better represented the natural variability of DO levels and tolerance of native fish and aquatic
invertebrates to low dissolved oxygen.
Figure 3. Daily-averaged dissolved oxygen (DO) concentrations at Acosta Bridge and Dames Point of
the marine portion of the LSJR. Graphics are from Appendix F of TMDL document (Magley
and Joyner, 2008) comparing observed data with model simulation of DO.
Data used to verify low DO levels in the marine portion of the LSJR is principally based on continuous
monitoring of DO concentrations between 1996 and 2003 at the Dames Point Bridge and the Acosta
Bridge. This zone of the river is where freshwater and saltwater mix, resulting in an environment
where neither freshwater algae nor marine algae do well. This often leads to depressed oxygen levels
as algae entering this reach of the river from mainly upstream sources die, because as microbes break
down the dead algae they consume oxygen from the water column. If more algae enter this mixing
zone, stimulated by higher nutrient levels upstream, then the additional dead algae are broken down
by the microbes, which can result in even higher consumption of oxygen and lower DO concentrations
University of Florida Water Institute and UF/IFAS Extension, June 2012
10
in the water. A comparison of data during this time period (Figure 3) to the minimum average 5 mg/L
standard for Class III waters indicated that both of these bridge stations had periods when the DO
concentrations were below acceptable levels.
Freshwater Zone: Chlorophyll-a and Trophic State Indices.
At the time of the LSJR TMDL (2004), protective criteria related to nutrient concentrations (specifically
nitrogen and phosphorus) were based on narrative criteria and not numeric criteria. The IWR states
that nutrient impairment can be inferred using certain biological response measurements, including
the amount of chlorophyll-a in the water column or by applying the Trophic State Index (TSI, see Figure
4). Nutrients such as nitrogen and phosphorus tend to be limiting resources in water bodies, so
increased concentrations tend to result in increased plant growth, followed by increases in subsequent
trophic levels. Therefore, a water body's TSI can often be used to estimate its biological condition.
TSI = (CHLATSI+ NUTRTSI)/2
Where:
CHLATSI = 16.8 + 14.4 * LN(Chl-a)
TNTSI = 56 + 19.8 * LN(N)
TN2TSI= 10 * [5.96 + 2.15 * LN(N+ 0.001)]
TPTSI = 18.6 * LN(P * 1000)-18.4
TP2TSI = 10 * [2.36 * LN(P * 1000)-2.38]
if N/P>30, then NUTRTSI = TP2TSI
if N/P<10, then NUTRTSI = TN2TSI
if 10<N/P<30, then NUTRTSI = (TPTSI+TNTSI)/2
Figure 4. Equations used to calculate TSI (62.303.200, F.A.C.). CHLA=chlorophyll-a;
NUTR=nutrients; TN=total nitrogen; TP=total phosphorus
As defined in the IWR, when applying either measure, there is a requirement of collecting at least four
samples seasonally throughout the year (January 1-March 31, April 1-June 30, July 1-Sept. 30, and Oct.
1-Dec. 31). In the monitoring of the LSJR to support the TMDL, sampling was performed at numerous
locations, and six samples were collected per season (24 samples per year), greatly strengthening the
statistical significance of the determination of impairment. In the case of chlorophyll-a, if average
concentrations are greater than 20 micrograms (µg)/L in freshwater (62-303.351, F.A.C.) or 11 µg/L in
an estuary or open coastal water (62-303.353, F.A.C.), or if data indicate annual mean chlorophyll-a
values have increased by more than 50% over historical values for at least two consecutive years (in
either fresh or marine systems), then the system is deemed to be impaired.
In the case of the TSI, average freshwater values greater than 60 in this system are considered to
indicate an impaired condition (62-303.352, F.A.C.). During the impairment verification period (1996-
University of Florida Water Institute and UF/IFAS Extension, June 2012
11
2003), water column samples were collected within each WIBID, and eleven of the 15 WBIDs in the
LSJR were verified as impaired when compared to the IWR criteria (Table 1).
Table 1. Chlorophyll-a or TSI annual average values for WBIDs in the LSJR determined to be impaired
(from Appendix I of LSJR TMDL document - Magley and Joyner, 2008). Impairment was verified for
WBIDs 2213A-2213F based on chlorophyll-a values greater than 11 or exceeding the historic minimum
by more than 50% for two or more consecutive years; for WBIDs 2213I-2213L based on TSI values
greater than 60; and for WBIDs 2213M and 2213N based on chlorophyll-a values greater than 20. See
Figure 2 for specific location of WBID.
Additional lines of evidence used to support listing of LSJR as impaired
In addition to dissolved oxygen, chlorophyll-a, and TSI data that were used to verify impairment of the
LSJR, other lines of evidence were presented in the TMDL document (Wayne and Magley 2008). These
lines of evidence came from various sources, including FDEP (2002); Burkholder and Glasgow (1997a;
1997b); Phlips et al. (2000); Bricker et al. (1999); USEPA (2001); Janicki and Morrison (2000);
Hendrickson and Konwinski (1998). In summary, the following list of impacts occurring in the LSJR
were identified, most of which are either a direct measure of elevated nutrient loads, or can be
attributed to excess nutrients resulting in eutrophication and an imbalance in flora and fauna:
 Submersed aquatic shoreline vegetation covered in algal mats
 Excessive epiphyte growth further blocking light from submerged aquatic vegetation
 Anecdotal accounts of shoreline vegetation losses and reduced recreational fishing quality
 River sediment conditions indicative of low benthic animal diversity
 Excessive organic matter sedimentation and prolonged anoxia (lack of oxygen)
 The presence of potentially toxic dinoflagellates such as Pfiesteria-like Crytoperidiniopsoids and
Prorocentrum minimum
University of Florida Water Institute and UF/IFAS Extension, June 2012
12




Fish kills
An increase in combined, within-basin point and nonpoint source nutrient pollution load to the
LSJR, 2.4 times over natural background for TN and 6 times for TP.
One of the highest reported areal nutrient loading rates in the southeastern United States, at
9.7 and 2.1 kilograms of nitrogen and phosphorus per hectare of watershed per year.
Changes in the amounts of river algae that appear to correlate significantly with changes in
inorganic nitrogen and DO, suggesting that algae use much of the nitrogen supplied to them for
growth. However, during this cycle of growth and ultimate death, the algae exert a dominant
influence over river oxygen content.
1.b. What assumptions, methods, and data were used to determine nutrient loads to the Lower St.
Johns River (LSJR)?
The modeling approach for estimating nutrient loads for the LSJR TMDL utilized statistical relationships
between rainfall, land use, and soils to estimate streamflow, stream-water nutrient concentrations and
stream-water nutrient load for each sub-basin of the LSJR basin. The formulation of this statistical
model has its roots in the spreadsheet watershed load screening model, referred to as the Pollution
Load Screening Model (acronym PLSM; Adamus and Bergman, 1995). Details regarding the data used
in the model are included below.
Input parameters:



Rainfall - Measured rainfall within the LSJR basin was apportioned over the basin using a
Theissen polygon approach (Chand et al., 2005) and aggregated into three seasons: December 1
through March 31, April 1 through July 31, and August 1 through November 30. Separate
streamflow, stream nutrient concentrations and stream nutrient load predictions were then
made for each season and summed to get annual totals.
Land use - Land throughout the LSJR basin was classified into 15 land uses: Low-density
residential, medium-density residential, high-density residential, low-intensity commercial,
high-intensity commercial, industrial, mining, forest range/open/barren, pasture, livestock, row
crops, miscellaneous agriculture, citrus, water surfaces, and wetlands. The 1995 land use map
for the LSJR basin was utilized in PLSM to estimate the starting loads for the BMAP, except
within the TCAA where agricultural acreage was updated using results from a 2000 field survey
(Hendrickson, personal communication, 2011).
Soil hydrologic groups – Existing Natural Resources Conservation Service (NRCS) drainage
classifications from the Soil Survey Geographic Database (SSURGO) were used. These
hydrologic classes include: A (well drained), B (moderately well drained), C (moderately
drained), and D (poorly drained).
University of Florida Water Institute and UF/IFAS Extension, June 2012
13

Figure 5.
2000 Land use (SJRWMD, 2002) for the five watersheds used for PLSM runoff coefficient
estimation, and sampling station locations used for PLSM water quality coefficient
estimation.
University of Florida Water Institute and UF/IFAS Extension, June 2012
14
Predicted outputs

Streamflow at the outlets of sub-basins throughout the LSJRB. Streamflow predictions were
evaluated using data collected from approximately 1992 through 1999 from 5 calibration
watersheds (See Figure 5)

Stream water nutrient loads at the outlets of sub-basins throughout the LSJRB. Stream water
quality coefficients were fit using data from 29 sampling stations (see Figure 5). Stream
nutrient load predictions were evaluated using data collected from approximately 1992 through
1999 from 5 calibration watersheds.
For a statistical model such as PLSM, model coefficients (in this case runoff coefficients and water
quality coefficients) that relate model input parameters to predicted model outputs are estimated
based on an assumed model structure and the available data. The PLSM model structure assumes that
for each land area that has a particular land use and is underlain by a particular soil class (i.e. unique
land use and soil group combination i, [see Figure 6]) the runoff and nutrient load are then estimated
as follows:

Seasonal Runoff Volume for land use and soil group combination i= Seasonal Rainfall depth*
Land Area for land use and soil group combination i*Seasonal Runoff Coefficient (the fraction of
rainfall that becomes runoff) for land use and soil group i*Conversion Factors

Seasonal Nutrient Load for land use and soil group combination i= Seasonal Runoff Volume for
land use and soil group combination i * Seasonal Water Quality Coefficient (Concentration) for
land use*Conversion Factors
where * represents multiplication. For each sub-basin, the seasonal pollutant load is estimated by
adding the contributions from each land area and soil group combination i in the sub-basin. The totals
from all the sub-basins are then summed to estimate the total seasonal load from the sub-basin to the
LSJR.
Figure 6. Example of two soil types (green and brown polygons) overlain with three land use
types (different hatchings), would yield six unique land use and soil group combinations
(i) and result in six runoff volume calculations and six nutrient load calculations.
University of Florida Water Institute and UF/IFAS Extension, June 2012
15
Estimation of Runoff coefficients
Runoff coefficients were estimated for each land use and soil group combination and each season
based on literature values; then model predictions were regressed against streamflow data measured
from August 1, 1992, and July 30, 1995, at the outlets of five “calibration watersheds” to assess the
accuracy of the results. The five calibration watersheds chosen by SJRWMD included Deep Creek
(includes the 16 mile creek), Black Creek North Fork, Black Creek South Fork, Ortega River, and Cedar
River (see Figure 5). These watersheds were chosen based on the quality of their data and
representativeness of the regions in the LSJR basin. The Black Creek North and South Fork basins are
primarily forested, draining the Trail Ridge to the west of the river. The Deep Creek basin is primarily
agricultural, pasture, and forested within the Atlantic Coast Flatwoods to the east of the river. The
Cedar and Ortega River basins drain the Duval Upland, a region of relic beach ridges and swales of the
Sea Island District, and are largely urbanized. Figure 5 shows the 2000 land use for these five
calibration watersheds.
Initial estimates for runoff coefficients for all land uses except row crops and silviculture were taken
from Adamus and Bergman (1995). These coefficients were based on estimates used in Chow (1964; a
standard hydrology reference) and Harper (1994; a literature review of Florida stormwater studies) as
well as ranges of rainfall-runoff ratios observed for undeveloped watersheds with relatively
homogeneous land use in the region (John Hendrickson, personal communication 2012). Estimates of
row crop runoff coefficients were derived from data collected during the SJRWMD's three-year TriCounty Agricultural Area (TCAA) Best Management Practices (BMP) Study, conducted under the Lower
St. Johns River Basin SWIM program (SWET, 1994). Estimates for runoff coefficients for silviculture
were obtained from a literature review of forested coastal plain watershed hydrologic studies (Beasley
and Granillo, 1988; Hollis et al., 1981; Riekerk, 1983; Riekerk, 1989; Schrieber et al., 1976; and Speir et
al., 1969).
The non-seasonal runoff coefficients (for each land use and soil group combination) described above
were modified to create three sets of seasonal coefficients to reflect the relative seasonal differences
observed in the St. Johns River Basin (Hendrickson and Konwinski, 1998). Values of runoff coefficients
for all land uses except row crop agriculture and water surfaces were “seasonalized” by adjusting up
for the August - November and December - March seasons ( to reflect higher rainfall expected during
these seasons), and down for the April - July season (to reflect higher evapotranspiration expected in
this season). The magnitude of the adjustment was based on the observed seasonal changes in runoff
for long term discharge monitoring watersheds (John Hendrickson, personal communication, 2012).
For row crop agriculture the runoff coefficients varied by season but not by soil type based on data
from the SWET (1994) study. For water surfaces, all rainfall was assumed to become runoff regardless
of season, therefore a runoff coefficient of 1.0 was assigned for all seasons.
Using these seasonal runoff coefficient estimates, seasonal model streamflow predictions from PLSM
were regressed against actual measured seasonal streamflow data from August 1992 to July 1995 for
the five calibration watersheds to assess the model fit for each season (9 seasons*5 watersheds = 45
data points). This regression yielded a model coefficient of determination (R2) of 0.615, indicating that
61.5% of the variability in the seasonal streamflow data is accounted for by the PLSM (Hendrickson and
Konwinski, 1998). The resulting season, soil, and land use-specific runoff coefficients for application of
University of Florida Water Institute and UF/IFAS Extension, June 2012
16
the PLSM to the lower St. Johns River Basin are shown in Table 2, and the fit of the model to the data is
shown in Figure 7. For comparison, the Harper (1994) coefficients are plotted with the SJRWMD season
1 coefficients in Figure 8. The final SJRWMD coefficients are generally similar to the Harper
coefficients, however they are based more closely on conditions in the LSJR basin. The Harper
coefficients are average values taken from reports of studies conducted throughout Central and South
Florida.
Figure 7. Station and season comparisons of PLSM-predicted and measured runoff for Lower St.
Johns River calibration watersheds (from Hendrickson and Konwinski, 1998).
University of Florida Water Institute and UF/IFAS Extension, June 2012
17
1.2
H et al A
1
H et al B
0.8
H et al C
H et al D
0.6
Harper
0.4
0.2
0
Figure 8. Comparison of SJRWMD Season 1 runoff coefficients by land cover for the four soil groups
(H et al A – D) (Hendrickson and Konwinski, 1998; Hendrickson et al, 2002) with Harper
(1994) runoff coefficients (which do not vary seasonally or by soil group).
In general, a coefficient of determination of 0.615 is considered ‘fair’ for pollutant screening models.
To improve this correlation, improvements were made to the PLSM runoff coefficients by Hendrickson
et al (2002). In the judgment of the SJRWMD modelers, the biggest shortcoming of the Hendrickson
and Konwinski PLSM model in estimating streamflow was the failure of the model to adjust for
seasonal to annual changes in shallow ground water storage due to long term trends in rainfall, and the
effect of this on soil saturation and the propensity of a rainfall event to generate runoff. To adjust for
this in the TMDL nonpoint source load modeling, Hendrickson et al (2002) developed an additional
empirical adjustment, referred to as the long-term rain ratio adjustment that accounts for changes in
antecedent moisture conditions associated with intra-annual patters in rainfall and evapotranspiration.
This improved the agreement between model predicted seasonal volume, and observed volume. After
applying this adjustment, the PLSM linear regression R2 statistic between model-predicted and
observed seasonal streamflow volume from 1992 through 1999 improved to 0.8037 as shown in Figure
9.
University of Florida Water Institute and UF/IFAS Extension, June 2012
18
Figure 9. Comparison of modeled versus observed seasonal flow for original PLSM (Hendrickson and
Konwinski, 1998) and long-term rain ratio adjusted PLSM (Hendrickson et al., 2002) for the
calibration watersheds (from Hendrickson et al., 2002).
The strength of the PLSM runoff model is that it is an empirical model constrained by observed rainfall
and streamflow data taken within the LSJB. Nevertheless the procedure used to develop the model,
estimate the seasonal runoff coefficients and estimate the long-term rain ratio adjustment factor
required many assumptions and best professional judgment calls as described in Hendrickson and
Konwinski (1998) and Konwinski et al. (2002).2 While the assumptions made are physically-reasonable
and observation based, the resulting model structure required that 180 runoff coefficients (15 land
uses*4 soil types*3 seasons= 180 runoff coefficients) be specified, with only 45 seasonal data points
available to evaluate the fit of the model to the data, resulting in a classic problem of model and
parameter non-uniqueness (Madsen et al., 2002; Beven, 2006). In other words a different set of
assumptions, different parameter estimation methodology, and different set of observed data from
the same system could have resulted in a different model that fit the data as well, or perhaps better
than the adopted model. Future work should formally test the sensitivity of PLSM streamflow volume
model predictions to model structure and parameter uncertainty to help build confidence among
agricultural stakeholders that the PLSM model developed for this application is as accurate as possible
and suitable for the purpose intended. In addition evaluation of the model’s predictive value should
be conducted with additional data that has been collected since the model was originally developed.
2
Examples of assumptions and best professional judgments include the selection of literature values and
experiments on which to base the runoff coefficients, the way runoff coefficients were modified to reflect
seasonal differences, the way runoff was modified to reflect long-term changes in rainfall, the way land uses
were categorized, etc.
University of Florida Water Institute and UF/IFAS Extension, June 2012
19
Table 2. Seasonal runoff coefficients for application of the Pollution Load Screening Model to the LSJRB
(from Hendrickson and Konwinski, 1998; Hendrickson et al., 2002).
Land Use
A
Soil Hydrologic Group
B
C
D
WellDrained
Moderately
Well-Drained
Moderately
Drained
Poorly
Drained
0.05
0.5
0.6
0.5
0.7
0.5
0.05
0.05
0.05
0.401
0.05
0.05
0.05
1
0.95
0.12
0.6
0.7
0.6
0.8
0.6
0.12
0.12
0.12
0.401
0.12
0.12
0.12
1
0.95
0.18
0.7
0.8
0.7
0.9
0.7
0.18
0.18
0.18
0.401
0.18
0.18
0.18
1
0.95
0.25
0.8
0.9
0.8
1
0.8
0.25
0.25
0.25
0.401
0.25
0.25
0.25
1
0.95
0
0.2
0.3
0.2
0.4
0.2
0
0
0
0.392
0
0
0
1
0.75
0
0.3
0.4
0.3
0.5
0.3
0
0
0
0.392
0
0
0
1
0.75
Season 1: December through March
Low Density Residential
Medium Density Residential
High Density Residential
Low Intensity Commercial
High Intensity Commercial
Industrial
Mining
Miscellaneous Agriculture
Pasture
Row Crop
Citrus
Livestock Feedlots
Forestry, Silviculture, Range, Barren
Water Surfaces
Wetlands
Season 2: April through July
Low Density Residential
Medium Density Residential
High Density Residential
Low Intensity Commercial
High Intensity Commercial
Industrial
Mining
Miscellaneous Agriculture
Pasture
Row Crop
Citrus
Livestock Feedlots
Forestry, Silviculture, Range, Barren
Water Surfaces
Wetlands
University of Florida Water Institute and UF/IFAS Extension, June 2012
0.05
0.4
0.5
0.4
0.6
0.4
0.05
0.05
0.05
0.392
0.05
0.05
0.05
1
0.75
0.1
0.5
0.6
0.5
0.7
0.5
0.1
0.1
0.1
0.392
0.1
0.1
0.1
1
0.75
20
Table 2 (continued). Seasonal runoff coefficients for application of the Pollution Load Screening Model
to the LSJRB (from Hendrickson and Konwinski, 1998; Hendrickson et al., 2002).
Land Use
A
Soil Hydrologic Group
B
C
D
WellDrained
Moderately
Well-Drained
Moderately
Drained
Poorly
Drained
0.05
0.55
0.65
0.55
0.7
0.55
0.05
0.05
0.05
0.512
0.05
0.05
0.05
1
1
0.15
0.65
0.75
0.65
0.8
0.65
0.15
0.15
0.15
0.512
0.15
0.15
0.15
1
1
0.25
0.75
0.85
0.75
0.9
0.75
0.25
0.25
0.25
0.512
0.25
0.25
0.25
1
1
0.35
0.85
0.95
0.85
1
0.85
0.35
0.35
0.35
0.512
0.35
0.35
0.35
1
1
Season 3: August through November
Low Density Residential
Medium Density Residential
High Density Residential
Low Intensity Commercial
High Intensity Commercial
Industrial
Mining
Miscellaneous Agriculture
Pasture
Row Crop
Citrus
Livestock Feedlots
Forestry, Silviculture, Range, Barren
Water Surfaces
Wetlands
Estimation of Water Quality coefficients
Water quality coefficients were originally estimated for each land use and each season using multiple
linear regression on seasonal flow-weighted mean in-stream nutrient concentrations calculated from
measurements taken from December 8, 1992, and July 30, 1995 from the five calibration watersheds
described above, as well as 24 additional watersheds where sufficient water quality data was available
to use in the estimation process (See Figure 5). For the 24 stations where streamflow was not available,
flow-weighted estimates were made using flow quartiles measured at the nearest of three flowmeasurement stations (South Fork, Ortega, and Deep Creek). Note that by calibrating water quality
coefficients to measured constituent loads at in-stream locations rather than edge-of-field boundaries
some attenuation/assimilation of nutrients between the source and the stream may be incorporated
into these coefficients.
To reduce the number of parameters that needed to be estimated by the multiple least squares
regression, land use types were combined into four categories (Hendrickson and Konwinski, 1998;
Konwinski et al., 2002):
1. Residential/urban (the sum of medium and high density residential, all commercial, and
industrial land uses);
2. Row crop agriculture (row crops, citrus, miscellaneous agriculture);
3. Dairy and pasture (sum of improved pasture, manure spray fields and livestock feed lots); and
University of Florida Water Institute and UF/IFAS Extension, June 2012
21
4. Undeveloped (the sum of low density residential, mining, silviculture, forestry, range/open
space, and wetlands).
Multiple linear regression was then performed to estimate water quality coefficients for each land use
and each season based on the fraction of areal cover of each of these aggregated land use types within
each of the 29 basins, and the seasonal flow-weighted nutrient concentrations calculated for each of
the basin outlets. In some cases (i.e. where there were only two predominant land uses in a basin)
simultaneous equations were also solved to obtain coefficients, either to compare to the regression
results or to provide coefficients when the regression results produced negative estimates
(Hendrickson and Konwinski, 1998; J. Hendrickson personal communication, 2012).
After the regression/simultaneous equation analysis, SJRWMD assigned separate water quality
coefficients to urban land uses that had been aggregated for the regression approach. Within the
aggregated urban use category, water quality coefficients for the component land uses were
distributed generally following trends reported by Harper (1994) for the given water quality
constituent, with deviations in some instances in the specific ordering to account for differences in
land use definitions. No distinction was made for water quality coefficients for silviculture, forestry,
range/open space, mining, and wetlands, which were all supplied the coefficients that were derived
from the aggregated undeveloped land uses in the multiple regression. Row crop agriculture was
directly assigned the water quality coefficients derived for the aggregated row crop land uses in the
multiple regression. For the relatively small areas of the basin identified as miscellaneous agriculture
and citrus, the average annual loading rates of Harper (1994) were used directly without seasonal
proportioning. Because water quality coefficients developed for dairy lands were influenced by areas
within the watershed of both pasture and higher-density animal feedlots, which are separately defined
in the land use coverage, SJRWMD assigned pasture lands a value somewhat lower than the mean,
while high-density livestock staging areas were assigned a value somewhat higher.
Table 3 presents the final seasonal water quality coefficients used in the PLSM model. For comparison,
Table 4 presents measured nutrient surface runoff concentrations for four farms that participated in
the SJRWMD TCAA BMP study (Livingston Way, 2001). Figures 10 and 11 compare the PLSM and
Harper water quality coefficients for TN and TP, respectively. Row crop agriculture coefficients fitted
through the multiple regression approach were found to be higher than those reported by Harper
(1994) from studies in South and Central Florida, but similar to those derived from the Tri-county
Agricultural Area BMP Study (SWET, 1994). High intensity livestock and intensive pasture were also
found to have higher water quality coefficients than those reported by Harper (1994).
University of Florida Water Institute and UF/IFAS Extension, June 2012
22
Table 3. Seasonal water quality coefficients used in the Pollutant Load Screening Model (PLSM) to
predict nonpoint source loads to the LSJR (to establish TMDL baseline load). All values represent flowweighted concentrations in mg/L (from Hendrickson et al., 2002). TN=total nitrogen, TP=total
phosphorus, BOD=biological oxygen demand, SS=suspended solids, TIN=total inorganic nitrogen,
TPO4=total orthophosphate.
Season 1: Dec. 1 through Mar. 31
Land Use
Low Density Res
Medium Density Res.
High Density Res.
Low Dens. Commercial
High Dens. Commercial
Industrial
Forest, Range/ Open, Barren
Pasture
Row Crop, Misc. Ag
Livestock
Water (Atmos Wetfall)
Wetlands
TN
0.8
1.4
1.8
1.1
1.2
1.2
0.7
3.9
2
4.5
0.28
0.7
TP
0.08
0.25
0.3
0.2
0.3
0.25
0.06
0.75
0.38
1.3
0.017
0.06
BOD
1
2
4
2
4
2
1
4
1
6
0
1
SS
6
15
20
15
25
25
25
15
15
15
0
3
TIN
0.04
0.35
0.4
0.35
0.4
0.4
0.02
1
0.7
1.5
0.28
0.02
TPO4
0.06
0.1
0.13
0.1
0.13
0.1
0.04
0.6
0.2
1
0.015
0.04
Season 2: Apr. 1 through Jul. 31
Land Use
Low Density Res.
Medium Density Res.
High Density Res.
Low Dens. Commercial
High Dens. Commercial
Industrial
Forest, Range/ Open, Barren
Pasture
Row Crop, Misc. Ag
Livestock
Water (Atmos Wetfall)
Wetlands
TN
0.8
1.6
2
1.2
2
1.2
0.7
3
10.7
6
0.49
0.7
TP
0.07
0.3
0.5
0.3
0.5
0.3
0.05
1.1
1.8
1.3
0.014
0.05
BOD
1
2
4
2
4
2
1
4
1
6
0
1
SS
6
30
40
30
50
40
40
10
54
30
0
3
TIN
0.04
0.2
0.3
0.3
0.4
0.4
0.04
1
4.7
1.2
0.49
0.04
TPO4
0.05
0.1
0.12
0.1
0.12
0.1
0.03
0.85
0.6
1
0.014
0.03
Season 3: Aug. 1 through Nov. 30
Land Use
Low Density Res.
Medium Density Res.
High Density Res.
Low Dens. Commercial
High Dens. Commercial
Industrial
Forest, Range/ Open, Barren
Pasture
Row Crop, Misc. Ag
Livestock
Water (Atmos Wetfall)
Wetlands
TN
0.8
1.5
1.7
1.3
1.7
1.3
0.8
3
4.4
5
0.47
0.7
TP
0.09
0.35
0.53
0.22
0.53
0.22
0.09
2
2.2
2.6
0.028
0.07
BOD
1
2
4
2
4
2
1
4
1
6
0
1
SS
5
25
35
15
35
25
3
15
26
3
0
3
TIN
0.06
0.41
0.55
0.3
0.55
0.3
0.04
1.2
0.55
1.7
0.47
0.04
TPO4
0.07
0.16
0.22
0.13
0.22
0.13
0.05
2.1
1.6
2.4
0.016
0.05
University of Florida Water Institute and UF/IFAS Extension, June 2012
23
Table 4. Measured annual average flow-weighted TN and TP concentrations in surface runoff before
and after implementation of varying combinations of water and nutrient management best
management practices for row crops on four study farms in the Tri-County Agricultural Area (from
Livingston Way, 2001).
Farm C
(Nutrient
BMP)
Farm G
(Water BMP)
Farm I
(Nutrient and
Water BMPs)
Farm J
(Nutrient and
Water BMPs)
Pre-BMP
TN (m)g/L
Post BMP
TN (mg/L)
Pre-BMP
TP (mg/L)
Post-BMP
TP (mg/L)
Source of data
5-13
5-12
0.5-2
4-12
read off Figure 16 , p30
2-13
4-30
0.5-2.5
4-14
read off Figure 17, p31
5-13
3-13
3.1-4.8
1.1-2.8
read off Figure 18, p32
1-5
1-10
0.6-1.8
1.0-1.8
read off Figure 19, p 33
12
10
8
6
4
Harper
H et al 1
2
0
H et al 2
H et al 3
Figure 10. Comparison of total nitrogen (TN) water quality coefficients (mg/L) used by Harper (1994)
and the three seasonal TN coefficients used by PLSM (Hendrickson et al., 2002).
University of Florida Water Institute and UF/IFAS Extension, June 2012
24
3
2.5
2
1.5
1
Harper
0.5
H et al 1
0
H et al 2
H et al 3
Figure 11. Total phosphorus (TP) water quality coefficients (mg/L) used by Harper (1994) and the three
seasonal TP coefficients used by PLSM (Hendrickson et al., 2002).
Model Evaluation
To evaluate model validity stream nutrient load was simulated with the calibrated PLSM and compared
to measured stream nutrient load. In the initial PLSM development (Hendrickson and Konwinski, 1998)
the model was run with the August 1992 through July 1995 seasonal rainfall coverages to simulate
individual constituent loads from the five calibration basins where pollutant load could be calculated,
i.e., for which both water quality and discharge quantity were available. . Based on the eight seasons
(the unusually dry Aug.-Nov. season of 1993 was omitted) and five stations, forty seasonal pollutant
mass loads were available for this comparison. Correlation coefficients relating PLSM predictions to the
flow-weighted means of measured data were fair, with values from 0.63 to 0.74 indicating that the
model explains 63% to 74% of the variability in the data. Figures 12-16 contain plots of PLSMsimulated versus measured flow-weighted seasonal mean areal mass loading from Hendrickson and
Konwinski (1998) and show the degree to which the model captures observed differences in nutrient
loading rates among watersheds and across seasons.
University of Florida Water Institute and UF/IFAS Extension, June 2012
25
Figure 12. PLSM-simulated versus calculated watershed flow-weighted seasonal mean total nitrogen
areal mass loading rates in grams per hectare (from Hendrickson and Konwinski, 1998).
Figure 13. PLSM-simulated versus calculated watershed flow-weighted seasonal mean total
phosphorus areal mass loading rates in grams per hectare (from Hendrickson and
Konwinski, 1998).
University of Florida Water Institute and UF/IFAS Extension, June 2012
26
Figure 14. PLSM-simulated versus calculated watershed flow-weighted seasonal mean total inorganic
nitrogen (TIN) areal mass loading rates in grams per hectare (from Hendrickson and Konwinski, 1998).
Figure 15. PLSM-simulated versus calculated watershed flow-weighted seasonal mean total
orthophosphate (PO4) areal mass loading rates in grams per hectare (from Hendrickson and
Konwinski, 1998).
University of Florida Water Institute and UF/IFAS Extension, June 2012
27
Figure 16. PLSM-simulated versus calculated watershed flow-weighted seasonal mean total suspended
solids (TSS) areal mass loading rates in grams per hectare (from Hendrickson and Konwinski,
1998).
The CE-QUAL-ICM model used to determine the assimilative capacity of, and the TMDL for, the LSJR
requires total nitrogen (TN) and total phosphorus (TP) to be specified in terms of labile (easily broken
down) and refractory (slowly broken down) components. Therefore Hendrickson et al. (2002) modified
the PLSM framework developed by Hendrickson and Konwinski (1998) to partition organic nitrogen
(TON=TN-TIN) and non-orthophosphate phosphorus (TP-TPO4) into its labile and refractory
components. To accomplish this, Hendrickson et al. (2002; 2007) developed water quality coefficients
for total organic carbon (TOC), labile total organic carbon (LTOC) and refractory total organic carbon
(RTOC) using the same methodology described above for the other constituents. They then estimated
labile and refractory organic nitrogen and non-orthophosphate phosphorus based on estimated
relationships of these quantities to observed ratios of LTOC and RTOC to TOC. Updated nutrient loads
for the full suite of labile and non-labile constituents were estimated for the 1992-1999 time period
using the new water quality coefficients and the new rainfall ratio adjusted streamflow estimates
described above. Note that the coefficients for TN, TP, BOD, SS, TIN and TPO 4 developed by
Hendrickson and Konwinski (1998) using the 1990-1995 data (see Table 3 above) were left unchanged
in this application. Unfortunately updated plots and regression statistics for observed versus
estimated nutrient loads that are directly comparable to those shown in Figures 12-16 were not
included in Hendrickson et al, 2002.
University of Florida Water Institute and UF/IFAS Extension, June 2012
28
As mentioned above in the discussion of the PLSM streamflow model predictions, due to model
structure assumptions3, data limitations and model fitting procedures, the PLSM constituent load
model structure and parameter values are non-unique. It should be noted that virtually all hydrologic
and water quality models suffer from non-uniqueness due to unavoidable model error, measurement
error and natural variability. Thus a formal sensitivity analysis of the PLSM model that examines the
sensitivity of model predictions to parameter values and model assumptions should be conducted to
help build agricultural stakeholder confidence that the PLSM model developed for this application is as
accurate as possible and suitable for the purpose intended. In addition, evaluation of model’s
predictive performance using independent flow and water quality data collected since the model was
developed is recommended.
1.b.i. What are the estimated relative load contributions for various land uses in the basin?
Figures 17 through 24 summarize the PLSM estimates of the 1995-1999 average TN and TP loads for
the fresh tidal region, oligohaline region, meso-polyhaline region, and the total Lower St Johns River,
respectively. All data is taken from Hendrickson et al. (2002). These data show that in the fresh tidal
region that encompasses the TCAA, natural nonpoint sources are the major contributor to TN loads at
48%. However, as pointed out by Hendrickson et al. (2002, 2007) the external load from this natural
nonpoint TN load is largely in the form of refractory organic N, which breaks down slowly and does not
promote algal growth as readily as labile organic N. Agricultural nonpoint loads and point-source loads
are each estimated to contribute 24%, with 71% of this TN estimated to be in the form of labile organic
N and inorganic (fertilizer) N, which is readily assimilated for algal growth (Hendrickson et al., 2002).
Because of the small fraction of urban area in this reach urban nonpoint source loads are estimated to
contribute only 4%. For TP, agricultural nonpoint source loads are estimated to be the highest at 39%,
with point-source loads estimated at 31% , natural nonpoint source loads estimated at 23%, and urban
nonpoint source loads estimated at 6%. In the oligohaline and mesohaline portions of the river
agriculture’s contribution to TN and TP loads are estimated to be much smaller because they are a
small fraction of the total land area.
3
Examples of assumptions and best professional judgments include the method to aggregate/disaggregate the land uses,
the method by which seasonal flow-weighted water quality was estimated when no streamflow was available, the method
used to disaggregate nutrients into labile and refractory forms, etc.
University of Florida Water Institute and UF/IFAS Extension, June 2012
29
Fresh Tidal Region Total N Loads
Point Source
24%
Other Nonpoint
0%
Urban Nonpoint
4%
Natural Nonpoint
48%
Agricultural
Nonpoint
24%
Figure 17. PLSM-estimated TN Loads in the Fresh Tidal Region (1995-1999) from Hendrickson et al.
(2002).
Fresh Tidal Region Total P Loads
Point Source
31%
Other Nonpoint
1%
Urban Nonpoint
6%
Natural Nonpoint
23%
Agricultural
Nonpoint
39%
Figure 18. PLSM-estimated TP Loads in the Fresh Tidal Region (1995-1999) from Hendrickson et al.
(2002).
University of Florida Water Institute and UF/IFAS Extension, June 2012
30
Oligohaline Region Total N Loads
Point Source
29%
Natural Nonpoint
47%
Other Nonpoint
0%
Urban Nonpoint
22%
Agricultural
Nonpoint
2%
Figure 19. PLSM-estimated TN Loads in the Oligohaline Region (1995-1999) from Hendrickson et al.
(2002).
Oligohaline Region Total P Loads
Point Source
38%
Other Nonpoint
0%
Natural Nonpoint
23%
Agricultural
Nonpoint
4%
Urban Nonpoint
35%
Figure 20. PLSM-estimated TP Loads in the oligohaline region (1995-1999) from Hendrickson et al.
2002).
University of Florida Water Institute and UF/IFAS Extension, June 2012
31
MesoPolyhaline Region Total N Loads
Natural
Nonpoint
14%
Agricultural
Nonpoint
1%
Urban Nonpoint
13%
Point Source
72%
Other Nonpoint
0%
Figure 21. PLSM-estimated TN Loads in the meso-polyhaline region (1995-1999) from Hendrickson et
al. (2002).
MesoPolyhaline Region Total P Loads
Natural Nonpoint
5%
Agricultural
Nonpoint
1%
Urban Nonpoint
15%
Other Nonpoint
0%
Point Source
79%
Figure 22. PLSM-estimated TP Loads in the meso-polyhaline Region (1995-1999) from Hendrickson et
al. (2002).
University of Florida Water Institute and UF/IFAS Extension, June 2012
32
LSJR Total N Loads
Natural Nonpoint
34%
Point Source
45%
Urban
Nonpoint
13%
Agricultural
Nonpoint
8%
Other Nonpoint
0%
Figure 23. PLSM-estimated TN Loads in LSJR (1995-1999) from Hendrickson et al. (2002).
LSJR Total P Loads
Natural
Nonpoint
14%
Point Source
57%
Agricultural
Nonpoint
12%
Urban Nonpoint
17%
Other Nonpoint
0%
Figure 24. PLSM-estimated TP Loads in LSJR (1995-1999) from Hendrickson et al. (2002).
University of Florida Water Institute and UF/IFAS Extension, June 2012
33
1.b.ii. Is there information available to show how these loads have changed over time?
Annual modeled loads have been estimated for 1995-1999. All of these estimates use the same land
use (i.e., the 1995 land use map for the LSJR basin, with agricultural land-uses within the TCAA
updated using 2000 survey data) but different annual rainfall totals. No runs have been made to date
that vary land use over time to look at estimated changes in load resulting from changes in land use.
However, time series for TN and TP measured by the SJRWMD at the outlet of the five calibration
watersheds from approximately 1989 to present are included in Figures 25 and 26 below. Statistics for
these times series are included in Tables 6 and 7 below. The year 2000 land use distributions
(SJRWMD, 2002) for these watersheds, aggregated into the lumped PLSM categories described in
section 1.b above, are indicated in Figure 27. For contrast, Figures 28 and 29 and Tables 8 and 9 show
time series and statistics for TN and TP measured at five agricultural watersheds in the LSJR basin.
These data indicate that sub-basins dominated by agricultural land uses have higher mean TN and TP
concentrations and higher peak concentrations than those dominated by forested or urban land uses.
The periodic peaks of TN and TP observed in agricultural watersheds indicate a potential to reduce
total loads by managing extreme events through the use of alternative nutrient management practices,
on-site water retention and/or regional stormwater treatment systems.
30
25
TN (mg/l)
20
TN in Calibration Watershed Outlets
North Fork Black Creek
South Fork Black Creek
Ortega
Cedar
Deep Creek
15
10
5
0
Figure 25. Measured total nitrogen (TN) time series from approximately 11/1989 through 11/2011 for
the stations of the five calibration watersheds in the LSJR basin (data obtained from
SJRWMD, Jan 12, 2012).
University of Florida Water Institute and UF/IFAS Extension, June 2012
34
Table 6. Total nitrogen (TN– mg/L) statistics for the five calibration watersheds in the LSJR basin
TN Statistic
Date range
# observations
mean
median
min
max
std dev
Deep Creek
(DPB)
11/7/1989 12/27/2011
531
1.89
1.47
0.43
26.81
2.19
Black Creek
North Fork
3/20/1991 8/24/2011
245
0.64
0.59
0.06
2.34
0.29
Black Creek
South Fork
3/13/1990 –
8/24/2011
252
0.60
0.59
0.09
2.71
0.30
Ortega
7/10/1995 6/16/2011
103
0.79
0.76
0.32
2.03
0.23
Cedar
8/24/1993
8/17/2011
122
1.16
1.15
0.44
1.75
0.24
TP in Calibration Watershed Outlets
4
3.5
TP (mg/L)
3
North Fork Black Creek
South Fork Black Creek
Ortega
Cedar
Deep Creek
2.5
2
1.5
1
0.5
0
Figure 26. Measured total phosphorus (TP) time series from approximately 11/1989 through 11/2011
for the stations of the five calibration watersheds in the LSJR basin. (data obtained from
SJRWMD, Jan 12, 2012).
University of Florida Water Institute and UF/IFAS Extension, June 2012
35
Table 7. Total phosphorus (TP – mg/L) statistics for the five calibration watersheds in the LSJR basin.
TP Statistic
Date range
# observations
mean
median
min
max
std dev
Deep Creek
(DPB)
11/7/1989 12/27/2011
530
0.55
0.45
0.08
1.78
0.35
Black Creek
North Fork
3/20/1991 8/24/2011
243
0.05
0.05
0.00
0.60
0.05
Black Creek
South Fork
3/13/1990 –
8/24/2011
249
0.13
0.13
0.03
0.31
0.04
Ortega
7/10/1995 6/16/2011
103
0.08
0.08
0.03
0.41
0.06
Cedar
8/24/1993
8/17/2011
122
0.187
0.18
0.055
0.476
0.075
Percent of Total Land Use
100
90
80
70
60
50
40
30
20
10
0
urban
livestock
row crop
undev upland
wetlands
Cedar
Ortega
N Fork Black S Fork Black
16mile
Deep Creek
Figure 27. Land use percentages for the five calibration watersheds in the LSJR basin based on analysis
of 2000 spatial data from SJRWMD (2002). Deep Creek land use in this graph also
encompassess the land use in the 16-Mile Creek watershed (see Figure 5).
University of Florida Water Institute and UF/IFAS Extension, June 2012
36
30
TN in Agricultural Watershed Streams
Deep Creek
Mocassin Branch
25
Outlet of Hastings Drainage District
Dog Branch
TN (mg/L)
20
16Mi
15
10
5
0
Figure 28. Measured total nitrogen (TN) time series for five agricultural watersheds in the LSJR basin
from approximately 11/1989 through 11/2011 (data obtained from SJRWMD, Nov 16,
2011).
Table 8. Total nitrogen (TN – mg/L) statistics for five agricultural watersheds in the LSJR basin
TN Statistic
Date range
# of observations
mean
median
min
max
std dev
Deep Creek
11/7/1989 12/27/2011
505
1.89
1.49
0.43
26.81
2.17
Mocassin
Hastings Drainage
Dog Branch
16 Mile Creek
Branch
District
10/22/1987 10/22/1987 2/5/1985 6/22/1999 12/21/2009
12/21/2009
12/21/2009
1/9/2012
286
293
306
131
1.53
1.40
1.95
1.21
1.26
1.18
1.35
1.05
0.08
0.15
0.18
0.49
8.80
19.84
16.89
7.07
1.30
1.39
2.09
0.77
University of Florida Water Institute and UF/IFAS Extension, June 2012
37
TP in Agricultural Watershed Streams
4
Deep Creek
Mocassin Branch
3.5
Outlet of Hastings Drainage District
Dog Branch
3
16Mi
TP (mg/L)
2.5
2
1.5
1
0.5
0
Figure 29. Measured total phosphorus (TP) time series for five agricultural watersheds in the LSJR
basin from approximately 11/1989 through 11/2011 (data obtained from SJRWMD, Nov 16,
2011).
Table 9. Total phosphorus (TP – mg/L) statistics for fiver agricultural watersheds in the LSJR basin.
TP Statistic
Date range
# of observations
mean
median
min
max
std dev
Deep Creek
11/7/1989 12/27/2011
502
0.56
0.46
0.08
1.78
0.35
Mocassin
Hastings Drainage
Dog Branch
16 Mile Creek
Branch
District
10/22/1987 –
10/22/1987 2/5/1985 6/22/1999 –
12/21/2009
12/21/2009
12/21/2009
1/9/2012
286
297
308
131
0.32
0.33
0.38
0.12
0.23
0.22
0.24
0.08
0.03
0.04
0.04
0.08
3.11
6.36
3.13
0.84
0.33
0.46
0.41
0.14
University of Florida Water Institute and UF/IFAS Extension, June 2012
38
1.c. What assumptions, methods, and data were used to develop the TMDLs for the LSJR BMAP and
for the TCAA in particular?
After verification of nutrient impairment, but before a TMDL for contaminants of concern can be
determined, a threshold concentration (mass per unit of volume) or load (mass per unit of time) for
each contaminant of concern must be determined. In the LSJR, the nutrients nitrogen and phosphorus
were identified as increasing the chlorophyll-a concentration and trophic state index above acceptable
levels in the freshwater section, and lowering the dissolved oxygen concentration in the marine
portion of the LSJR below acceptable levels. This determination was based on guidance provided by
the IWR. However, the thresholds outlined in the IWR are not specifically nutrient standards; they
need not be and were not used as targets for TMDLs. This section of the report outlines the methods,
assumptions, and data that were used to determine a threshold nutrient level that would be protective
of the designated use in the LSJR as it relates to chlorophyll-a and the site-specific alternative dissolved
oxygen standard that was developed for the marine section of the river. The Trophic State Index,
although used in the impairment verification process, was not used to establish the maximum loading
level for the freshwater section of the river.
Determination of nutrient impairment threshold for the freshwater zone in the LSJR
Prior to establishment of the initial TMDL for the LSJR in 2003, the SJRWMD developed a Pollutant
Load Reduction Goal (PLRG) for the freshwater zone of the LSJR, based on chlorophyll-a values
(Hendrickson et al. 2003). In that document, the authors outlined the maximum algal biomass levels
that would:
 Maintain the diversity of the plankton community – findings based on monitoring data and
studies conducted by Phlips and Cichra (2002)
 Allow for upward transfer of primary production to higher trophic levels while maintaining
zooplankton diversity – findings based on monitoring data and studies by Phlips and Cichra
(2002)
 Minimize the potential dominance of detrimental algal species and production of algal toxins –
data and findings provided by Phlips and Cichra (2002) and Pearl et al. (2002).
Hendrickson et al. (2003) concluded that:
 these three characteristics were met in the Lower St. Johns River if chlorophyll-a concentrations
did not exceed 40 µg/L (micrograms per liter) more than 10% of the time, and
 if levels rise above 40 µg/L for extended periods there is a shift in algal species toward
undesirable blue-green and toxic algal species and a decline in zooplankton communities, all of
which would signal an imbalance in the aquatic flora and fauna as well as potential impairment
of recreational activity due to potentially harmful algal blooms.
The protective chlorophyll-a threshold concentration identified from the Hendrickson et al. (2003)
study differed considerably from the criteria outlined in the IWR, which used a level of 20 µg/L.
However, since the 40 µg/L for no more than 10% of the time determination was based on specific
monitoring data from the LSJR, and the IWR rule is based on state wide averages (not exclusively local
data) and supports the determination of site specific alternative criteria if warranted, the Stakeholder
Committee, TMDL Executive Committee, and ultimately FDEP agreed to support the idea that
University of Florida Water Institute and UF/IFAS Extension, June 2012
39
maintaining chlorophyll-a levels below 40 µg/L 90% of the time would provide a reasonable
expectation that an imbalance in natural populations of flora and fauna would be avoided. This
threshold was then used as a reference against which modeling of chlorophyll-a dynamics in the LSJR
could be evaluated relative to nutrient loads.
The actual determination of TMDL targets for nitrogen and phosphorus is based on a suite of models,
including a water quality model that simulates the transformation of nutrients and processes affecting
eutrophication in the river. This model is a three-dimensional, time-variable process model developed
by the U.S. Army Corps of Engineers (USACE) and titled: Quality Integrated Compartment Model (CEQUAL-ICM) version 2 (Cerco and Cole, 1993). This model is among the more sophisticated water
quality process models available and allows assessment of benthos, zooplankton, submerged aquatic
vegetation, as well as the more standard interaction of phytoplankton, oxygen dynamics, nutrients,
and organic carbon. In addition to the CE-QUAL-ICM water quality process model and the previously
discussed PLSM that calculates seasonal runoff and nutrient loads for each sub-basin, a hydrodynamic
model (Environmental Fluid Dynamics Code - EFDC) was used to simulate the mixing and transport of
nutrients in the river (Hamrick, 1992; Sucsy and Morris, 2002).
By using these three models in combination, comparing output variables of the CE-QUAL-ICM against
the chlorophyll-a threshold criteria of less than 40 µg/L for 90% of the time, and verifying that the
alternative site-specific dissolved oxygen criterion in the marine section of the LSJR was being met, a
maximum annual load (the TMDL) of 8,571,563 kg/yr TN and 500,325 kg/yr TP in the freshwater
portion and 1,376,855 kg/yr TN in the marine portion was determined.
1.d. What assumptions, methods, and data were used to develop the load reduction allocations for
the LSJR BMAP and for the TCAA in particular?
At its simplest, a TMDL is comprised of a waste load allocation (WLA) that represents the portion of the
allowable loading capacity assigned to: 1) existing and future point sources and nonpoint sources
holding a stormwater permit under the National Pollutant Discharge Elimination System (NPDES), 2) an
allowable load allocation (LA) that signifies the portion of the loading capacity assigned to existing and
future non-NPDES nonpoint sources and natural background conditions, and 3) a margin of safety.
Correspondingly, load reductions are developed for the respective stakeholders when actual loads to a
waterbody are greater than the allowable load allocations.
The Florida Watershed Restoration Act (FWRA) mandates that a TMDL include the “establishment of
reasonable and equitable allocations…among point and nonpoint sources…” Point sources include
direct discharges such as industrial and municipal wastewater treatment facilities (WWTFs), municipal
separate storm sewer systems (MS4s), and other non-MS4 stormwater discharges, while nonpoint
sources include agricultural runoff, urban runoff, septic tanks, atmospheric deposition, and natural
background contributions. Although the Florida Administrative Code states that “an even-handed and
balanced approach to attainment of water quality objectives” should be taken by FDEP [Rule 62302.300(10)(c), F.A.C.), this does not mean that all pollutant sources are to be managed the same, and
University of Florida Water Institute and UF/IFAS Extension, June 2012
40
the FDEP TMDL Protocol document asserts that allocations should consider the relative contributions
of each source and impacts on the receiving water body (FDEP, 2006).
To address policy issues involving the different options related to the TMDL program, including the
allocation of point and nonpoint sources, the FWRA required FDEP to form an Allocation Technical
Advisory Committee (ATAC) in 1999.
In Florida, the Basin Management Action Plan (BMAP) developed for a TMDL is the associated
implementation plan, which details the allowable loads to the impaired water body, the required load
reductions to achieve water concentrations that meet water quality standards, and the actions that
stakeholders will take to meet their load reductions. After the starting loads were determined for the
LSJR basin, based on various assumptions outlined in Table 10, and the TMDL was determined based
on the methodology outlined above, the required load reduction was allocated according to guidance
developed by the ATAC (FDEP, 2001).
Table 10. Assumptions made to determine starting loads and allocations as listed in the LSJR
BMAP (LSJR TMDL Executive Committee, 2008).
Water quality targets were established based on ecological health as it related to cultural
eutrophication because there are no numeric criteria for nutrients. In the freshwater section, the
target was based on chlorophyll a concentrations. In the marine section, the target was based on
dissolved oxygen conditions.
The initial loads for point sources are based on the loads from 1997-1998. For those facilities that
were operating at or near capacity during that time, the permitted load was used as the starting
point load. In areas where growth was expected, the starting point load was the load that would
be needed by the facility through 2008.
Future growth is addressed as part of the allocation process by increasing the “starting point”
loads before reductions are applied. Point source starting point loads project out for five years
beyond the initial loads from 1997-1998. MS4s and nonpoint source starting point loads utilize
loadings from projected 2008 land uses.1
The allocations do not include any required reductions in the load from atmospheric deposition
or natural background.
Achieving the TMDL in the LSJR Basin is contingent on reductions from the Middle St. Johns River
Basin. The Middle St. Johns and Ocklawaha basins are provided a single allocation in the
freshwater reach of the LSJR.
Wastewater plants below the head of tide were considered to have a share in the responsibility
for the overall load to the river. The water quality model was not designed to partition the nearfield impacts of individual facilities and was therefore not a good way to measure the influence
of individual sources.
The wasteload allocations for all domestic wastewater facilities in the marine portion are based
on the facilities meeting an equivalent TN concentration of approximately 5 mg/L.
Most of the point source reductions in the freshwater portion reflect a “committed” level of
reductions offered by the individual domestic and industrial waste facilities prior to development
of the LSJR TMDL. These committed reductions were sufficient to meet the TMDL requirement,
University of Florida Water Institute and UF/IFAS Extension, June 2012
41
Table 10. Assumptions made to determine starting loads and allocations as listed in the LSJR
BMAP (LSJR TMDL Executive Committee, 2008).
and additional Step 3 reductions were not required from those point sources.
The wasteload allocations to all industrial wastewater facilities in the marine portion are based
on the facilities making a reduction equivalent to the average percent reduction of domestic
facilities (approximately 49%), rather than industry-specific best available technology
economically achievable (BAT) levels because BAT for nutrients was not defined for these
industries.
The wasteload allocation originally attributed to the closed Smurfit-Jacksonville facility was
redistributed to the other industrial wastewater facilities in the marine section (Smurfit-Stone
Container and Anheuser Busch).
For facilities with effluent TN concentrations below 5 mg/L, an adjusted load was determined
that was based on the target TN concentration and their 2008-adjusted flow, thereby generating
a load higher than their starting loads (a “credit”). These facilities are allowed to trade these
credits, either within their jurisdiction or with other entities.
Technical Advisory Committee (ATAC) Report (February 2001). “Step 2” reductions were only
applied to nonpoint sources as they were considered to provide treatment approximately
equivalent to the treatment required for domestic and industrial wastewater facilities. The Step 2
reductions were based on the reduction from nonpoint sources expected from BMP
implementation and did not result in meeting the assimilative capacity, and additional (Step 3)
reductions were necessary. The same overall percentage reduction was required from point and
nonpoint sources for Step 3 reductions.
Counting load reductions associated with phase out of failing septic tank systems will be
reviewed on a case-by-case basis to determine the load amount that can be attributed to the
removal, accounting for the uncertainty associated with the estimate.
While it is recognized that nitrogen and phosphorus are present in the environment in different
forms of variable bioavailability, reductions do not specify these specific forms and the final
TMDL is expressed only in terms of TN and TP. Because TMDL reductions were only determined
for anthropogenic sources, which are high in bioavailable nutrient forms (for example, N and P in
domestic waste are 98% bioavailable), it was felt that it was not necessary to make this
distinction.
The LSJR TMDL is expressed as an annual load. While daily loads can be computed as defined in
the adopted TMDL, expression of the TMDL on a mass per day basis is for information purposes
only. The TMDLs to be implemented are those expressed on a mass per year basis, and effluent
limits for wastewater discharges will be based on the annual expression. While the loads for the
individual MS4s were calculated, the allocations to the MS4s are expressed as a percent
reduction rather than loads.
TMDLs will be developed for several of the tributaries to the LSJR, which were independently
verified as impaired for nutrients. While implementation of these TMDLs will be addressed in a
separate BMAP, it should be noted that these subsequent TMDLs may require additional
reductions in nutrient loading to achieve applicable water quality standards in these tributaries.
Certain non-structural BMPs and environmental education efforts were given provisional credit
for load reductions in this BMAP while additional research is being conducted to quantify their
University of Florida Water Institute and UF/IFAS Extension, June 2012
42
Table 10. Assumptions made to determine starting loads and allocations as listed in the LSJR
BMAP (LSJR TMDL Executive Committee, 2008).
effectiveness. These reduction estimates may change as the additional research results are
available. Activities that qualified for provisional credit included street sweeping, continuous
deflective separation (CDS) units, and second generation baffle boxes.
1
Although this text comes directly from the Lower St. Johns River Basin BMAP, we believe that t this should read urban nonpoint source starting
point loads utilize loadings from projected 2008 land uses. According to the SJRWMD (J. Hendrickson personal communication, 2011), 1995 land
use data updated with results from a 2000 land use survey of the TCAA was used for agricultural nonpoint source loads for the starting point loads
The ATAC recommended that agricultural and urban nonpoint sources should be expected to
implement BMPs to provide minimum levels of treatment comparable to domestic and industrial
wastewater point sources before point sources were required to implement additional reductions
since point source discharge is already significantly treated and controlled (FDEP, 2001). The ATAC
allocation process consists of a three-step approach:
Step 1. Require the amount of load reduction that would be achieved if 45% of all agricultural
operations implemented Best Management Practices (BMPs) to control nonpoint source
pollution, all urban areas implemented BMPs to control nonpoint source pollution to the
Maximum Extent Practicable (MEP), and all homes with septic tanks within the 100-yr
floodplain were hooked up to new sewers.
Step 2. If reductions in step 1 are not sufficient to meet TMDL, then require the additional load
reduction that would be achieved if 90% of all agricultural and urban areas and homes with
septic tanks within the 100-yr floodplain implemented practices specified in Step 1.
Step 3. If these reductions are still not enough, allocate additional reductions to all point and nonpoint
sources in increments of 10 percent until the TMDL is met (unless the loads from a source are
at background levels). Those that provide treatment beyond Best Available Technology (BAT)
are exempt from step 3.
The ATAC protocol was followed for the most part when determining allocations for the LSJR TMDL.
The exception was in regard to step two, where instead of the recommended 90% implementation,
100% of the urban and agricultural areas were expected to adopt BMPs.
As shown in Table 11, for agricultural lands in the TCAA discharging to freshwater portions of the LSJR,
the initial starting load was estimated at 683,540 lbs/yr of Total Nitrogen (TN) and 183,601 lbs/yr of
Total Phosphorus (TP). The allowable loading from agricultural land uses – while still meeting
protective water quality targets in freshwater portions of the LSJR – are 427,538 lbs/yr of TN and
156,143 lbs/yr of TP. The difference between the initial load and the allowable load is 256,001 lbs/yr
of TN and 27,458 lbs/yr of TP, or a required reduction of 37.5% TN and 15% TP. Implementation of
agricultural BMPs on 100% of agricultural lands was estimated to result in a 36% (246,074 lbs/yr)
reduction in TN and an 11% (20,196 lbs/yr) reduction in TP. This leaves 9,927 lbs/yr of TN and 7,262
lbs/year of TP as remaining load reduction required for agriculture not met by BMP implementation.
This residual amount is expected to be treated using Regional Stormwater Treatment systems (RSTs)
according to the BMAP.
University of Florida Water Institute and UF/IFAS Extension, June 2012
43
Table 11: Agricultural load reduction summary for freshwater reaches of the LSJR.
TN (lbs/year) TP (lbs/year)
Initial load from agricultural lands
683,540
183,601
Allowable loading from agriculture lands
427,538
156,143
Difference = What needs to be reduced from agricultural
256,001
27,458
land use loads
(37.5%)
(15%)
Assumed reduction by implementing agricultural BMPs on 246,074
20,196
100% of agricultural lands
(36%)
(11%)
Remaining load reduction to be achieved by. Regional
9,927
7,262
Stormwater Treatment systems
The BMAP based the starting point for TCAA agricultural loads on 2000 land use, but land use changes
since that time may have reduced the number of acres under production. In addition to the
conversion of agricultural lands to residential use, several dairies that were initially in the load
allocation calculations are no longer in operation. This trend in decreased acreage in agriculture is
expected to continue. Consequently, agricultural load estimates will need to be adjusted in the future
to account for this reduction in agricultural land area. The BMAP acknowledges this situation and
states that agricultural loads will be adjusted appropriately during the next cycle of the process.
However, even if agricultural land area decreases in future cycles of the TMDL, this will not change the
requirement that agricultural BMPs be implemented as the principal means by which agricultural land
uses reduce their contribution of overall nutrient loads to the LSJR. Furthermore, even after
agricultural lands are taken out of production there may be nutrient residuals in the soil, especially
phosphorus, that continue to leach out over time. These potential residual loads, sometimes called
“legacy loads” should be assessed and addressed, as necessary, in the next cycle of the TMDL/BMAP
process.
1.d.i. Is the required load reduction aggregated by total land use in the basin, or is it a required
reduction per unit acre of a specific land use?
The TMDL process allows for similar land use categories with comparable hydrologic and water quality
characteristics to be aggregated into representative classifications. Such was the case in the LSJR
BMAP, as agricultural operations were assigned load reduction responsibilities as a general land use
category. Thus, the allocations and required load reductions are not applied on a unit acre basis.
FDACS and the SJRWMD are expected to be partners with the agricultural community collectively to
assist in the implementation of practices designed to meet the agricultural load reduction allocation.
While a general load reduction target exists for the agriculture sector in the LSJR basin (37.5% TN, 15%
TP in the freshwater section, 67.4% TN in the marine section), individual load reductions have not been
assigned to specific agricultural operations. Under the FWRA, agricultural producers in a BMAP area
either must adopt BMPs to protect water quality or monitor their water quality to demonstrate
compliance with water quality standards. Producers who implement applicable FDACS BMPs receive a
“presumption of compliance” and have fulfilled their responsibilities. It should be noted that the BMPs
University of Florida Water Institute and UF/IFAS Extension, June 2012
44
upon which the TMDL reductions were based included limits on nitrogen applications and zero
phosphorus additions, elements that are now a point of contention with the growers.
Where nutrient load reductions from BMP implementation in the basin are not expected to achieve
the designated load reductions to meet the TMDL, regional treatment strategies have been undertaken
by the water management district and/or state agencies to further remove the target pollutant from
runoff. This is based on the premises that agricultural operations are fully implementing all applicable
BMPs, that the cost of further treatment is not economically feasible without funding assistance, and
that agriculture serves a critical public purpose. Under F.S. 403.067(7)(a)5, public works, including
capital funding is one of the programs identified as part of the TMDL implementation process. In
addition the ATAC report to the Governor and Legislature had a consensus that, “Public funds should
be used to provide any needed reductions in NPS loadings beyond that achieved by implementation of
approved BMPs because BMPs take economic constraints into account and further reductions by
individual landowners are likely not practical.” It should also be pointed out under F.S. 403.067(7)(d)
“Where water quality problems are detected despite the appropriate implementation, operation, and
maintenance of best management practices and other measures according to rules adopted under this
paragraph, the Department of Agriculture and Consumer Services shall institute a reevaluation of the
best management practice or other measure.”
1.d.ii. Was the feasibility of reducing nutrient loads and effectiveness of implementing agricultural
and urban BMPs considered when allocating load reductions?
It was assumed that FDACS-adopted agricultural BMPs are economically feasible. It was similarly
assumed that urban nonpoint sources would reduce loads to the Maximum Extent Possible (MEP)
using BMPs. One hundred percent implementation of BMPs was required for agriculture to cover their
“fair share” of load reductions. The remainder of the reduction required for agriculture is planned to
be achieved by regional projects.
The following BMP efficiencies were assumed in the LSJR BMAP calculations:
1) Urban stormwater: 30% Nitrogen Removal, 50% Phosphorus Removal (SWFWMD, 1990).
2) Agricultural stormwater: 36% Nitrogen Removal, 11% Phosphorus Removal (Livingston Way,
2001).
It should be noted that these efficiency estimates are uncertain and should be re-evaluated over time
as additional data are collected.
1.d.iii. What is the performance of the two existing RSTs?
As part of the BMAP process adopted by rule, several regional stormwater treatment systems (RSTs)
were to be constructed in an effort to make up for the difference between agricultural load reductions
(kg of nutrients per year) expected from the implementation of BMPs and the total load reductions
allocated to agriculture. As a result, the St. Johns River Water Management District constructed two
facilities, the Deep Creek West RST and the Edgefield RST. Both sites are operational, but only the
University of Florida Water Institute and UF/IFAS Extension, June 2012
45
Deep Creek West RST has water flow (cubic feet of water per second [cfs]) data available to calculate
load (Steinmetz and Livingston-Way, 2009; Steinmetz and Livingston-Way, 2011). Therefore, only the
Deep Creek West RST will be evaluated with regard to this question.
The Deep Creek West RST facility is located in St. Johns County near Hastings, Florida (Figure 30) and
has been operational since October 2006. The facility receives drainage from a 1,196-acre catchment
within the larger 38,928-acre Deep Creek Basin. The facility consists of a 15-acre wet detention pond
followed by a 38-acre constructed treatment wetland (Figure 30). Water entering the facility is
pumped into the wet detention pond from a 0.25 acre sump area, which is gravity fed by Canal 1 of the
Hastings Drainage District. The system is designed to capture up to 90% of the flow on Canal 1 with
peak flow rate pumped into the RST of 20 cfs and baseflows of 10 cfs.
A
B
Figure 30. Regional stormwater treatment areas (RSTs) in the Tri-County Agricultural Area (pink
outline in map). A) is the Deep Creek West RST and B) is the Edgefield RST (SJRWMD, 2012).
The target goals of the facility, as adopted by FDEP Secretarial order in the 2008 LSJR BMAP were to
reduce annual loads to the LSJR by 1,000 kg/yr total nitrogen and 818 kg/yr total phosphorus (LSJR
TMDL Executive Committee, 2008). Monitoring of ambient (non storm-flow) conditions began in 2008,
and the monitoring of storm-flow conditions began in 2009. Performance of the RST appears to vary
University of Florida Water Institute and UF/IFAS Extension, June 2012
46
considerably between ambient conditions (Table 12) and storm-flow conditions (Table 13), as well as
between growing and fallow season. When evaluating the overall performance of the RST to treat
loads, both ambient and storm flows are simply combined allowing for an overall assessment of RST
efficacy.
Combined annual load reduced by the facility during both ambient and storm-flow conditions is shown
in Table 14. Only 2009 and 2010 (through August) data are available (no storm-flow data is available in
2008). During this period, total nitrogen loads were reduced by 3625 kg (2009) and 2409 kg (2010),
and total phosphorus loads were reduced by 1668 kg (2009) and 974 kg (2010). This equates to a twoyear average (not including last four months of 2010) of 3017 kg total nitrogen and 1321 kg total
phosphorus, or a 302% and 162% reduction in total nitrogen and total phosphorus loads above the
BMAP performance target. Actual reduction rates would likely be higher if the months of SeptemberDecember 2010 were included in the calculation.
Table 12. Ambient loading per year into and out of Deep Creek West RST 2008-2010. Table from
Steinmetz and Livingston-Way (2011). Growing season is from January to May. Fallow season is from
June to December.
University of Florida Water Institute and UF/IFAS Extension, June 2012
47
Table 13. Stormflow event loading per year into and out of Deep Creek West RST, 2009-2010. Table
copied from Steinmetz and Livingston-Way (2011). Growing season is from January to May. Fallow
season is from June to December.
Table 14. Total mass removal of nitrogen and phosphorus per year at Deep Creek West RST, 20092010. Table copied from Steinmetz and Livingston-Way (2011). Growing season is from January to May.
Fallow season is from June to December.
University of Florida Water Institute and UF/IFAS Extension, June 2012
48
2. Were the assumptions, data, and models used to develop the loads for the LSJR consistent with
best professional practices for development of other TMDLs? Did the process make use of the best
information available? If not, how could the assumptions, data, models, and development process
be improved?
Verification of impairment in the Lower St. Johns River was based principally on monitoring data
collected between 1996 and 2003. These data indicated that chlorophyll-a concentrations and values
for the Trophic State Index were at levels that indicate impairment of the LSJR relative to Class III water
quality standards. The approach used to make this determination was in line with guidance provided
in the State of Florida’s Impaired Waters Rule and was supported by sufficient water quality
monitoring information to draw the conclusion that the Lower St. Johns River was not meeting its Class
III designated use.
Use of the PLSM is generally consistent with best practices for TMDL and BMAP development. The
PLSM is a statistical model that has been used to estimate watershed loadings at various locations in
Florida. The Hydrologic Simulation Program - Fortran (HSPF), Surface Water Management Model
(SWMM), and Watershed Assessment Model (WAM) (Borah and Bera, 2004; Graham et al., 2009) are
more process-oriented models that have also been recognized by EPA and FDEP as useful for
estimating watershed loads for TMDL development. However these models require much more data
for calibration and validation than the PLSM, and are not guaranteed to be unique or more accurate.
SJRWMD engineers and scientists examined both HSPF and WAM water quality prediction capability
for the LSJR TMDL, and ultimately concluded that the PLSM was more accurate, and had the capability
to predict the water variables needed for input to the Lower St. Johns River CE-QUAL-ICM application
that were unavailable in these other models.
At the time the PLSM was first developed and applied to the LSJR (during 1995-2000), it utilized the
readily available land use, soils, rainfall, streamflow, and water quality data to determine inputs and
estimate runoff and water quality coefficients. However, since the model was developed land use has
changed and additional rainfall, streamflow, and water quality data should be available. Thus, for the
next round of TMDL/BMAP development the land use in PLSM should be updated, and the runoff and
load predictions validated with new rainfall, streamflow, and water quality data to evaluate the
predictive reliability of the model, and to recalibrate the runoff and water quality coefficients if
necessary. In addition the sensitivity and uncertainty of PLSM predictions to model structure and
parameter uncertainty should be formally assessed in order to build agricultural stakeholder
confidence in the model and efficiently target model improvement efforts. If possible, additional fieldscale runoff and nutrient water quality data should be gathered to help refine land-use-specific runoff
and water quality coefficients.
Determination of the TMDL for the LSJR was the first developed for a large river system in Florida that
combined both freshwater and marine mixing zones. Due to the complexity of this interaction, an
appropriately complex modeling effort was used to evaluate the nutrient and biological response
dynamics of the system in an effort to best determine the TMDL that would be protective of Class III
waters designated use. The modeling effort applied was, and still is, one of the most sophisticated
University of Florida Water Institute and UF/IFAS Extension, June 2012
49
models available to predict the effects of estuarine eutrophication, and its results were confirmed with
a substantial suite of monitoring data collected from the LSJR. While there is uncertainty in the
estimates provided by the PLSM for the TCAA, available measured data and model estimates indicate
that agricultural practices have contributed to nutrient loads and decline in water quality in the LSJR.
The threshold chlorophyll-a concentrations used to set the TMDL were based on research findings from
monitoring data and studies conducted in the LSJR, and are expected to be protective of a wellbalanced aquatic flora and fauna. The application of these site-specific criteria rather than thresholds
based only on the IWR guidelines demonstrates the level of effort in this basin to use the best available
site-specific information.
Adopting Site Specific Alternative Criteria for dissolved oxygen levels in the marine section of the LSJR
is appropriate based on known natural variability and occasionally low levels of oxygen in estuarine
systems. By applying a “persistent exposure criterion” yet maintaining the minimum 4 mg/L dissolved
oxygen standard provided a means to better quantify acceptable dissolved oxygen exposure levels
between the range of 4 and 5 mg/L based on growth and recruitment of species of specific concern,
and yet not allow a potentially undesirable acute exposure level below 4 mg/L.
Although this TMDL was one of the first to be developed, and therefore did not have many documents
to compare to at the time, the approach to verifying impairment and establishing appropriate nutrient
thresholds established the precedent for TMDLs being developed now.
One additional aspect of the development of this TMDL was that there was a considerable amount of
high-quality monitoring and research data available to support both chlorophyll-a threshold
development and modeling efforts. Continued monitoring to confirm nutrient dynamics and biological
and dissolved oxygen response (as outlined in the TMDL) is highly recommended so that confidence in
modeled performance predictions of the system can be established and future TMDLs can be set
appropriately.
University of Florida Water Institute and UF/IFAS Extension, June 2012
50
References
Adamus, C.L., and M.L. Bergman. 1995. Estimating nonpoint source pollution loads with a GIS screening
model. Water Resources Bulletin 31(4):647-655.
Beasley, R.S. and A.B. Granillo. 1988. Sediment and water yields from managed forests on flat coastal plain sites.
Water Resources Bull. 24(2):361-366.
Bevin, K., 2006. A manifesto for the equifinality thesis. Journal of Hydrology. 320(1-2):18-36
Borah, D.K., and M. Bera. 2004. Watershed-scale hydrologic and nonpoint-source pollution models:
Review of applications. Trans. ASABE 47(3): 789‐803.
Bricker, S.B, C.G. Clement, D.E. Pirhalla, S.P. Orlando, and D.R.G. Garrow. 1999. National estuarine
eutrophication assessment: Effects of nutrient enrichment in the nation’s estuaries. Silver Spring,
Maryland: National Oceanic and Atmospheric Administration, National Ocean Service, Special
Projects Office and The National Centers for Coastal Ocean Science.
Burkholder, J.M., and H.B. Glasgow, Jr. 1997a. Pfiesteria piscicida and other Pfiesteria-like
dinoflagellates: Behavior, impacts, and environmental controls. Limnology and Oceanography.
42(5 part 2): 1052-1075
Burkholder, J.M., and H.B. Glasgow, Jr. 1997b. Trophic controls on stage transformations of a toxic
ambush-predator dinoflagellate. J. Euk. Microbiology. 44:200-205.
Chand, R., G.K. Hodlur, M.R. Prakash, N.C. Mondal, and V.S. Singh. 2005. Reliable natural recharge estimates
in granitic terrain, Curr. Sci., 88(5), 821–824.
Chow, V. T. (Editor), 1964. Handbook of Applied Hydrology. McGraw-Hill, New York, New York.
Cerco, C., and T. Cole. 1993. Three-dimensional eutrophication model of Chesapeake Bay. Journal of
Environmental Engineering. 119:1006-1025.
Florida Department of Environmental Protection (FDEP). 2001. A Report to the Governor and the
Legislature on the Allocation of Total Maximum Daily Loads in Florida. Bureau of Watershed
Management, Tallahassee, Florida.
FDEP. 2002. Basin status report for the Lower St. Johns River Basin. Tallahassee, Florida: Bureau of
Watershed Management.
FDEP. 2006. TMDL Protocol. Version 6.0. (Task Assignment 003.03/05-003). Accessed at:
http://www.dep.state.fl.us/water/tmdl/docs/TMDL_Protocol.pdf
Graham, W.D., Donigan, A.S., Muñoz-Carpena, R., Skaggs, W. and Shirmohammadi, A. 2009. Peer Review of the
Watershed Assessment Model (WAM). Final Panel Report. University of Florida.
University of Florida Water Institute and UF/IFAS Extension, June 2012
51
Hamrick, J.M. 1992. A three-dimensional environmental fluid dynamics computer code: Theoretical
and computational aspects. Special Report 317. Gloucester, Virginia: College of William and Mary,
Virginia Institute of Marine Science.
Harper, H.H. 1994. Stormwater Loading Rate Parameters for Central and South Florida. Environmental Research
and Design, Inc., Orlando, Florida.
Hendrickson, J., and J. Konwiniski. 1998. Seasonal nutrient import-export budgets for the Lower St.
Johns River, Florida. St. Johns River Water Management District.
Hendrickson, J., N. Trahan, E. Stecker, and Y. Ouyang. 2002. Estimation and assessment of the current
and historic external load of nitrogen, phosphorus and organic carbon to the Lower St. Johns River
for the fulfillment of TMDL and PLRG objectives. Palatka, Florida: St. Johns River Water
Management District, Department of Water Resources.
Hendrickson J., E.F. Lowe, D. Dobberfuhl, P. Sucsy, and D. Campbell. 2003. Characteristics of Accelerated
Eutrophication in the Lower St. Johns River Estuary and Recommended Targets to Achieve Water Quality
Goals for the Fulfillment of TMDL and PLRG Objectives. St. Johns River Water Management District, Palatka,
FL.
Hendrickson, J.C., N. Trahan, E. Gordon and Y. Ouyang. 2007. Estimating the relevance of organic carbon,
nitrogen, and phosphorus loads to a blackwater river. J. Amer. Water Res. Assoc. 43(1): 264-279.
Hendrickson, J., personal communication, November 2011.
Hollis, C.a., R.F. Fischer, and W.L. Pritchett. 1981. Effects of some silvicultural practices on soil-site properties in
the lower coastal plains. pp. 585..606, In: Forest Soils and Land Use, Proc. of the 5th N. Amer. For. Soils
Conf., C.T. Youngberg, ed. Forest and Wood Science, Colorado State Univ., Fort Collins.
Janicki, A., and G. Morrison. 2000. Developing a Trophic State Index for Florida’s estuaries. Florida
Department of Environmental Protection. St Petersburg, Florida.
Jordan, N.R., C. S. Slotterback, K.V. Cadieux, D.J. Mulla, J.O. Kim, D.G. Pitt and L. S. Olabisi 2011. TMDL
implementation in agricultural landscapes: A Communicative and Systemic Approach,
Environmental Management 48:1-12
Livingston-Way, P. 2001. Water Quality Monitoring and Assessment of Agricultural Best Management
Practices in the Tri-County Agricultural Area. Phase II Final Report. SJRWMD report submitted to
FDEP.
LSJR TMDL Executive Committee. 2008. Basin Management Action Plan for the Implementation of
Total Maximum Daily Loads for Nutrients Adopted by the Florida Department of Environmental
Protection for the Lower St. Johns River Basin Main Stem.
Madsen, H., G. Wilson and H. C. Ammentorp 2002. Comparison of different automated strategies for
calibration of rainfall-runoff models, Journal of Hydrology 261(1-4):48-59
University of Florida Water Institute and UF/IFAS Extension, June 2012
52
Magley, W. and D. Joyner 2008. TMDL Report: Total Maximum Dailiy Load for Nutrients for the Lower
st. Johns River. Watershed Assessmetn Section, bureau of Watershed Management, Florida
Department of Environmental Protextion. Tallahassee FL.
National Research Council (NRC). 2001. Assessing the TMDL Approach to Water Quality Management,
Water Science and Technology Board.
NRC. 2008. Public Participation in Environmental Assissment and Decision Making.
Pahl-Wostl 2009. A Conceptual framework for analyzing adaptive capacity and multi-level learning
processes in resource governance regimes, Global Environmental Change 19:354-365
Pearl, H.W., M.F. Piehler, W.W. Carmichael, J. Dyble, P. H. Moisander, J. Leonard and A. Waggener.
2002. Phytoplankton and zooplankton in the St. John’s River System: Factors affecting community
structure and function. Year 2 Final Report to St. Johns River Water Management District under
contract No. SD154RA.
Phlips, E.J., M. Cichra, F. J. Aldridge, J. Jembeck, J. Hendrickson, and R. Brody. 2000. Light availability
and variations in phytoplankton standing crops in a nutrient-rich blackwater river. Limnology &
Oceanography 45:(4) 916-929.
Phlips, E.J. and M. Cichra. 2002. Spatial and temporal patterns in the plankton community of the
Lower St. Johns River. Final Report to the St. Johns River Water Management District No. 97W165.
Riekerk, H. 1983. Impacts of silviculture on flatwods runoff, water quality, and nutrient budgets. Water
Resources Bulletin 19(1):73-79.
Riekerk, H. 1989. Influence of silvicultural practices on the hydrology of pine flatwoods in Florida.
Water Resources Res. 25(4):713-719.
Southwest Florida Water Management District (SWFWMD), 1990. Urban Stormwater Analysis and
Improvements for the Tampa Bay Watershed. Southwest Florida Water Management District, Brooksville,
Florida.
Speir, W.H., W.C. Mills and J.C. Stephens. 1969. Hydrology of three experimental watersheds in southern Florida.
Agricultural Research Service Bulletin # ARS 41152, USDA. 50 pp.
St. Johns Regional Water Management District (SJRWMD). 2012. Regional stormwater treatment areas.
Obtained on 1/20/2012 from weblink: http://www.sjrwmd.com/lowerstjohnsriver/regionalstormwater.html
SJRWMD. 2011. 2009 Land cover and land use, St. Johns River Water Management District. GIS dataset
obtained on 1/4/2012 from weblink: http://www.sjrwmd.com/gisdevelopment/docs/themes.html
SJRWMD. 2002. SJRWMD Land Use and Land Cover (2000). GIS dataset obtained on 1/3/2012 and 1/25/2012
from weblink: http://www.sjrwmd.com/gisdevelopment/docs/themes.html
University of Florida Water Institute and UF/IFAS Extension, June 2012
53
SJRWMD. 1999. SJRWMD Land Use and Land Cover (1995). GIS dataset obtained on 1/3/2012 from weblink:
http://www.sjrwmd.com/gisdevelopment/docs/themes.html
Steinmetz, A. and P. Livingston-Way. 2009. Edgefield Regional Stormwater Treatment (RST) Facility,
Tri-County Agricultural Area, Water Quality Draft Summary December 2009. St. Johns River Water
Management District.
Steinmetz, A. and P. Livingston-Way. 2011. Deep Creek West Regional Stormwater Treatment (RST)
Facility, Tri-County Agricultural Area, St. Johns County Water Quality Report 2008-2010. Draft
Report. St. Johns River Water Management District.
Sucsy, P.V., and F.W. Morris. 2002. Calibration of a three-dimensional circulation and mixing model of
the Lower St. Johns River. Technical Memorandum. St. Johns River Water Management District.
Palatka, Florida.
SWET. 1994. Tri-county Agricultural Best Management Practices Study, Final Report, Phase I. Soil and Water
Engineering Technology, Inc., Gainesville, FL. 126 pp.
United States Environmental Protection Agency (USEPA). 2001. Nutrient criteria technical guidance
manual: Estuarine and coastal marine waters. EPA 822-B01-003. Office of Water, Washington, DC.
University of Florida Water Institute and UF/IFAS Extension, June 2012
54
Appendix A - Documents Reviewed
Primary documents:
1. Magley W. and D. Joyner 2008. Total Maximum Daily Load for Nutrients for the Lower St. Johns
River, Florida Department of Environmental Protection, Tallahassee FL. Link to pdf
2. Lower St. Johns River TMDL Executive Committee. 2008. Basin Management Action Plan for the
Lower St. Johns River Basin Main Stem. Link to pdf
3. Livingston-Way, P. 2001. Water Quality Monitoring and Assessment of Agricultural Best
Management Practices in the Tri-County Agricultural Area. Phase II Final Report. SJRWMD
report submitted to FDEP. Link to pdf
Support documents:
4. Adamus, C.L., and M.L. Bergman. 1995. Estimating nonpoint source pollution loads with a GIS
screening model. Water Resources Bulletin 31(4):647-655. Link
5. Hendrickson, J., and J. Konwiniski. 1998. Seasonal nutrient import-export budgets for the Lower
St. Johns River, Florida. St. Johns River Water Management District. Link to pdf
6. Hendrickson J., E.F. Lowe, D. Dobberfuhl, P. Sucsy, and D. Campbell. 2003. Characteristics of
Accelerated Eutrophication in the Lower St. Johns River Estuary and Recommended Targets to
Achieve Water Quality Goals for the Fulfillment of TMDL and PLRG Objectives. St. Johns Rivers
Water Management District, Palatka, FL. Link to pdf
7. Hendrickson, J., N. Trahan, E. Stecker, and Y. Ouyang. 2002. Estimation and assessment of the
current and historic external load of nitrogen, phosphorus and organic carbon to the Lower St.
Johns River for the fulfillment of TMDL and PLRG objectives. Palatka, Florida: St. Johns River
Water Management District, Department of Water Resources. Link to pdf
8. Sucsy, P., and J. Hendrickson. 2004. Calculation of nutrient reduction goals for the Lower St.
Johns River by application of CE-QUAL-ICM, a mechanistic water quality model. Palatka, Florida:
St. Johns River Water Management District, Department of Water Resources. Link to pdf
Additional Documents
9. Magley W. and D. Joyner 2004. Total Maximum Daily Load for Nutrients for the Lower St. Johns
River, Florida Department of Environmental Protection, Tallahassee FL. Link to pdf
10. Steinmetz A. and P. Livingston Way 2009. Deep Creek West Regional Stormwater Treatment
Facility: Treatment Wetlands Design and Monitoring – Tri-County Agricultural Area, Lower St.
Johns River Basin. Technical Publication SJ2009-6. Link to pdf
11. Steinmetz A. and P. Livingston Way 2009. Edgefield Regional Stormwater Treatment (RST)
Facility Tri-County Agricultural Area St. Johns River Water Management District Water Quality
Draft Summary, December 2009 (DRAFT). Link to pdf
12. Steinmetz A. and P. Livingston Way 2011. Deep Creek West Regional Stormwater Treatment
(RST) Facility Tri-County Agricultural Area, St. Johns County Water Quality Report 2008-2010.
(DRAFT) Link to pdf
University of Florida Water Institute and UF/IFAS Extension, June 2012
55
Appendix B - Grower Questions from November 1, 2011 Listening Session
The following questions were collected from growers during a meeting held at the Putnam County
Agricultural Center on November 1, 2011.
Questions raised by growers during the listening session
The following questions were submitted in writing.

How have loads been determined off of agricultural land when over 50% of the land draining to
the canals is silviculture lands or homes?

How have loads changed from early 1990s to 2010 due to real estate changes? (i.e.,
development)

Nutrient loading in rainfall is at times higher than farm runoff. Most rainfall is captured on sites
not discharged. Can growers get credit for cleaning rainwater?

Where did the nutrient load data originate?

When were agricultural loads established to create the TMDL?

What is the Margin of Safety (MOS) for TMDLs in the LSJR? How does this MOS effect the
portions of the TMDL related to agriculture? For instance, how many pounds of N & P were
allocated to agriculture? Was this an equitable allocation among the land uses?

Agriculture is identified as a contributor to the nutrient load. This may be true a portion of the
year (maybe 1 to 2 months). Are we credited for sequestering nutrients (10-11 months)?

How is the source of impairment determined?
The following questions / comments were heard during the general discussion (somewhat
paraphrased).

Because so little water appears to run off the farms, why do flow numbers seem so high?

Isn’t the rain concentration higher than runoff concentration?

Big ditches run through some of the agricultural lands and receive water from off site. How do
you tease apart nutrients from a farm from neighboring land or other offsite sources?

Assimilative capacity changes depending on rainfall, can you explain?

We would like to see data and graphs in forms we can understand.

What data was used in models to make the TMDL?

How do they monitor and keep up with how things are going?

When will LSJR basin TMDL be revised? What will be done? Do they redo the allocation
process?

Do they take into consideration land use changes? The model may need to reallocate the load
to urban from agriculture.
University of Florida Water Institute and UF/IFAS Extension, June 2012
56

Coal Plant – how much N is it contributing (from atmosphere)? There are new scrubbers on it
now…

Benchmark farms: can someone explain this data and how it was used in TMDL/BMAP?

Can you explain the fertilizer data in the BMP study?

Why not put more macrophytes in the river, like water hyacinths to improve river water
quality? Then harvest for compost to help reduce loads to the water column.

Why did we not get compensation for farming for 3 years without P? (comment related to
fertilizer study where they were asked to reduce P application?)

How were the coefficients for the agricultural loads determined? What did land use look like
and how did it change?

Can you show us data that shows that agriculture is contributing more nutrients in runoff in a
way we can understand? E.g. at the mouth of Deep Creek there are multiple contributing
factors. What are the assumptions? When and where sampled? How do we know about the
quality the data? How do they attribute what quantity of nutrients to agriculture?

Can they credit agriculture for reducing acreage under production just as water treatment gets
credit for implementing a technology?

What is their (agriculture’s) load right now?

Doesn’t our land hold the nutrients when we have the land planted? (crop and summer cover
crop)

How will the next round of the BMAP process be conducted? Will nutrient loads be determined
using the same assumptions, along with the same model and coefficients? Will different land
use maps be considered?

Is there a consistent number assigned to each Margin of Safety? Is it subjective?

Are we over-protecting the system?

How effective are RSTs? Are they working? Is the cost for RSTs required as part of the BMAP the
best allocation of money or could it be better spent on agriculture lands to cost share
practices?
University of Florida Water Institute and UF/IFAS Extension, June 2012
57