A WATER AND NUTRIENT BUDGET FOR CADDO LAKE

 

 

 

Report Prepared for:

 

Texas Commission on Environmental Quality

12100 Park 35 Circle

Austin, Texas  78753

 

And

 

Northeast Texas Municipal Water District

PO Box 955

Hughes Springs, Texas  75656

 

 

 

Report Prepared by:

 

HDR Engineering, Inc.

4401 West Gate Blvd., Suite 400

Austin, Texas  78745

 

 

 

 

 

November 9, 2007

 

 

TABLE OF CONTENTS

 

Executive Summary	Page 1
1.0 Introduction	Page 2
1.1 Project Objectives	Page 2
1.2 Project Background and Significance	Page 2
2.0 Methods and Materials	Page 14
2.1 Data Collection and Analysis	Page 14
2.2 Modeling	Page 16
3.0 Results and Discussion	Page 18

 

List of Figures

Figure 1-1 Caddo Lake Watersheds	Page 4
Figure 1-2 Caddo Lake Water Quality Monitoring Stations	Page 5
Figure 1-3  Daily Average Streamflow in Big Cypress Creek at USGS Gage             07346000	Page 8
Figure 1-4  Combined Gaged Daily Average Inflows to Caddo Lake Since Closure of Ferrells Bridge Dam	Page 8
Figure 1-5 Nutrient Sampling Locations in the Ungaged Caddo Lake Watersheds	Page 11
Figure 2-1  Caddo Lake Water and Nutrient Budget Special Study Monitoring Stations	Page 15
Figure 2-2 Input Streamflows and Concentrations Used in the LOADEST Model	Page 20
Figure 3-1 Average Daily Inflows to and Outflows from Caddo Lake, July 2006-June 2007	Page 19
Figure 3-2 Cumulative Caddo Lake Inflow and Outflow	Page 19
Figure 3-3 Total Nitrogen Concentrations and Streamflows Measured in Black Cypress Bayou, July 2006-June 2007	Page 22
Figure 3-4 Modeled Annual Total Phosphorus Load Entering Caddo Lake 1980-2006	Page 25
Figure 3-5 Annual Total Phosphorus Loads in Caddo Lake Tributaries, 1980-2006	Page 26
Figure 3-6 Average Monthly Total Phosphorus Caddo Lake Tributary Inflows	Page 26
List of Figures Continued
Figure 3-7 Annual Average and Median Water Surface Elevations of Caddo Lake, 1980-2006	Page 28
Figure 3-8 Caddo Lake Annual Turnover and Areal Phosphorus Loads, 1980-2006	Page 28
Figure 3-9 Annual Inflows and Predicted Annual Average Total Phosphorus Concentrations, 1980-2006	Page 29

 

List of Tables

Table 1-1  Historical Total Phosphorus Loading of Caddo Lake	Page 10
Table 2-1 Watersheds Modeled Using LOADEST	Page 16
Table 3-1   Average, Median1 and Range of Field Measured Parameters in the Caddo Lake Nutrient Study, July 2006 Through May, 2007	Page 20
Table 3-2  Average, Median1 and Range of Water Chemistry Results From Samples Collected in the Caddo Lake Nutrient Study, July 2006 Through May, 2007	Page 21
Table 3-4  Nitrogen Phosphorus Ratios (mg-at/l) at Nutrient study Stations July 2006-2007	Page 24

 

List of Appendices

 

Appendix A    Water Quality Sampling Results	Page 31
Appendix B    Cypress Creek Basin Clean Rivers Program QAPP Appendix G, Appendix G Revision 1, and Amendment 1 to Appendix G	Page 44
Appendix C    LOADEST FILES – in Caddo Nutrient.pdf file only
LOADEST Output Files	Page 116

Executive Summary

 

Intensive water quality monitoring was conducted from July, 2006 through May 2007 at locations in each of the major tributaries of Caddo Lake, and at Caddo Lake Dam to document ambient nutrient concentrations entering and exiting the lake.  This information, together with USGS stream gage data and water quality data from TCEQ’s  Surface Water Quality Monitoring Information System (SWQMIS) database was used as input to a water quality model, LOADEST, to develop a time series of nutrient loading into Caddo Lake.  That information was used as input to the Vollenweider model, a regression model developed from a large data base of temperate zone lakes and reservoirs to predict lakewide phosphorus concentrations and assess trophic condition.

 

Total phosphorus loading into Caddo Lake was found to be at a level suggestive of eutrophic conditions.  Caddo Lake was found to be experiencing annual total phosphorus loads about one third of those seen in the TMDL study of the nearby Lake O’ the Pines, while it is exhibiting lakewide phosphorus concentrations nearly 2/3 of those in Lake O’ the Pines.  If natural levels of phosphorus loading in the Caddo Lake watershed are similar to that in the Lake O’ the Pines watershed, the present loading rate into Caddo Lake is about three times as great as it would be without human occupation of the Cypress Creek Basin.

 

1.0 Introduction

 

1.1 Project Objectives

 

The Caddo Lake Water and Nutrient Budget Special Study described here has been developed to determine the quantities, seasonal patterns and source watersheds of nutrients (nitrogen and phosphorus) presently being supplied to plant and microbial populations in Caddo Lake over an annual cycle.  This information will be used to support the development of a nutrient budget to assess the trophic status (i.e., stage of reservoir eutrophication or nutrient enrichment) of Caddo Lake in comparison with similar systems within the Cypress Creek Basin and the surrounding area.

The special study objectives are as follows:

 

·        Determine the volume and seasonal patterns of water delivery to the Caddo Lake;

·        Measure the quantities of nutrients entering Caddo Lake from each of the major tributary watersheds;

·        Measure the amounts of water and nutrients lost from Caddo Lake through releases and spills from Caddo Lake Dam;

·        Identify the watersheds most influential in driving hydrologic and biological processes in Caddo Lake;

·        Estimate the annual gain or loss of phosphorous and nitrogen in Caddo Lake;

·        Assess the trophic status of Caddo Lake with respect to nutrient loading now being experienced and compare with similar water bodies;

·        Placement of one to three stage recorders on James Bayou (TCEQ Station No. 10319) according to manufacturer’s installation recommendations and similar to recent USGS equipment installation at four locations on Big Cypress Creek to obtain streamflow information from the last major, ungaged watershed tributary to Caddo Lake; and

·        Directly assess the nutrient output of Lake O’ the Pines and to evaluate gain/loss of water and nutrients between Ferrells Bridge Dam and the new USGS gage station (07346080) on Big Cypress Creek upstream of State Highway 43 near Karnack, Texas.

 

This study was funded by the Texas Commission on Environmental Quality (TCEQ) as part of the Cypress Creek Basin Clean Rivers Program, which is administered and directed by the Northeast Texas Municipal Water District (NETMWD).

 

1.2 Project Background and Significance

 

Big Cypress Creek is impounded to form Caddo Lake near Mooringsport, Louisiana.  Caddo Lake Dam was completed in 1914 just above the confluence of Big Cypress Creek and Black Bayou to form Twelve Mile Bayou, a tributary of the Red River (Figure 1-1).  Dam construction was initiated to prevent drainage of the lake and the loss of navigation between Mooringsport and Jefferson, Texas following the clearance of log jams on Twelve Mile Bayou and the Red River.[1]  The present Caddo Lake Dam, completed as a replacement of the original dam in 1971, consists of an earthen embankment with an uncontrolled spillway at an elevation of 51.36 meters above mean sea level (msl).  The spillway structure was designed to allow future crest elevation increases to permit future enlargement of lake storage capacity.  Caddo Lake has a total surface area of approximately 108.4 km² of which the Texas portion includes approximately 56.6 km² in Harrison and Marion Counties.  It has been classified as a Resource Category 1 Habitat by US Fish and Wildlife Service, and the upper portion (in Texas) is designated a “Wetland of International Importance” by the Ramsar Convention, an intergovernmental treaty covering international conservation and management of wetland habitats.  Although shallow, the Louisiana portion of Caddo Lake consists primarily of open water and exhibits significant shoreline residential development (Figure 1-2).  It is also the site of the world’s first offshore oilfield (1911), which continued into the 1970’s with the most extensive well development along the Texas-Louisiana state line.[2]  Texas Parks and Wildlife Department owns and manages the portion of the lake designated as Ramsar Wetland.

 

Caddo Lake encompasses shallow, open water habitats, marginal habitats, and in the transitional area of the upper reaches of the lake, riverine and backwater habitats.  Open waters, which are represented in the TCEQ monitoring network by the mid-lake Station 10283 near uncertain, are typically on the order of 2 meters deep and relatively free of aquatic vegetation other than scattered bald cypress trees.  Marginal habitats surrounding the main lake basin are characterized by varying densities of bald cypress and generally exhibit depths of less than 1.5 meters and more or less dense stands of aquatic vegetation. Marginal habitats are represented by stations 10285, 10286, 15249, and 15275.  Riverine habitats are located in the channel of Big Cypress Creek, are typically more than two meters deep, relatively free of aquatic vegetation and experience nearly constant flowing water (i.e., stations 15021 and 15248).  Backwater habitats are located in the deltaic complex of the upper lake where the natural levees of Big Cypress Creek have partially isolated the large embayments of Clinton and Carter lakes.  These habitats are shallow (less than 2 meters), receive varying inflows from Big Cypress and Kitchen Creeks and typically support dense stands of bald cypress and aquatic vegetation.  Backwater habitats are most similar to those on the lake margins, differing primarily in their relative isolation from the main basin.  Station 10294 in Carter Lake is most typical of this environment, while Station 14236 is more similar to the marginal habitats.

 

In 2000, portions of Caddo Lake, particularly marginal habitats in embayments at the mouths of direct tributaries, were identified in the Texas Water Quality Inventory and 303(d) list as exhibiting depressed dissolved oxygen concentrations.[3]    Section 303(d) of the federal Clean Water Act requires that States periodically identify waters that do not or are not expected to meet applicable water quality standards, and to include them on a list of

 

 

 

 

USGS 07346280

10286 Mouth of Harrison Bayou

15249 Junction of Boat Lanes 1K & 1L

15275 Goose Prairie

14236 Devil’s Elbow

Louisiana

Texas

10283 Caddo Mid Lake

Figure 1-2 Caddo Lake Water Quality Monitoring Stations

Impaired waters, the 303(d) list (40 CFR subsections 130.7).  Lake O’ the Pines, impounded by Ferrill’s Bridge Dam on Big Cypress Creek about 50 stream miles above Caddo Lake was also put on the 2000 303(d) list for depressed dissolved oxygen concentrations in its upper reaches.  A study to determine the nature and source of the problem initiated at that time determined that excessive levels of phosphorus, a plant nutrient, originating from wastewater treatment plant discharges and from many non-point sources as a result of agricultural practices, was stimulating high levels of photosynthesis and respiration in the lake, resulting in large daily (diel) changes in dissolved oxygen concentration.[4]  During critical periods, primarily summer, these diel changes were sufficient to drive oxygen levels below the standards set for the segment.  The level of phosphorus input to Lake O’ the Pines that would result in the maintenance of dissolved oxygen standards, and the necessary amounts of reduction in phosphorus entering the lake from point and nonpoint sources in the watershed required to reach the “Total Maximum Daily Load” (TMDL) was established and adopted by TCEQ in April, 2006.[5]

 

This study was initiated to assess the sources and extent of nutrient input (loading) into Caddo Lake and to begin to evaluate the role(s) played by those nutrients in maintaining the lakes’ biological communities, particularly the potential relationship between nutrient loading and depressed dissolved oxygen levels.  Of the primary plant nutrients, nitrogen and phosphorus, the latter is the focus of this study for practical reasons.  First, phosphorus is the primary suspect in freshwater situations where excessive plant growth may be a problem, and second, nitrogen occurs in many forms, all of which undergo both chemical and biological transformations, and can enter and leave a water body directly from the atmosphere as a result of biological processes.  Those characteristics make nitrogen “bookkeeping” difficult and require methods and equipment beyond the scope of this study. 

 

The watershed of Caddo Lake consists of four major subwatersheds: Big Cypress Creek (also impounded to create Lake O’ the Pines, Lake Bob Sandlin, and Lake Cypress Springs), Little Cypress Creek, Black Cypress Bayou and James Bayou, (Figure 1-1).  Little Cypress Creek and Black Cypress Bayou are tributaries of Big Cypress Creek, joining that stream between Jefferson and the headwaters of Caddo Lake.  The first three of these subwatersheds have United States Geological Survey (USGS) streamflow gaging stations located near Caddo Lake, but James Bayou, which enters Caddo Lake through a large embayment on its northern shore in Louisiana, does not.  These four subwatersheds provide most of the inflows to the lake, but numerous, small creeks, the largest of which are Kitchen Creek and Harrison Bayou, enter the main body of the lake directly.  The total watershed areas of the three gaged streams drain 5418 km², or about 77.6% of Caddo Lake’s 6977 km² watershed.  The largest of these streams, Big Cypress Creek, accounts for 2468 km² (35.4%) of the Caddo Lake Watershed.  James Bayou and the minor drainages respectively account for 875 and 684 km² (12.6% and 9.8%) of the total drainage area.  However, because the stream gages are located substantial distances upstream of Caddo Lake, the drainage areas from which flows are measured are actually smaller, the gaged area being 4895 km², or 90.3% of the three major watersheds.

 

Figure 1-2 summarizes the daily average streamflows during the period of record (1924-2007) in Big Cypress Creek at USGS Gauge 07346000.  This gauge is located near Ferrells Bridge Dam which impounds Lake O’ the Pines, upstream of Jefferson, Texas.  The relatively low, uniform flows in Big Cypress Creek following 1979 reflect the presence of the dam, which has a maximum controlled release rate of about 85 cubic meters per second (cms).  Although the range and variance of the Big Cypress Creek discharge is diminished substantially by Lake O’ the Pines, the average annual discharge was not found to be significantly different (Students T, unequal variance, P<0.05) during the uncontrolled (before 1959) and controlled (after 1979) periods, based on an analysis made for the 2003 Cypress Creek Basin Summary Report  However, inclusion of more recent data (through May 2007) results in a substantially lower average flow during the controlled period that is statistically less than the flows recorded prior to dam construction, illustrating the importance of both reservoir operations and climatic conditions.

 

Streamflow in Black Cypress Bayou, which joins Big Cypress Creek downstream from the City of Jefferson, has been monitored near its mouth at USGS Gage 07346045 since October, 1968.  Little Cypress Creek enters Big Cypress Creek below both the City of Jefferson and the confluence with Black Cypress Bayou.  Streamflow has been monitored near the mouth of Little Cypress Creek at USGS Gage 07346070 since 1946.  Figure 1-3 shows the combined daily average inflow from all three gaged streams during their common periods of record (October, 1979 through September, 2004).  Even with the flow regulation exerted by Lake O’ the Pines, large seasonal inflow differences occur annually, resulting in continuing winter-spring wet periods and a summer low flow season.

 

Based on the sizes of their drainage basins relative to Black Cypress Bayou (the gaged watershed closest and most similar in terms of average rainfall, agricultural development and vegetation), James Bayou, the direct tributaries, and the ungaged portions of the major watersheds exhibited a total average inflow to Caddo Lake (October 1979-May 2007) of 62.354 cms, or an average annual volume of 1,967,742,590 m³/yr.  Assuming a volume on the order of 156,641,800 m³ at the spillway elevation of 51.36 meters msl, results in an average annual turnover for the period of record (annual inflow/lake volume), or lake volume replacement of 12.6/year.

 

Average total phosphorus and nitrogen concentrations (and other parameters) are available for the lowermost stations of most of the streams tributary to Caddo Lake.  This data, collected primarily by TCEQ Region 5 staff and the Cypress Creek Basin Clean Rivers Program from the mid-1990s to present, is archived in the Surface Water Quality Monitoring Information System (SWQMIS) database maintained by TCEQ and was used to obtain the average nutrient concentrations listed in Table 1-1.  Relative to the daily average streamflow records from the USGS gages, the data is relatively sparse in time; Big Cypress, Little Cypress, and Black Cypress Creeks being represented by quarterly

 

 

 

Figure 1-4 Combined Gaged Daily Average Inflows to Caddo Lake Since Closure of Ferrells Bridge Dam

 

 

records, while James Bayou and the minor streams generally have data covering much shorter periods.

 

A first-order estimate of nutrient loading into Caddo Lake can be made by using the average annual streamflows from each of the three gaged drainages and the respective average total phosphorus concentrations measured at each stream gage. In Table 1-1, which summarizes this information, flows for James Bayou and the watersheds below the USGS gages (the lower part of TCEQ Segment 0402 and Segment 0401), were estimated

from Black Cypress Bayou flows adjusted to reflect the ungaged drainage areas (Figure 1-4).

 

The assignment of a total phosphorus concentration to the ungaged drainages is problematic since the highest values are averaged from relatively few measurements.  Alternatively, weighting the results from the more frequently sampled locations by averaging overall values gives a total phosphorus concentration of 0.132 g/m³, a somewhat lower annual phosphorus load of 55,836 kg/yr from the lower 0402/0401 watersheds, and a total Caddo Lake load of 202,396 kg/year.  Either estimate implies a relatively high areal loading rate, 1866-2170 mg/m² of lake surface.  By way of comparison, the raw areal loading rate into Lake O’ the Pines developed during the TMDL studies of that lake was 2554 mg/m².

 

However, other factors, primarily lake depth and annual flushing rate, are important in determining the effective loading of the lake.  To estimate expected phosphorus concentrations in Caddo Lake from this level of loading, a simple model based on data from large numbers of lakes and reservoirs developed by Vollenweider can be used:[6]

 

 

TP (mg/m³) =    (mgTP/m²*year)

                               Z (r + s)

 

Z = average depth (reservoir volume/reservoir area, 1.43 m),[7]

r = flushing rate (reservoir volume/annual inflow, 12.6), and

s = sedimentation coefficient (10/Z)

 

TP_low =   ______1866_______

              1.43(12.6 + 10/1.43)

         = 66.60 mg/m³  or 0.067 mg/l

 

TP_high =   _____2170______

              1.43(12.6 + 10/1.43)

          = 77.45 mg/m³  or 0.077 mg/l

 

Table 1-1

Historical Total Phosphorus Loading of Caddo Lake

 

 

 

 

 

 

 

This model is intended to predict open water average concentrations in a completely mixed basin.  As noted above, average total phosphorus concentrations exceeding 0.05 mg/l in open lake waters are generally considered indicative of eutrophic conditions.  Application of the same model to Lake O’ the Pines predicted a lake average total phosphorus concentration of 0.140 mg/l,[8]  while observed lakewide phosphorus concentrations averaged 0.114 mg/l, and values exceeding 0.200 mg/l were common in the upper, vegetated shallows.

 

With respect to measured nutrient levels within Caddo Lake, total phosphorus concentration at TCEQ Caddo Mid-Lake Station 10283 (1993-2006, n=60) averaged 0.078 mg/l, and 0.075 mg/l in open boat lanes (Station 15249, n=26).  Total phosphorus data collected by the U.S. Geological Survey (USGS) during 1974-75 and 1991-92 (n=15) in open water near Caddo Lake Dam averaged 0.06 mg/l.  However, substantially higher values have been measured in vegetated shallows and embayments and at the mouths of tributaries, presumably reflecting local deposition, recycling, and limited horizontal mixing in the lake:

 

Devil’s Elbow, Station 14236 (n=8), 0.133 mg/l

Goose Prairie, Station 15275 (n=5), 0.229 mg/l

Mouth of Harrison Bayou, Station 10286 (n=6), 0.233 mg/l

 

Darville et al reported lakewide average total phosphorus levels to average 0.126 mg/l, noted substantially higher nutrient levels in the wetlands than in the open waters of Caddo Lake, and also judged the lake to be eutrophic.[9]  The TMDL studies in Lake O’ the Pines suggested that lakewide total phosphorus concentrations greater than 0.07 mg/l significantly increased the probability of occurrence of algal blooms exhibiting chlorophyll a concentrations exceeding the state screening level (0.021 mg/l).  That average was also assumed to include the higher phosphorus concentrations present in the vegetated shallows where inflowing nutrients are first deposited, supporting intense biological activity in sediments, higher plants and the attached community (periphyton). 

 

That nutrient and sediment should accumulate in the portions of the lake that receive the bulk of inflowing water, with its load of dissolved and suspended materials, and that eutrophic conditions should result are not surprising, nor do these conditions necessarily warrant action.  Such eutrophic conditions are a natural and largely unavoidable feature of the aging of a lake or reservoir, and Caddo Lake is old relative to other regional, lentic water bodies.  The central question here is weather the natural process of eutrophication is being significantly accelerated by human activities in the watershed.

 

The purpose of this study is to further our understanding of the present condition and potential future of Caddo Lake by examining nutrient loading in greater detail.  By collecting water quality samples with greater frequency at additional locations, important information on nutrient loading and export from the lake, and the relationship of nutrient concentrations to (for example) ambient streamflow, season or antecedent conditions, and nutrient transport during high flows can be developed that will allow much better use of the historical data.  This will permit better overall loading estimates to be made, and initiate steps to begin to examine the effects of annual variations in loads, the importance of wet and dry years, and the locations of the most important, or potentially most easily controlled sources of nutrients.

 

2.0 Methods and Materials

 

2.1 Data Collection and Analysis

 

Water samples were collected for nutrient analysis from each of the four major tributaries to Caddo Lake, and from the uncontrolled spillway outlet of Caddo Lake in Louisiana.  Water sampling stations, which are designated by their TCEQ station numbers in Figure 2-1, were located adjacent to the stream gaging stations on three of the tributaries.  Daily average streamflow data was obtained from long-term stream gages maintained by the USGS on Big Cypress Creek at Ferrells Bridge Dam (07346000), and at the US 59 crossings of Black Cypress (07346070) and Little Cypress (07346080) Creeks that are short distances above the confluence of the three streams.  Discharge from Caddo Lake, and changes in lake volume, were calculated from daily water surface elevations available from the web site rivergages.com and the spillway rating curve obtained from the Vicksburg District US Army Corps of Engineers.[10]

 

Streamflows on James Bayou, the ungaged areas below the three USGS gages, and the direct tributaries to Caddo Lake were estimated by adjusting the Black Cypress Creek data for drainage area.  An attempt to install pressure transducers in James Bayou to directly gage streamflow was abandoned.  The sampling site at CR 3312 proved unsuitable due to multiple channels, numerous obstructions and backwater effects from downstream controls.  The James Bayou assessment was conducted during a site visit by an HDR Engineering, Inc. hydrologist, and supported by a HEC-RAS model of expected water surface elevations based on plan and profile drawings of the recently constructed bridge at that location obtained from TxDOT.  The result of the assessment indicated that drainage area adjustment was a more practical and potentially more accurate method than direct gaging at that location.

 

In 2006 the USGS also installed an additional stream gage in Big Cypress Creek below this confluence to better monitor inflows of water, nutrients and other materials to Caddo Lake.  However, rating this gage has proved difficult for reasons similar to those encountered on James Bayou, therefore Big Cypress Creek flows originating below the gage at Lake O’ the Pines, and those of the minor drainages were also estimated by drainage area adjustments of the gaged discharges of adjacent streams.

 

Water samples were collected monthly when “low flow” conditions prevailed at the Little Cypress Creek gage (streamflows less than 100 cfs, roughly the 40^th percentile flow for that stream), weekly at flows above 100 cfs, and daily at high flows (above the 75^th percentile flow of 660 cfs) to assure adequate sampling at higher flow regimes, and particularly, to obtain information on nutrient concentrations over one or more storm hydrographs.  Samples at the Caddo Lake outlet were collected only during periods when the lake was spilling.  Parameters analyzed for are the major plant nutrients, including the commonly encountered compounds of nitrogen, phosphorus, carbon and chloride, a conservative (not biologically active) constituent useful in model calibration since its concentration is affected only by addition and dilution.   Field parameters (dissolved

 

oxygen, temperature, pH, conductivity) were measured concurrent with water sample collections.

 

Sampling procedures, preservation and holding times, methods of chemical analysis, laboratory reporting limits, and QA/QC procedures conformed to the practices presently in use in the Cypress Creek Basin Clean Rivers Program and are documented in Appendix G, Revision 1 of the current (2006-2007) QAPP (Appendix B).

 

2.2 Modeling

 

To estimate the phosphorus (P) load into Caddo Lake, Load Estimator (LOADEST), a FORTRAN program developed by the USGS (Runkel et al., 2004), was used.  LOADEST uses daily streamflow and P concentration values to calculate a regression model to estimate constituent load (Runkel et al., 2004).  Output from the model includes load estimates, standard errors, and 95% confidence intervals on a monthly basis.

 

LOADEST incorporates three statistical methods, Adjusted Maximum Likelihood Estimation (AMLE) and Maximum Likelihood Estimation (MLE), and Least Absolute Deviation (LAD).  The AMLE and MLE methods are used when the observed versus simulated residuals are normally distributed.  The LAD method may be used when the residuals are not normally distributed.  The model and supporting model documentation may be found at the file transfer protocol site at ftpdcolka.cr.usgs.gov.

 

For the study LOADEST simulations were completed for the Little Cypress, Big Cypress, Black Cypress, and James Bayou watersheds.  As shown in Table 2-1, the simulations used USGS streamflow gages 07346045, 07346000, and 07346070.  James Bayou streamflow was estimated based on a drainage area ratio of Black Cypress Bayou to James Bayou. 

 

Table 2-1

Watersheds modeled using LOADEST

 

Watershed	USGS ID	Watershed Area (km2)	Period of record for Phosphorus data
Big Cypress	07346000	1,360	1979-2007
Little Cypress	07346070	1,080	1974-2007
Black Cypress	07346045	584	1974-2007
James Bayou	ungaged	540	1972-2007

 

To model the total phosphorus loading for each watershed, both streamflow and phosphorus concentration values are required.  Figure 2-2 shows the input streamflow and concentration used for each watershed and plotted versus each other.  As is typical in most settings, total phosphorus concentration decreases as streamflow increases.  The phosphorus data used in this model were collected from the 1970s through May 2007 and include the information from the SWQMIS data base in addition to the data collected during this study (Appendix A).

 

Phosphorus concentrations that were below the detection limits of the laboratory analysis method were censored.  LOADEST provides a method by which to incorporate these censored data into the calibration input file (calib.inp).  For each concentration that was below the detection limit, a less-than sign, “<”, was placed before the detection limit to denote the censored data within the LOADEST model.

 

Figure 2-2

Input streamflows and concentrations used in the LOADEST model

 

 

 

 

 

3.0 Results and Discussion

 

Figure 3-1 presents average daily inflows to Caddo Lake reported from the gaged watersheds,  estimated inflows from the ungaged areas, and outflow measured at Caddo Dam during the study year, effectively illustrating the abrupt end of the regional drought.  The overall water balance input – output) is better illustrated by comparing combined inflows with Caddo Dam out in a cumulative format; adding the total daily flows over the study period (Figure 3-2.  Combined inflow volume (776,680,070 m³⁾, together with the change in lake volume between the lowest level reached (October 25, 2006) and the volume at the end of May, 2007 (53,652,900 m³), agrees with the cumulative discharge from Caddo Dam during the study period (876,490,118 m³) to within 5.6%.

 

Table 3-1 lists average values and ranges of parameters measured in the field during the period July 2006 through May 2007 concurrent with the collection of water samples for laboratory analysis.  The results of those analyses are summarized in Table 3-2.  The complete data set is presented in Appendix A; field measured parameters are in Table A-1, while Table A-2 contains the results of laboratory water chemistry analyses.

 

During the 2006-2007 study period, lowest dissolved oxygen concentrations tended to occur during the summer – fall low flow period, while low pH and clarity (Secchi depth) tended to coincide with the initial high flow period in January and February.  The episodes of dissolved oxygen supersaturation common during the latter period reflect the influx of relatively cold runoff water (Appendix Table A-1).  A large proportion of the inorganic nitrogen (i.e., ammonia, nitrite and nitrate) data consisted of censored values, that is, while the material may have been present, the concentrations were less than could be reliably measured using the prescribed laboratory techniques (Appendix Table A-2).  Mean and median values were therefore calculated with (1) censored values set to one-half their respective detection limits, and (2) censored values set to zero to span the range of probable results.  The summary results in Table 3-2 show that the bulk of nitrogen entering Caddo Lake is bound to organic matter (living algal and bacterial cells and organic detritus) rather than as inorganic nitrogen immediately available for uptake and growth.  Dissolved phosphorus was not monitored during this study as experience in the region has shown that detectable levels of this more available form are rarely obtained.  The consistently low levels of these nutrients suggest that growth rates in both inflowing water and that exiting Caddo Lake are limited by the rates at which they are recycled from living and dead tissue.  The lowest average levels of both total nitrogen (nitrate + TKN) and total phosphorus occurred in Big Cypress Creek just below Lake O’ the Pines (Station 15135), while the highest levels were present in Black Cypress Bayou (10245) and at Caddo Dam (18843).

 

Linear regression analyses of total phosphorus, total nitrogen and total organic carbon concentrations of streamflow during the study period consistently showed negative relationships (lower concentrations at higher flows).  While the slopes of the regressions were often significant (P<0.05), the r² numbers were also consistently low, most likely because the relationships are non-linear, as is the case in Figure 3-3, where both linear and polynomial trend lines are plotted.

 

 

 

Table 3-1

Average, Median¹ and Range of Field Measured Parameters in the Caddo Lake Nutrient Study, July 2006 Through May, 2007

 

 

 

 

 

 

 

 

Table 3-2

Average, Median¹ and Range of Water Chemistry Results From Samples Collected in the Caddo Lake Nutrient Study, July 2006 Through May, 2007

 

 

 

 

Table 3-2 continued

 

¹ Censored values assumed to equal one-half lab detection limit

 

 

 

 

A comparison of monitoring stations using a nonparametric analysis of variance (Kruskal-Wallis AOV) indicated statistically significant (P<0.05) differences in total phosphorus concentration between the Big Cypress Creek station below Lake O’ the Pines and the stations on Little Cypress (10332) and Black Cypress Creeks.  A single, very high value for total phosphorus (1.06 mg/l) was obtained from the Caddo Dam station on October 29, 2006.  Its elimination from the data set results in an average total phosphorus concentration of 0.11 mg/l in the water exiting Caddo Lake, somewhat more consistent with the results from inflowing waters than the 0.15 mg/l shown in Table 3-2.

 

The results for total nitrogen showed that average concentrations were lowest at the Lake O’ the Pines station, but were statistically different (P<0.05) only from the Caddo Dam station.  Average nutrient concentrations at the lower Big Cypress Creek station were similar to those in the other streams monitored (Table 3-2).  Total organic carbon concentrations were significantly lower at the Lake O’ the Pines station than at all other monitored locations.

 

With the exception of the uppermost station on Big Cypress Creek, total phosphorus averages between 1979 and 2006 (Table 1-1) tend to be somewhat lower than those measured during this study.  Other parameters measured sufficient times during the period 1979 through 2005 to permit comparison with the data collected during this study include chloride, TKN and TOC at all the stations except 10319, James Bayou.  These averages, compared in Table 3-3, exhibit substantial differences between particular pairs, but no general pattern is evident.

 

Table 3-3

Average Parameter Values (mg/l) Recorded in the 1980-2006 Period of Record (POR) and During the Caddo Lake Nutrient Budget Study

 

Nutrient data are particularly sparse for the minor drainages surrounding Caddo Lake.  Including results collected within the last year, but excluding outlying values that obviously represent anomalous events and the data from Station 10295 that is dominated by Big Cypress Creek water, only 27 data points are available, consisting of samples from Harrison Bayou, and data collected Haggerty and Kitchen Creeks in 1998 and 1999 Special Studies.  Using only the data from the lowermost Harrison Bayou Station (15509), and the data from Haggerty (16253, 3 samples) and Kitchen Creeks (14998, 4 samples), gives a data set of 18 values and an average total phosphorus concentration of 0.123 mg/l.  This latter value, rather than the more inclusive (and substantially higher) average given in Table 1-1, is used in modeling the annual total phosphorus loads originating in the minor drainages.

 

Although it is desirable to balance nutrient inflows to Caddo lake with corresponding outflows, inflows from the minor ungaged drainages can only be estimated by application of drainage area and nutrient concentration ratios, resulting in substantial uncertainty.  Aggregate total phosphorus loading during the study period totaled 114,622  kg, of which only 78,025 kg originates in the four major watersheds, while outflow from Caddo Dam totaled 68,627 kg, a difference of 45,995 kg.  TMDL studies in Lake O’ the Pines indicated that an even larger proportion of the inflowing phosphorus load was trapped in that reservoir.  That this may represent a real difference is suggested by the higher turnover of Caddo Lake and the generally high total phosphorus concentrations in Caddo Dam outflows versus the low concentrations measured below Ferrills Bridge Dam (Table 3-2).

 

Living tissues, particularly rapidly growing plankton populations, tend to exhibit a nitrogen-phosphorus (N:P) ratio of 16:1 on an atom for atom basis, and are expected to take up those nutrients in that ratio for growth.  Comparison of total nitrogen and total phosphorus concentrations on that basis in the samples collected during the study period shows large variations in the N:P ratio among sample dates, but the average and median values tended to show the expected balance except in the lower Big Cypress Creek station (10295) and Little Cypress Creek (10332) in which nitrogen appears to be in short supply (Table 3-4).  With respect to potential nutrient limitation or control on production and coupled respiration, both total phosphorus and TKN represent nutrient reservoirs that are of relatively limited availability; the phosphorus contained in living and dead organic material, adsorbed to particulates, or present as polyphosphate, and the nitrogen present primarily as organic matter.  In this condition, the nutrient loads entering Caddo Lake represent potential nutrients, that is while they can eventually be utilized for population growth and the accumulation of biomass, the actual rates at which photosynthesis and respiration can occur will to depend on the rates of nutrient regeneration from suspended particles, sediments and living populations, given appropriate conditions of temperature, light, and other nutrients.

 

Table 3-4

Nitrogen Phosphorus Ratios (mg-at/l) at Nutrient Study Stations, July 2006- June2007

 

 

Initial runs of the LOADEST model were performed to select the best regression models based on the data collected during the 2006-2007 study period, and applied to the historical data available from the TCEQ data base.  Residuals for the four model simulations were normally distributed so, the AMLE load estimates were used and are shown in Figures 3-4 and 3-5.  LOADEST files, including the initial regression development and validation and the total phosphorus load output files from four locations, Big Cypress Creek, Little Cypress Creek, Black Cypress Creek, and James Bayou are included on electronic media as they extend over several hundred pages.    Although all available streamflow data was included in the LOADEST model, for loading analysis only the 26-year period 1980-2006 is considered here because streamflow data is available for all three gaged watersheds then and the both the frequency and reliability of nutrient data tends to become much less prior to 1980.

 

Nitrogen loads were not modeled due to the lack of interpretable data on the various nitrogen species, and inability to accurately assess the fate of what appears to essentially be an organic nitrogen load entering Caddo lake.  In addition, the phosphorus concentrations present in Caddo Lake and its tributaries and our experience in the Cypress Creek Basin with the widespread phosphorus loading of agricultural soils that occurred with past fertilization practices and indicated that, given the constraints of time and resources, this study would be best served by focusing the modeling effort on phosphorus loading.

 

During the period including 1980-2006, modeled total phosphorus loads averaged 182,786 kg/year.  The substantial variation among years is shown in Figure 3-4, and the distribution of loads among the four major tributaries and the aggregate minor drainages closest to Caddo lake is presented in Figure 3-5.  This can be compared with the estimated total loading for the study period of 114,622 kg.  The three gaged watersheds (Big Cypress Creek, Little Cypress Creek and Black Cypress Bayou), which were independently modeled, exhibit relatively similar patterns of annual loading, reflecting the dominance of climatic conditions.  This figure also indicates that the largest contributor of phosphorus to Caddo Lake consists of the small streams flowing directly into the lake and the lower portions of the major watersheds below the stream gages and water quality monitoring stations.  However, because of the lack of stream gages and regular water quality monitoring in this area, it could not be included in the LOADEST model.  Total phosphorus loading in the aggregated minor drainages was estimated by adjusting the Black Cypress Bayou loading results for drainage area and for average total phosphorus concentration differences during the same periods (see discussion on page 21).  While these results may serve to fill out the analysis, and may actually reflect reality, it certainly suggests  where some additional sampling effort might be focused.

 

Seasonal patterns in total phosphorus loads entering Caddo Lake are also climatically driven, with the largest loads being delivered with the highest streamflows (Figure 3-6), in spite of the tendency for higher streamflows to exhibit lower nutrient concentrations (Figure 3-3).  This pattern is the same as that documented in Lake O’ the Pines, and as in that reservoir, the nutrients delivered during the winter-spring wet season will support photosynthesis and respiration during the summer when water temperatures are highest, non-algal clarity is maximum and lake levels are lowest.

 

In order to better assess the impact of phosphorus loading on Caddo Lake, historic water surface elevation data was used to calculate annual median lake areas and volumes.  The lake elevation is generally above the spillway elevation, and using that static datum to estimate turnover (the number of times per year the lake volume is replaced by inflows) will result in an overestimate.  In addition, varying inflows insure that lake elevations, and consequently lake area and volume, will vary, so an annual estimate is desirable to use for input to the Vollenweider model.  Median annual levels, rather than average levels were chosen to avoid the effects of brief, but intense inflow events, since we assume that the physical factors (water depth, turnover) affecting nutrient delivery

 

 

 

 

 

 

(winter-spring) and utilization (summer) will be a function of the conditions prevalent over an entire year (Figure 3-7).

 

Over the 26 year period examined, lake volumes were replaced an average of 8.1 times/year, substantially less than the 12.6 times/year estimated in the introduction using the spillway elevation and a long term average inflow estimate.  Likewise, the 26-year average areal phosphorus load calculated from the modeled phosphorus loads and lake areas derived from annual median water surface elevations is 1587 mg/m³, about 85% of the low end of the range estimated in the introduction.  Figure 3-8 summarizes the turnover and areal phosphorus load in Caddo Lake by year for the 26 year period.  The larger lake surface areas resulting from higher water surface elevations during periods of high inflows (years of high turnover) do not do much to oppose the simultaneously larger phosphorus loads also delivered to during high inflow periods.

 

Annual predictions of Caddo Lake total phosphorus concentrations using the Vollenweider model are shown in Figure 3-9.  Although the predicted concentrations vary widely throughout the period of record, the 26 year average of 66.7 mg/m³ (0.067 mg/l) is the same as the low end of the estimated range derived from long term averages in the introduction, and is about 2/3 the whole lake total phosphorus level estimated for Lake O’ the Pines in the TMDL study of that reservoir.  The high end of that range (77.0 mg/m³) corresponds closely with the total phosphorus concentration calculated for Caddo Lake (79.16 mg/m³) by TCEQ in “Trophic Classification of Texas Reservoirs, 2004 Water Quality Inventory and 303(d) List” (May 13, 2005).   Finally, the phosphorus levels from the TCEQ mid lake station (10283) and the USGS station near the dam reported in the introduction are also quite similar to our 26-year average, giving some

 

 

 

 

 

 

 

confidence in the present assessment of the level of total phosphorus loading and the trophic condition of Caddo Lake.

 

The phosphorus concentrations observed in Caddo Lake and its tributaries, and the whole lake concentrations derived in this study, are at levels commonly recognized as potentially problematic in lentic habitats.  For example, summer total phosphorus concentrations on the order of 0.04-0.05 mg/l are widely recognized to indicate potentially eutrophic conditions,[11] or to entail a high probability of severe algal blooms,