Water Resources U.S. Department of the Interior, U.S. Geoiogical Survey WRIR-^6-4273

STUDY OF NONPOINT SOURCE NUTRIENT LOADING IN THE PATUXENT RIVER BASIN, MARYLAND Stephen D. Preston J INTRODUCTION quality sampling effort. The modeling component since September 1985. In July 1995, the primary includes both watershed and estuarine water-quali­ responsibility for the study of nonpoint source nutri­ Background ty models that are calibrated and verified with the ent loading in the State was transferred to the appropriate monitoring data. Together, the data col­ Maryland Department of Natural Resources In 1985, the U.S. Geological Survey (USGS), in lection and modeling projects are used by the State (MDNR), and the USGS and MDNR have conduct­ cooperation with the Maryland ed the Patuxent study jointly since that Department of the Environment (M3E). date. The collection of water-quality data initiated a study of nonpoint source is expected to continue at least through (NFS) nutrient loading in the Patuxent September 1997. River watershed. The purposes of the study are to: (1) quantify NFS nutrient loading at various locations in the Purpose and Scope watershed; (2) gain further understand­ This report provides an overview of the ing of the relation between land use major components of the study of nutri­ and water quality; and (3) evaluate the ent loading in the Patuxent watershed. potential effectiveness of management The main focus of the joint USGS, MDE, practices. In order to achieve these and MDNR Patuxent study has been the goals, hydrologic, meteorologic, geo­ collection of surface-water discharge and graphic, and water-quality data for the water-quality data from representative Patuxent watershed were collected locations throughout the watershed. The and compiled. A model was then data-collection effort has provided developed that allows simulation of the Water-quality detailed hydrologic time-series data that hydrology and water quality of the Monitoring Sites are necessary for model development. A watershed and enables evaluation of Gaging Stations second major focus of the study has various land-use alternatives on water been the development of a detailed quality. watershed model that has been used to evaluate various land-use and land-man­ The Patuxent watershed was selected agement alternatives. Water-quality syn­ for study because of its importance to optic studies were conducted by collect­ the State of Maryland and because it ing a small number of samples over a is representative of other subbasins of broad spatial area. The synoptic studies the Chesapeake Bay. The Patuxent were designed to provide information traverses the corridor between the two about spatial water-quality patterns in the major metropolitan areas of Baltimore, watershed, and about the causes of Md., and Washington, D.C., and com­ those patterns. All of the components of mercial, industrial, and residential the Patuxent watershed study are cur­ development within the watershed has rently underway and are in various been substantial. Development in the stages of completion. Detailed descrip­ Chesapeake Bay region is expected to tions of the results of the study compo­ continue, and the State has a strong nents have been presented to State and interest in evaluating the effects of local agencies and will continue to be future growth on water quality and in Figure 1 The Patuxent River watershed, monitoring site locations and sur­ rounding area. presented and published as the study defining methods for planning growth to continues. minimize potential adverse effects. to analyze alternative management strategies and Maryland's overall strategy for the study of nutrient to evaluate progress for meeting nutrient-load- LONG-TERM MONITORING loading in the Patuxent watershed calls for compre­ reduction goals. hensive monitoring, modeling, and research to Purpose The USGS and the MDE have addressed the need reduce scientific uncertainty about nutrient loading Long-term surface-water discharge and water-qual­ and the response of the watershed to management for water-quantity and water-quality data in the non- tidal part of the Patuxent by initiating a long-term ity data collection was implemented in the Patuxent actions. The monitoring program developed to meet watershed in order to provide the data necessary the requirements of the Patuxent nutrient-reduction study of the hydrology and water quality of the watershed. Water-quality data have been collected for evaluating various aspects of nutrient loading, strategy includes both a tidal and nontidal water- including trend detection. Data collection was intended to be long-term because the Patuxent and River Input studies has processes that control NFS nutrient resulted in a more detailed data record at loading can occur over an extended the Bowie site than at the other sites. period of time, and the detection of Land use in the Bowie subbasin during water-quality changes related to land 1985 was 44 percent forest, 30 percent use may require observations over agriculture, and 22 percent urban and decades. Because of the need for long- residential. Several large sewage treat­ term water-quality data, significant ment plants are located immediately resources were expended to build mon- upstream of the Bowie site, and could itoring-site housing and equipment con­ have a significant effect on water quality figurations that would allow long-term at the site. or even permanent data collection. A water-quality-monitoring site was In part, long-term water-quality data EXPLANATION established on Western Branch at Upper Urban were collected to provide the statistical Agriculture Mariboro, Md., to represent an area of basis for documenting changes in urbanization in the Coastal Plain. Forest water quality and relating those Western Branch is a relatively large Water changes to land-use patterns. Stream Wetlands stream that discharges directly to the nutrient and sediment loads have been Barren Land Patuxent estuary. The Western Branch estimated each year that data have Other subbasin has received a substantial been collected, and the estimates have amount of development because of its been compared among the sites to location near Washington, D.C. Land use investigate potential relations between in the Western Branch subbasin in 1985 the land-use patterns of the monitoring included 38 percent forest, 30 percent subbasins and loads. In addition, trend agriculture, and 24 percent urban and analysis has been planned to detect residential area. current trends in nutrient and sediment concentrations and loads. Future trend Two monitoring sites were selected to analyses have been planned to detect represent the numerous small Coastal water-quality trends related to anticipat­ Plain subbasins along the Patuxent estu­ ed implementation of management ary. In Calvert County, the Hunting Creek practices. subbasin was selected, and a monitoring site was established at the Md. Route Water-quantity and water-quality data 263 crossing. In St. Marys County, the were also collected to provide the data Killpeck Creek subbasin was selected, base necessary for calibration of a Figure 2. Land-use distribution in the Patuxent River watershed during 1985 and a monitoring site was established detailed watershed model. The water­ (data from Maryland Office of Planning). near Huntersville. In both subbasins, land shed model was developed as a tool for use in 1985 was predominantly forest (63- relating land-use practices to water quality and for The farthest upstream site is near Unity, Md., 76 percent) and agriculture (20 -26 percent), with understanding the controls of water quality. The immediately upstream from the Tridelphia the remainder being urban and residential. watershed model will also be useful for investigat­ Reservoir. The Unity site represents an area of rural land in the Piedmont part of the watershed. ing the effects of potential land-use change and Approach land-management alternatives on water quality, Land use in the Unity subbasin during the year that and to simulate hydrologic conditions and water the study was initiated (1985) was primarily agricul­ Stream discharge and water-quality data are col­ quality in unmonitored parts of the watershed. ture (64 percent) and forest (32 percent). lected at each of the monitoring sites. Discharge data are collected using standard USGS stream- One site was selected to monitor the water quality flow-gaging procedures in which high-frequency Site Descriptions of the Middle Patuxent and Little Patuxent Rivers, stage measurements are converted to discharge Surface-water discharge and water quality were which are major tributaries of the Patuxent main using a rating curve (Buchanan and Somers, monitored at six stream sites that are representa­ stem. The site is south of Savage, Md., down­ 1982). Water-quality samples are collected manual­ tive of conditions in the Patuxent watershed (fig. 1). stream of the confluence of the two tributaries. The ly during low-flow conditions and automatically dur­ The area of the watershed is about 930 square area draining to the Savage site has received sig­ ing high-flow events. Data on the concentration of miles, and land use or land cover in 1985 consisted nificant development pressure over the recent past nutrients during high-flow events are critical to the of approximately 46 percent forest, 32 percent agri­ so that the site represents a subbasin that has accuracy of nutrient loading estimates (Preston and culture and 15 percent urban/residential (fig 2). The recently experienced a shift in much of its area Summers, in press) and significant effort has been northern; third of the watershed is in the Piedmont from agricultural to urban and residential land uses. expended to ensure reliable collection of high-flow Physiographic Province and the remainder is in the Land use in the Savage subbasin during 1985 was samples. Coastal Plain. The monitoring sites were selected mostly agriculture (35 percent), but a large fraction to represent various subbasins, physiographic set­ of land area was urban and residential (31 per­ To ensure the long-term reliability of sample collec­ tings, and mixes of land uses in the watershed. cent). tion, considerable resources were expended to Two of the sites are on the main stem of the establish and maintain stable infrastructure at the A second main-stem monitoring site is located near monitoring sites. The potential for damage from Patuxent River, two are on major tributaries to the Bowie, Md., downstream from the Fall Line and Patuxent, and two are on streams in small Coastal high, fast-flowing water and storms creates a large upstream from the tidal part of the Patuxent River. potential for failure of sampling equipment and the Plain subbasins that discharge directly to the Data collected at the Bowie site are used by both Patuxent estuary. Five additional streamflow-gaging loss of important water-quality data. To minimize the Patuxent study and a separate study of nutrient the potential for sample loss, the housing and stations in the watershed are supported by other loads to the Chesapeake Bay, called the River studies. equipment configurations at the monitoring sites Input Study (Zynjuk, 1995). The combined effort of were designed to withstand extreme weather and streamflow conditions. At each of the sites, semi­ using the automatic sampler. Comparison of the Annual runoff (mean discharge divided by subbasin permanent housing was established to shelter results indicated that the two methods provided area) is a measure of the relative amount of water data-collection equipment (fig. 3). Sampler intake data that were nearly equivalent throughout a exported per unit area of the subbasin. Annual lines were installed between the housing and the range of discharges. Suspended sediment was the runoff is similar among the six sites. Slight differ­ stream, and were stabilized to avoid damage dur­ only constituent that exhibited slight differences ences at Western Branch and Hunting Creek may ing floods. between the two methods; suspended-sediment be due in part to differences in the periods of data concentrations in samples collected by the auto­ collection. Slightly higher annual runoff at Killpeck Water-quality samples were collected during low- matic sampler were slightly lower than those in Creek may be due to greater runoff caused by the flow and high-flow conditions to ensure sample samples collected manually from the stream. steep terrain and incised valleys in the subbasin. coverage over the full range of discharge. Low-flow samples were collected monthly using the equal- All laboratory analyses were performed by the Suspended-sediment concentration is a measure of Maryland Department of Health and Mental the paniculate content of the water, and suspend­ Hygiene laboratory, except for the suspended-sedi­ ed-sediment load is a measure of the mass of sedi­ ment analyses, which were performed by the ment transported by the stream particularly during USGS sediment laboratories in Lemoyne, Pa., and high discharge periods. Mean annual yield of sus­ Louisville, Ky. Data are reported annually in the pended sediment was highest at the Savage, USGS State report for Maryland, Delaware, and Western Branch, and Killpeck Creek sites. The District of Columbia (for example, James and oth­ subbasins represented by the Savage and Western ers, 1991). All data are stored electronically in the Branch sites are the most urbanized. High sus­ USGS National Water Information System (NWIS) pended-sediment yields may reflect construction in data base. those subbasins or particulate matter washed off impervious urban surfaces, such as roofs, roads, Summary of Selected Long- and parking lots. High suspended-sediment yield from the Killpeck Creek subbasin may be a reflec­ Figure 3. Housing was installed at each monitoring Term Monitoring Data tion of streambank erosion caused by rapid runoff site to protect automatic sampling equipment and to Hydrologic and water-quality data were collected in from steep slopes and ongoing valley incision. allow automated water-quality sampling during the Patuxent watershed from October 1985 through high-flow events. September 1996. Data collection was temporarily Total phosphorus yields show a pattern similar to width-increment (EWI) method (Edwards and discontinued at the Western Branch site because that observed for suspended sediment. Total phos­ Glysson, 1988), which is designed to provide hori­ of construction during the period 1989-91. Data col­ phorus concentration is often related to suspended- zontally and vertically integrated samples that are lection began in 1989 at the Hunting Creek site sediment concentration because total phosphorus representative of the entire stream cross section. when the streamflow-gaging station was first tends to be mostly in the particulate form. High-flow samples were collected using automated installed. Consequently, phosphorus is readily transported by sampling devices that initiated sampling as the high discharge when particles are entrained and stream stage reached a specified height that is carried by rapidly flowing water. As with suspend­ indicative of surface-water runoff. ed-sediment yields, total phosphorus yields are highest at the Savage, Western Branch, and High-flow sampling equipment at each site includes Killpeck Creek sites (fig. 5). These high total phos­ a flow meter and an automatic sampler (fig. 4). The phorus yields may be related to sediment transport flow meter records the stream hydrograph and from construction areas, runoff from urban areas, graphically indicates the time when samples are and, in the case of Western Branch, the discharge collected. The automatic sampler pumps water from sewage-treatment plants. from the stream or mixing chamber and stores it in separate sample bottles. As many as 24 samples Total nitrogen yields exhibit a spatial pattern that is can be collected; however, large numbers of sam­ different than that observed for suspended sedi­ ples are usually composited to yield from one to ment and total phosphorus. Most of the nitrogen five samples per storm. Automatic samplers are transport occurs through a different pathway than refrigerated at all sites to preserve the samples as Figure 4. Automatic sampling equipment at each the other two constituents. Total nitrogen consists they are collected. Samples are retrieved and site includes a flow meter for continuously primarily of nitrate at most sites and nitrate is com­ recording discharge and a refrigerated automat­ monly transported to streams through ground processed for laboratory analysis within 48 hours of ic sampler for collecting and preserving water collection. water. As a result, geology or other natural factors samples. may play an important role in the amount of nitrate Storm samples are collected in proportion to flow Selected hydrologic and water-quality data collect­ that reaches the stream and is subsequently trans­ rate by collecting a sample each time that a specif­ ed through September 1994 are summarized in ported from the subbasin. ic volume of flow is measured by the flow meter. table 1. Data summaries for mean daily discharge, Total nitrogen and nitrate-nitrite yields were great­ Thus the sampling frequency is increased during suspended sediment, total phosphorus, total nitro­ est at the Unity, Savage, and Bowie monitoring high flows and the probability that samples will be gen and nitrate-nitrite are included in table 1 sites (fig. 5). Most of the area of these subbasins is collected during the highest flows is also increased. because of their importance in estimating and in the Piedmont Physiographic Province; the other When necessary to minimize the number of sam­ understanding nutrient loading. Summaries include three subbasins are entirely within the Coastal ples, individual samples are composited to repre­ the number of data values available, the mean, Plain Physiographic Province. Some of the differ­ sent specific parts of the hydrograph, such as the standard deviation, mean annual load and mean ences in nitrogen yield between the Piedmont and rising and falling limbs, and the peak discharge. annual areal yield. Load estimation methods have Coastal Plain sites are most likely due to the pres­ been described by Preston and Summers (in Tests were performed to verify that samples collect­ ence of point sources upstream (for example, the press). Areal yield is simply the mean annual load ed through the sampler intake were representative Bowie site). Some differences also may be due to divided by the subbasin size. Yields are more com­ of the water in the stream (Koterba and others, land-use distribution and nonpoint sources. parable than loads among the sites because vari­ 1994). As part of these tests, samples were collect­ However, it is possible that geologic differences ability due to subbasin size is eliminated. ed manually in the stream and simultaneously between the two physiographic provinces affect the r, Table 1. Summary of discharge and selected water-quality constituent data collected at the Patuxent monitoring sites ing. Because many environmental processes affect Western Hunting Killpeck nutrients as they pass through the watershed, it is Unity Savage Bowie Branch Creek Creek difficult to quantify any one process independently Daih Diwbti I ^m 1 to determine its relative importance. The watershed Number of data 3,286 3,286 3,286 2,221 2,429 3,286 model integrates all of the major environmental Discharge - Mean (ft7sj 38.3 106 345 82.9 10.5 3.84 processes and allows them to be quantified sepa­ - Standard Deviation (ftVs) 59.7 167 421 134 15.2 4.62 rately. Thus the model could be used to quantify Annual Runoff (in/yr) 14.9 14.6 13.5 12.5 15.2 16.0 r ~ E ~m the relative importance of nonpoint source and niKipeniitd Sciiiiiitii point source loading and the effects of urban and Number of Data 175 173 273 145 127 176 Concentration - Mean (mg/L) 110 241 69.6 193 25.6 309 agricultural land use on loading. - Standard Deviation (mg/L) 287 468 89.1 275 29.5 575 The model also provides a tool for spatial extrapo­ Load (Mg/yr) 5170 34,400 22,100 16,400 286 987 lation of available information. The model is first 24.6 70.5 11.8 117 Yield (g/m2-yr) 57.4 135 calibrated where data are available in order to Total Phosplmrus Number of Data 252 163 551 285 114 147 determine the appropriate parameter values for a Concentration - Mean (mg/L) 0.126 0.273 0.213 0.238 0.128 0.290 representative area. The model and parameter val­ - Standard Deviation (mg/L) 0.265 0.404 0.163 0.419 0.155 0.0374 ues defined by calibration at one site can then be Load (Mg/yr) 7.55 35.0 60.1 31.2 1.27 1.06 applied to other nearby areas with similar charac­ Yield (g/nf-yr) 0.084 0.137 0.067 0.134 0.052 0.126126 I teristics. In that way, the entire watershed can be TVlftl Nftru£en H modeled and the relative importance of all nutrient Number of Data 239 153 465 276 108 140ri source areas to the Patuxent can be determined. Concentration - Mean (mg/L) 2.84 2.51 3.52 1.21 0.685 1.96 - Standard Deviation (mg/L) 0.923 0.868 1.54 0.686 0.627 0.691 The model can also be used to estimate the effects Load (Mg/yr) 106 254 835 118 6.69 6.84 of changes in the watershed on nutrient loading. Yield (g/m'-yr) 1.18 1.00 0.926 0.508 0.275 0.810 The relations between land use and nutrient load­ ^~ -3 Nitrate I \ftrite I ing are quantified by the model and these relations Number of Data 238 159 547 270 114 146 can be extrapolated to estimate nutrient loads Concentration - Mean (mg/L) 2.24 1.49 2.37 0.412 0.129 1.08 under past or future land-use scenarios. Currently, - Standard Deviation (mg/Lj 0.602 0.591 1.21 0.244 0.095 0.566 MDE is applying the Patuxent watershed model to Load (Mg/yr) 75.5 133 546 37.6 1.37 3.64 simulate nutrient loads under 100-percent forest Yield (g/m'-yr) 0.838 0.522 0.606 0.162 0.056 0.431 cover to simulate pristine water-quality conditions, and under a hypothetical land-use pattern in the amount of nitrate that reaches the stream through cal framework for integrating and extrapolating the year 2000 to estimate future water-quality condi­ ground water. Synoptic sampling studies that have collected information. The model was designed to tions. MDE is also using the model to determine been carried out to expand the spatial extent of simulate the hydrologic conditions and water quali­ the potential effects of management practices such available data, confirm the differences observed ty of the watershed based on forcing functions as nutrient management and conservation tillage among the long-term monitoring sites, and provide such as precipitation, land use, nutrient sources, on nutrient loading. additional information for understanding the con­ and other basin characteristics. Mathematical simu­ trols of stream nitrogen concentration. lation of the related environmental processes pro­ vides a mechanism for improving the understand­ Model Description WATERSHED MODELING ing of factors affecting nutrient loading and for pre­ The Patuxent watershed model is a continuous dicting the effects of land-use changes. simulation model that was designed to simulate Purpose water-quality constituent concentration or load on The watershed model is being used to improve the the basis of surface-water runoff, ground-water Watershed modeling was implemented as part of understanding of the relative importance of envi­ input, point-source input and instream processes. the Patuxent study in order to provide a mathemati­ ronmental processes or sources for nutrient load­

PATUXENT NUTRIENT YIELDS Monitoring Basin Comparisons

00 Unity Savage Bowie Western Hunting Killpeck Unity Savage Bowie Western Hunting Killpeck Branch Craek Craek Branch Creek Creek

Figure 5. Nutrient loads at each site can be compared by calculating the nutrient yield (load divided by area). The differences among yields can then be related to differences in basin characteristics such as land use or the implementation of management practices. The hydrologic component of the model is based tions. Low-flow synoptic sampling was intended to ness. The evaluation of NPS nutrient loading in the on precipitation and other meteorologic variables, provide additional information for understanding the watershed has been comprehensive and has and directs water through multiple flow pathways to causes of water-quality spatial variations. included long-term monitoring, detailed watershed predict stream discharge. The water-quality compo­ modeling, and synoptic sampling studies. A large nent of the model simulates instream and land-sur­ Samples were collected during August and amount of information has been compiled for the face biological and chemical processes such as September in each of water years 1993,1994, and watershed and that information is being used to plant nutrient uptake, denitrification, and sediment 1995. Summers are usually dry in the Patuxent identify the primary controls of and efficient man­ interactions. watershed area and base flow usually dominates agement strategies for NPS nutrient loading. streamflow during August and September. During The Patuxent watershed model is considered to be 1993, emphasis was placed on the Piedmont part The Patuxent NPS studies have produced tools spatially detailed. The overall model is composed of the watershed and most sampling locations were that will allow ongoing evaluations in the Patuxent of segments among which input data and parame­ located in that physiographic province. During 1994 River watershed and in the State of Maryland. The ter values can vary. The model is composed of 39 and 1995, emphasis was placed on the Coastal NPS evaluation tools generated by the Patuxent segments that average approximately 20 square Plain part of the watershed and fewer streams studies include extensive data bases and a miles in size (fig. 6). Segmentation of the water­ were sampled in the Piedmont. shed was based upon many factors that affect the spatial variability of the model, such as data-collec­ Selected Results tion locations, point-source locations and variations Nutrient Analyses in basin characteristics such as land use. Results of nitrate analyses of The Patuxent watershed model was calibrated at samples collected during the all locations where load or discharge data were synoptic sampling studies gen­ available. All of the long-term monitoring sites erally confirm the spatial pat­ described above were used as hydrologic and terns observed among the long- water-quality calibration sites. In addition, data from term monitoring sites. The five other stream-gaging stations were used for nitrate distributions for each hydrologic calibration. The model was calibrated for physiographic province and for various water-quality constituents including four each water year are shown in forms of nitrogen, three forms of phosphorus, total figure 8. As was observed for suspended solids, dissolved oxygen, biological oxy­ the long-term sites, nitrate val­ gen demand, and chlorophyll a. ues in the Piedmont part of the Water-quality Patuxent watershed were con­ Monitoring Sites Input to the model included meteorologic time sistently higher than those in the Gaging Stations series, point-source discharge time series, and Coastal Plain. Median nitrate geographic data. Hourly precipitation time series values were always more than data were compiled from seven rainfall monitoring 1.5 milligrams per liter (mg/L) in sites as input to the hydrologic component of the the Piedmont, but never exceed­ model. Discharge information from 33 point ed 0.5 mg/L in the Coastal Plain. sources of nutrients in the watershed was included Nitrate exceeded 1.0 mg/L in 96 as input to stream reaches. Geographic information percent of the Piedmont sam­ used in the model included land use, soils, and ples and reached as high as 12 topography. Runoff from 20 types of land use, mg/L in one sample. Nitrate including 6 types of impervious area and 6 types of exceeded 1.0 mg/L in only 12 agricultural row crops, was simulated to predict percent of the Coastal Plain nutrient inputs from nonpoint sources. samples and exceeded 2.0 mg/L in only 2 percent of the samples. LOW-FLOW SYNOPTIC SAMPLING Several factors could cause the differences in nitrate concentra­ tions between the two physio­ Purpose and Approach graphic provinces. Land-use Figure 6. The Patuxent watershed model is designed to account for spatial Synoptic sampling was performed to provide patterns differ and the percent­ variabilitv in nutrient loading by dividing the basin into segments that can be greater spatial detail on water-quality distributions age of agricultural land is higher calibrated seperatefy. in the Patuxent watershed. Differences were in much of trie Piedmont. The detailed watershed model. These tools have pro­ observed among the long-term monitoring sites, geology of the two provinces is also substantially vided information on the current status of water especially for nitrogen (fig. 5). To confirm these different and ground-water flow and chemistry may quality in the Patuxent watershed and on the con­ results, water-quality samples were collected at differ as a result. Wetlands are more prevalent in trols of water quality and loads. The Patuxent data many sites in the Patuxent watershed at nearly the the Coastal Plain and they may increase the poten­ bases and watershed model have the potential to same time, but over a broad spatial scale (fig. 7). tial for denitrification. Studies are underway to eval­ continue to provide information on water quality uate the potential causes of lower nitrate values in and nutrient loading. Thus, the cost of previous Low-flow conditions were the primary emphasis of the Coastal Plain. trie Patuxent synoptic-sampling effort. Where point work in the Patuxent watershed represents an sources of nutrients are not present, low-flow con­ investment that can continue to aid the manage­ ditions are generally indicative of ground-water SUMMARY, CONCLUSIONS AND ment of NPS nutrient loading in other areas of the input to streams (base flow). Spatial variations in FUTURE DIRECTIONS State. base-flow water quality may be due to variations in Study of nonpoint source (NPS) nutrient loading in Results of the Patuxent NPS study components to ground-water quality that are also related to geo­ Maryland has focused on the Patuxent River water­ logic, physiographic, or other environmental condi­ date have identified many important aspects of shed because of its importance and representative­ NPS nutrient loading. Point sources of nutrients REFERENC point of tidal influence. However, land-use prac­ tices are most important at all other sampling loca­ tions. Sediment transport and ground-water dis­ charge appear to be the most important mecha­ FALL LINE -*ft ^ rr,snt: U.S. nisms for phosphorus and ¥ * Report 86-531, 118 p. nitrogen transport, respec­ tively. Basin characteristics * , James, R.W., Homlein, J.F., S'mmons, R.K, and such as percentages of Strain, B.F., 1991, Water resowe*-- #i!a, urban and agricultural Maryland and Delaware, water y*ar 1991: areas, and type of geology U.S. Geological Survey Water-Data Report , affect those processes and MD-DE-91-1,449p. } are related to spatial varia­ Koterba, M.T., Dysart, J.E., Phillips, S.W., and tions identified by long- Zynjuk, L.D., 1994, Use and value of quality- term and synoptic water- control data in studies in the Chesapeake quality data-collection Bay region, in Pederson, G.L, ed.: National efforts. Ongoing evaluation Symposium on Water Quality, Herndon, Ma., of the collected data are American Water Resources Association, p. expected to continue to 65-74. provide information that will assist the manage­ Preston, S.D., and Summers, R.S., in press, ment of NPS nutrient load­ Estimation of nutrient and suspended-sedi­ ing. In particular, ongoing ment loads in the Patuxent River Basin, evaluation of the water­ Maryland, Water Years 1986-90: U.S. shed model output is Geological Survey Water-Resources expected to provide Investigations Report 96-4175, p. 69. detailed information on the Zynjuk, L.D., 1995, Chesapeake Bay: Measuring relative importance of Pollution Reduction: U.S. Geological Survey nutrient sources and trans- DOrt Dathwavs Figure 7. Low-flow synoptic sampling was performed at many sites across the Fact Sheet FS-055-95. Patuxent basin in order to detect spatial patterns in water quality. Planned future directions of NPS evaluation in the State of Maryland include continued study of water quality in the Patuxent watershed and a shift in emphasis to a statewide approach. Long-term data PATUXENT LOW-FLOW SYNOPTIC SAMPLING collection in the Patuxent watershed is planned to Nitrate Comparison continue in order to extend the data records for cc trend detection. However, future study will include LU 1- 5 5 other areas of the State to provide NPS information Piedmont _ 12mg/l_ Coastal Plain on a broader spatial scale. The statewide approach K LU will eventually become the primary approach used CL 4 4 by the State to evaluate NPS loading. The informa­ tion gained in the Patuxent study and the tools 1 developed will represent valuable assets in devel­ 2 3 3 ° oping the statewide NPS assessment program. !' 2 3 Ir Fi -f 1 3 1 T LU JL 1 1 1 y tp g 1 0 0 b z 1993 1994 1995 1993 1994 1995 Water Year Water Year

Figure 8. Base-flow nitrate concentrations tended to be lower at sites in the Coastal Plain than at sites in the Piedmont. The median nitrate value is illustrated as the centerline of each box and the 25'* and 75" percentiles are illustrated as the top and bottom of each box. Lines extending above and below each box indicate the 90"' and 10"" percentiles, respectively.