ABSTRACT: * author to whom all correspondence should be addressed.

Transcription

1 Estimation of contribution of total phosphorus from selected landuses to observed concentrations in tributaries to a coastal lagoon: Indian River Lagoon, Florida USA Gregory A. Graves * Department of Environmental Protection 1801 S. E. Hillmoor Drive, Suite C-204 Port St. Lucie, Florida telephone fax Greg.Graves@dep.state.fl.us ABSTRACT: An iterative calculation method is proposed to estimate total phosphorus in runoff from six landuses in six drainage basins based upon the acreage of those landuses, their respective runoff coefficients, and measured values from samples collected as part of various water quality monitoring programs. A computer routine was created to identify from among the multitude of possible combinations, the set of landuse-specific runoff concentrations that most suitably corresponded with monitoring data. The degree of agreement between calculated and observed concentrations was evaluated by examining cumulative error. Cumulative error was defined as the sum of the differences between the calculated basin concentrations and basin concentrations from water quality sampling data. An exhaustive trial and error process was used to identify a set of six values that would result in the least cumulative error. Each iteration of the computer program produced one million combinations of average phosphorus concentration values for the landuse types considered. Results from published literature specific to Florida were used to limit the range of possible concentrations. Multiple iterations were employed to explore relationships among calculated and observed factors, and to identify a reasonable set of estimated runoff concentrations possessing minimal cumulative error. The set of values thus identified was proposed for use as interim landuse-specific runoff concentrations until a more comprehensive model is developed. These values were subsequently used to estimate the effects of proposed changes in landuse due to future development, and to guide restoration efforts of the Comprehensive Everglades Restoration Plan (CERP) in the Indian River Lagoon subregion. Key Words Indian River Lagoon, St. Lucie Estuary, phosphorus, stormwater runoff, landuse, runoff coefficients, Comprehensive Everglades Restoration Plan, CERP * author to whom all correspondence should be addressed. 1

2 INTRODUCTION: The Indian River Lagoon is an elongated estuarine waterbody separated from the Atlantic Ocean by barrier islands (Figure 1). Its estuarine character is maintained by connection to the ocean through widely spaced inlets. Before 1892, shifting sand often closed these inlets and the lagoon was predominately freshwater. When inlets re-opened the Indian River became brackish again. Since European-descended settlers began maintaining permanent inlets, the lagoon has remained estuarine (Woodward-Clyde Consultants, Inc., 1994). Originally, the watershed of the lagoon was narrow and well defined. Since 1910, the size of the drainage basin has been increased by the construction of a network of canals (Woodward-Clyde Consultants, Inc., 1994). These canals lowered ground water, created arable lands, and reduced flooding. The area of the Indian River Lagoon considered herein is that portion of the lagoon within Martin and St. Lucie counties. The present-day watershed of the Indian River in the Martin and St. Lucie counties area consists of twenty-nine thousand hectares, most of which is developed for agricultural use (Woodward-Clyde Consultants, Inc., 1994). Figure 1. Study area. Primary "C" canals are shown as black solid lines, relict natural streams as black Canal C-25 Atlantic Ocean Indian River Lagoon North Fork South Fork Canal C-24 Canal C-23 St. Lucie River Estuary Lake Okeechobee Canal C-44 dotted lines. Background gray lines reflect primarily the network of wetlands, and secondary and tertiary canals. 2

3 A network of numerous small tertiary canals drain to larger secondary canals, which in turn drain into two relict semi-natural waterways and four primary "C" canals (Figure 1). The primary canals were constructed by the Central and Southern Florida Project (C&SF project) executed by the U. S. Army Corps of Engineers (ACOE), and are currently managed and operated by the South Florida Water Management District (SFWMD). The alteration of natural flow pathways and hydroperiod facilitated a shift in landuse from wetland-forest to agricultural-urban, which dramatically increased the concentrations and loads of nutrients delivered to the receiving waterbodies (Woodward-Clyde 1994, Janicki et al. 1999). This has led to a deterioration of water quality, particularly within the St. Lucie River Estuary basin where five of the primary water delivery systems (the North Fork of the St. Lucie River, the South Fork of the St. Lucie River, and the canals C-23, C-24, and C-44) discharge. To address the often-conflicting issues of flood protection, water supply, and water quality, a multi-agency effort was undertaken to re-examine the assumptions and intents of the original construction effort within the context of enhancing environmental benefits and values. This effort, the Central and Southern Florida Project Comprehensive Review Study (ACOE and SFWMD, 1999) resulted in CERP, which is now being implemented by the ACOE and SFWMD. One of the first subregions addressed under CERP was the Indian River Lagoon area through the Indian River Lagoon Feasibility Study. In order to evaluate the water quality benefits of the various alternatives considered during the Feasibility Study process, it was necessary to evaluate current and future (year 2050) nutrient conditions in the absence of any change in the existing C&SF project. Evaluation of changing landuse patterns and activities due to various restoration scenarios was also required. To do so required estimation of the average concentration of nutrients in runoff from each landuse type in the basin. An effort was initiated in year 2000 by URS Consulting under contract to SFWMD to develop a calibrated, detailed model of the Indian River Lagoon basin for the purposes of developing pollution load reduction goals (PLRGs). It was hoped that the URS model, upon completion, might provide a tool with which to perform water quality comparisons among various alternatives and to accurately predict current and future condition with and without C&SF project modifications. However, the most optimistic timeframe for completion of this effort was a minimum of eighteen months. As a result of conflicting timelines for completion and progress, the Indian River Lagoon Feasibility Study Team sought an alternative methodology to establish interim average phosphorus concentration values in runoff from specific landuse types present within the basin. 3

4 METHODS: The working hypotheses to be tested were: (1) The sum of the proportioned average concentration of total phosphorus () in stormwater runoff for each landuse within a basin could approximate the average concentration in runoff from the basin as a whole. (2) Appropriate values of phosphorus concentration in runoff from individual landuse types (Ci) could be ascertained from an iterative computer analysis which sought the best agreement (i.e., minimum cumulative error) between calculated values of average total phosphorus concentration in basin runoff (Ψ) and known values obtained from long-term monitoring programs. The formulae evaluated were thus: Ψ j = CE = 1 i 1 i 1 j AijRijCi AijRij j Ψj where, Ψ j = calculated for basin j Aij = acreage of landuse i in basin j Rij = runoff coefficient of landuse i in basin j Ci = concentration of in runoff from landuse i where, CE = cumulative error j = measured ten-year average of basin j Ψ j = calculated for basin j For the purposes of this analysis, six major drainage basins were evaluated, namely the drainage basin associated with the North and South Forks of the St. Lucie River and the SFWMD canals C-44, C-23, C- 24 and C-25. Each basin consists of "capturable" and "non-capturable" acreage. Capturable areas were defined as portions of a basin whose runoff waters could conceivably be retained and treated by some undefined CERP facility, and non-capturable areas were the portion of a basin where runoff could not be reasonably retained nor treated by any conceivable CERP facility. This analysis was confined to that part of each basin designated as capturable. The most recent available (1995) Geographic Information System (GIS) landuse coverages were queried to provide the capturable acreage of each landuse in each basin (Table 1). Table 1. Area (in acres) of each landuse in the capturable portion of each basin (from SFWMD 1995 GIS Landuse coverage maps). Landuse South North C-25 C-44 C-24 C-23 Fork Fork Forest Grove Pasture Urban Impervious Urban, Pervious Wetland Within each basin, landuses were divided into the six groups used in SFWMD s Hydrological Simulation Program - FORTRAN (HSPF), namely forest, grove, pasture, wetland, and pervious and impervious urban (T. Conboy and K. Konyha, SFWMD, personal communication, 2001). HSPF was 4

5 used to simulate in hourly timesteps runoff quantity, evapotranspiration, irrigation demand, water table elevation, and canal stage. Runoff coefficients for each landuse type within each basin over the period 1989 through 1998 are presented in Table 2. Runoff coefficients differ among the same landuses in different basins due to variations in irrigation demand and rainfall pattern. The dominant cause is rainfall variability between basins (Sculley 1986, Trimble 1990). Table 2. Runoff coefficients by landuse by basin, inches per year. Landuse South North C-25 C-44 C-24 C-23 Fork Fork Forest Grove Pasture Urban Impervious Urban, Pervious Wetland Acreages (Table 1) and runoff coefficients (Table 2) were used to calculate average acre-feet/year runoff per landuse in each basin (Table 3). Results of Table 3 were used to calculate the percent contribution of runoff from each landuse in each basin to the total runoff from each basin (Figure 2). Table 3. Runoff, average acre-feet per year by landuse by basin. Landuse South North C-25 C-44 C-24 C-23 Fork Fork Forest Grove Pasture Urban Impervious Urban, Pervious Wetland

6 Figure 2. Percent contribution of runoff (acre-feet/year) of each landuse in each basin. The objective of this analysis was to determine the set of six Ci (one for each landuse) which resulted in a calculated average basin concentration (Ψj) that best approximated the average concentration measured downstream of the capturable area of each basin. The sample collection point used to calculate the average measured basin concentration coincided with the easternmost control structure for canals C-23, C-24, C-25 and C-44. Average basin concentrations are presented in Table 4. Finding the minimum cumulative error (MCE) is proposed as an appropriate method to identify the set of six Ci that best approximate measured basin condition; the MCE being defined as the lowest observed sum of the difference between each basin's calculated concentration and its target concentration. Table 4. Ten-year (1989 through 1998) mean total phosphorus concentration (mg/l) for each basin as determined by various approaches. Storet a SFWMD b Mean Mean c Culled d SFork NFork C C C C a Includes combined SFWMD and FDEP grab data. b Data was not available for all waterbodies for these analyses. c Mean interpolated SFWMD grab and composite samples. d Calculated from samples collected only during eastward flow when discharge from Lake Okeechobee was not occurring The set of literature values of in runoff from various landuses that were deemed appropriate for use in Florida (Table 5) were employed to calculate upper and lower 95% confidence intervals for each landuse type. These upper and lower confidence intervals were used as allowable limits for acceptable Ci values. The allowable minimum for the wetland landuse was increased from mg/l to mg/l to reflect real-world expectations, as per those values observed in Everglades (Davis 1994) and Savannas Reserve (Graves et al. 1998) wetland studies. Once the allowable range for Ci for each landuses was set, an iterative computer routine was executed which sequentially explored the possible combinations of Ci within those limits (i.e., the adjusted 95% confidence interval limits given in Table 5). Each set of Ci was used to calculate average total phosphorus (Ψ) for each basin. The difference between Ψ and the observed average total phosphorus () concentration determined from water quality monitoring was summed across basins (i.e., the cumulative error). Each iteration of the computer routine produced one million distinct sets of Ci, the corresponding Ψ, and the cumulative error among the basins being evaluated. The set of Ci that produced the minimum cumulative error (MCE) for each set of input parameters (e.g., allowable ranges of Ci or monitoring-based basin concentration estimates) was identified. This process was repeated numerous times to explore response characteristics of Ci values as well as the set of monitoring data itself. Ci values that consistently went to their maximum or minimum allowable ranges were considered suspect. For brevity, results from all computer runs performed are not presented; only those runs which elucidate the process used to arrive at the final proposed set of Ci values are presented. Computer software enabling these calculations is available from the author. 6

7 Table 5. Summary statistics of literature value total phosphorus in runoff, mg/l, from Florida stormwater studies a Forest b Grove c Pasture Impervious Pervious Wetland f Urban d Urban e Mean Median Maximum Minimum Number Values Lower 95% CL g Upper 95% CL Std. Deviation a Bahk, 1997; Bahk and Kehoe, 1997; CDM Consultants, Inc., 1986; Fall, 1987; Fall, 1990; Fall and Hendrickson, 1987; Fall et al., 1995; Harper and Miracle, 1993; Hendrickson, 1985; Hendrickson, 1987; Kehoe et al., 1994; St. Johns River Water Management District and South Florida Water Management District, 1987; White, b Includes recreation, open space, undeveloped rangeland. c Includes two row crop values. d Combined single- and multi-family residential. e Combined low- and high-intensity commercial, and industrial. f Includes background Savannas marsh and Everglades concentrations. g Wetland lower 95% adjusted upward from to minimum observed (0.008). RESULTS AND DISCUSSION: The sequence of tables presented herein is intended to convey the decision processes used to arrive at the final proposed set of Ci. An initial computer run was performed employing the USEPA Storet means (Table 4) for basin targets with which to compare Ψ and calculate cumulative error, and the 95% confidence limits (Table 5) as the upper and lower bounds (Table 6). Table 6. Set of Ci which evidence minimum cumulative error (MCE=0.346 mg/l) where basin targets were USEPA Storet means (Table 4), and allowable landuse runoff minimums and maximums were the 95% confidence limits set forth in Table 5. Landuse MCE Ci Min. Max. Basin Target Calcd % Diff Forest SFork Grove NFork Pasture C UrbImp C UrbPerv C Wetland C The Ψ for South Fork, C-25, and C-44 significantly overestimated the expected Storet-based mean. In addition, Ci for forest, grove, pasture, and wetland were at their respective allowable limits. Figure 2 indicated that C-23 and C-24 possess similar landuse distributions, and the Ψ for C-23 and C-24 (Table 6) were almost identical. Subsequently, a detailed analysis of the SFWMD grab and composite sample data from C-23 and C-24 yielded a pair of concentrations (Table 4: SFWMD mean) more closely in 7

8 agreement with each other, and thus, more closely in line with the landuse runoff patterns. Using amended values for C-23 and C-24 resulted in a 12% decrease in MCE (Table 7). Table 7. Set of Ci which possess minimum cumulative error (MCE=0.303 mg/l) where basin targets were Storet means (Table 4) except C-23 and C-24 which were SFWMD means (Table 4). Landuse MCE Ci Min. Max. Basin Target Calcd % Diff Forest SFork Grove NFork Pasture C UrbImp C UrbPerv C Wetland C The Ci values in Table 7 produced Ψ for the North Fork, C-24 and C-25 that satisfactorily approximated the amended basin target values, but overestimated Ψ for the remaining basins. The Storet mean for the C-44 basin was based on samples collected at the S80 structure located at the east end of the C-44 canal. However, the C-44 canal flows westward approximately half the time, thus half of the values used to calculate the mean concentration observed at S80 were inappropriate. In addition, because the C-44 canal is used periodically to lower water levels in Lake Okeechobee, some periods of eastward flow do not reflect basin runoff but only that of Lake Okeechobee water quality (due to the tremendous quantities of water that may be discharged through the canal). In an attempt to estimate a creditable C-44 basin target concentration, an average total phosphorus concentration of mg/l was calculated using culled data (B. Gunsalus, SFWMD, personal communication, 2000; Table 4). This amended basin target for C-44 was used in all remaining analyses. The mean total phosphorus concentration for the South Fork included data from both the FDEP and the SFWMD. Many of the FDEP samples were not collected at the same sampling point as those collected by SFWMD, but were collected at a point about two miles further upstream. These samples would only represent basin runoff for that part of the basin upstream of the FDEP sampling point, and these datum should be excluded from the dataset used to calculate an appropriate basin mean. Recalculation using only SFWMD data resulted in a basin mean of mg/l for the South Fork (Table 4). However, the SFWMD sampling point was located within the tidally influenced section of the river and where water from C-44 and Lake Okeechobee could potentially mix with what would otherwise constitute South Fork basin runoff. Unlike elsewhere in the Indian River basin, approximately 73% of the groves in the C-25 watershed are on reservoirs (Figure 3; D. Smith, United States Department of Agriculture Natural Resources Conservation Service, USDA-NRCS, personal communication, 2000). Total phosphorus removal rates for wet detention reservoirs have been estimated at between 60 and 70% (Harper and Herr, 1997). Eliminating South Fork and C-25 from the analysis and employing the SFWMD-derived basin targets where available resulted in a much lower MCE (Table 8). Based upon these results, the set of basin targets (10 year mean total phosphorus concentrations derived from monitoring data) presented in Table 8 appeared suitable, and were subsequently employed for the remainder of this effort to arrive at an estimation of acceptable landuse-specific phosphorus concentration values. 8

9 Figure 3. Distribution of citrus in C-25 basin with and without onsite storage reservoirs (courtesy Donna Smith, USDA-NRCS) Table 8. Set of Ci which possess minimum cumulative error (MCE=0.037 mg/l) where basin targets were from Table 9 (SFWMD-based targets for C-23, C-24 and C-44). South Fork and C-25 not considered. Landuse MCE Ci Min. Max. Basin Target Calcd % Diff Forest SFork Grove NFork Pasture C UrbImp C UrbPerv C Wetland C The Ci values identified in Table 8 are values that resulted in the least cumulative error, provided that the various assumptions and input conditions were appropriate. However, an examination of the one hundred sets of Ci having the least cumulative error among the 1,000,000 generated reveals that grove and pasture remained constant, wetland nearly so, while the remainder exhibited considerable variation (Table 9). 9

10 Table 9. Variation in the one hundred sets of Ci values identified using the parameters established in Table 11 that possessed the least cumulative error. Cumulative error ranged from to within this interval. Landuse Ci Mean Std Dev Minimum Maximum Forest Grove Pasture UrbImp UrbPerv Wetland Numerous computer runs were subsequently performed to explore relationships among Ci. These efforts showed that in those instances where forest Ci increased, the two urban Ci decreased. Forcing the pasture Ci to a higher value did not result in a lowered grove Ci, but did result in increasing the cumulative error and underestimating the observed North Fork concentration. Further analysis indicated that it was possible to arrive at a reasonably low MCE utilizing any of the allowable Ci values for wetland, forest, urban impervious and urban pervious landuses. Conversely for grove and pasture, there existed a value or range of values where MCE was lower. The conclusion drawn was that identification of the Ci having the least cumulative error hinged on identification of suitable Ci values for pasture and grove, since both of these landuses constituted significant portions of the six watersheds and could thus be expected to contribute a substantial phosphorus load. Based on the assumption that runoff from natural, undisturbed lands would be of good quality, runoff from the forest and wetland landuses were expected to be relatively low in phosphorus. Since the formula used herein involves the product of landuse phosphorus concentration multiplied by the landuse runoff potential, the contribution of wetland and forest to the overall solution would be expected to be small. Similarly, although urban pervious and urban impervious might be expected to contribute relatively high concentrations of phosphorus to the receiving waters, the amount of urban landuse in each basin is small. Within the context of identifying a set of six Ci possessing the lowest MCE, the effect of forest, wetland, and urban Ci values was minimal compared to the effect of Ci values for grove and pasture. For example, if a pair of Ci values of grove and pasture were arbitrarily selected from among the allowable values determined by the min.-max. constraints, the "best" solution having the lowest cumulative error was resolved by identifying that set of wetland, forest and urban Ci (compatible to the selected grove and pasture Ci) that resulted in the lowest cumulative error. Because grove and pasture dominate the amount of phosphorus in basin waters, errors in identifying the most appropriate concentration value for grove and pasture landuse types resulted in corresponding but proportionally larger errors in the values assigned to the remaining four landuses. Thus, finding the most appropriate set of six Ci depends on identifying the most appropriate Ci for grove and pasture. From numerous iterations of the evaluation process, it was clear that within the best set of Ci having the least cumulative error, pasture Ci remained consistently between 0.2 and 0.3 mg/l. Grove Ci on the other hand consistently went to the maximum allowable value of the min-max constraints established by the 95% confidence interval of the literature values. Although allowing the grove Ci to increase to a value slightly above the 95% confidence interval (Table 10) resulted in only a minor reduction in cumulative error, the possibly more significant result was that the urban Ci were not maximized upon the highest allowable value, a condition assumed to be suspect. Additionally, these lowered urban values appeared more in line with FDEP runoff samples (unpublished data) collected in the St. Lucie Estuary basin, where the mean total phosphorus from the residential portion of the urban landuse was 0.23 mg/l. Residential areas constitute the majority of the urban landuse within this basin. 10

11 A further arguable justification for allowing the grove Ci to increase beyond the 95% confidence interval of Table 5, was that the grove" landuse category is actually comprised of several types of grove irrigation practices. Based upon FDEP St. Lucie basin runoff quality assessment (unpublished data), drip, jet, and flood irrigation practices resulted in differing runoff phosphorus concentrations, namely 0.216, 0.270, and mg/l, respectively. The procedure used herein must assume, given the restrictions of grove as a single landuse category, that the three irrigation practices are equally represented among all basins. This was not the case. Thus, the most appropriate Ci for grove is that particular value that best distributes the underlying error among the basins. Ostensibly, the reason that the grove Ci identifiable by this approach lies above the 95% literature-based confidence limit is the amount of flood-irrigated citrus in the North Fork basin. Raising the upper allowable range of the grove Ci resulted in a lower cumulative error, being able to acceptably estimate the observed concentration of the four basins, and a reduction in the two urban Ci values (Table 10). Table 10. The sets of Ci which possess minimum cumulative error (MCE=0.035 mg/l) where basin targets were from Table 11; South Fork and C-25 not considered. Grove allowable range altered. Landuse MCE Ci Min. Max. Basin Target Calcd % Diff Forest SFork Grove NFork Pasture C UrbImp C UrbPerv C Wetland C The values presented in Table 11 were obtained by narrowing the allowable min-max range above and below the Ci values presented in Table 10. This resulted in a slight improvement in MCE. The landusespecific total phosphorus concentration values proposed in Table 11 are within the 95% confidence interval of the published literature values that are most applicable to Florida (except for the grove landuse). The upper confidence limit for grove is mg/l while the value proposed is 4% higher (0.338 mg/l). Table 11. Set of Ci which possess minimum cumulative error (MCE=0.034 mg/l) where the allowable min-max range was narrowed around the Ci value identified in Table 10. Landuse MCE Ci Min. Max. Basin Target Calcd % Diff Forest SFork Grove NFork Pasture C UrbImp C UrbPerv C Wetland C Having identified a preliminary set of landuse runoff Ci values (Table 11) via the process described, it was necessary to evaluate the the utility of these values as being anything more than mathematical artifacts. To determine if similar results could be achieved by simply employing literature values, the mean of the values presented in Table 5 were substituted for the Ci values arrived at through the iterative estimation process described herein. Clearly, the Ci values identified in Table 11 appear to be better 11

12 suited to approximate the measured mean basin concentration than those of Table 12. The cumulative error using literature value means was over four times the error using the Ci values in Table 11. Table 12. Comparison of basin targets to calculated targets using mean of literature values presented in Table 4. MCE is mg/l (South Fork and C-25 differences not included). Landuse Lit. Mean Basin Target Calcd % Diff Forest.080 SFork Grove.232 NFork Pasture.476 C UrbImp.280 C UrbPerv.495 C Wetland.036 C The Ci values of Table 11 were also evaluated by examining the C-25 basin, which was excluded from the calculations used to arrive at Tables 10 and 11. Ten-year mean total phosphorus concentrations for C-25 were based upon samples collected at the control structure of a canal that flowed in only one direction. There was little justification to question these values, as the two means (Storet-based and SFWMD extrapolated) differed by less than 10%, well within expected analytical precision. A closer examination of the landuses in the C-25 basin revealed that 73% of the basin s citrus utilized reservoirs. If a reservoir treatment efficiency of 60-70% total phosphorus removal is assumed (Harper and Herr 1997), the corresponding C-25 grove-runoff correction factor would lie between and Calculation using grove runoff treatment efficacy factors of and 0.489, and the Ci values in Table 11 yielded an adjusted C-25 basin concentration of and mg/l, respectively. This range bracketed the SFWMD flow-adjusted mean concentration of mg/l. Assuming the calculated and SFWMD flow-adjusted mean concentration was the true mean basin concentration, a grove reservoir treatment efficiency of 62% would be required. A treatment efficiency of 62% is within the expected range for these type of reservoirs in Florida. Using the Ci values from Table 11, the calculated basin concentration for the South Fork was mg/l. This value is 23% higher than indicated by monitoring data. Among the possible explanations for this are (1) the Ci values presented in Table 11 are inappropriate for use in the South Fork, (2) the landuse acreages and/or runoff coefficients do not reflect current conditions, or (3) simple averaging of the data from South Fork monitoring data does not accurately represent basin mean total phosphorus concentration. It was not possible to definitively resolve this issue, although ongoing basin analysis and modeling may provide answers when completed. Calculations by the author indicate that if about 30% of the pasture in the South Fork were effectively functioning as forest insofar as phosphorus concentration in runoff was concerned, then the Ψ for the South Fork would be in agreement with monitoring data. Runoff quality of unimproved pasture or lightly stocked rangeland can be as good as forest runoff (Ken Konya, SFWMD personal communication, 2001). The answer may lie in how much of the pasture in the South Fork actually was functioning pastureland, or is in a fallow condition (due to various developmental and restoration pressures), and what proportion of pasture was improved versus unimproved in comparison to that proportion for the other five basins considered herein. CONCLUSIONS: The procedure described herein calculated a set of surrogate mean phosphorus concentrations in runoff from specific landuses that closely corresponded to mean basin runoff values derived from monitoring data. The values represent an improvement over a priori utilization of mean literature values. This 12

13 method of estimating landuse-specific mean runoff concentrations may be valuable in those situations where time, money, or both do not permit the detailed, lengthy modeling exercises that would otherwise be required. How well the values presented herein correspond with subsequent values derived from indepth efforts remains to be determined. However, all of the proposed values are similar to those found in actual basin sampling efforts, are in general agreement with published literature values, and appear to adequately estimate the quality of water being discharged from the basin by its six major tributaries. Thus, in the absence of competing alternatives, the proposed landuse-specific total phosphorus concentration values (as given in Table 11) were employed to estimate effects of changes in landuse due to future development and restoration efforts of the Indian River Lagoon sub-region of the Comprehensive Everglades Restoration Plan (USACOE 2001). More accurate estimates of Ci might have been achieved if the division among basin landuses more appropriately represented key basin water quality runoff potentials. There are numerous landuses within the Indian River Lagoon watershed. Grouping these landuses into general categories is necessary for modeling, but how they are grouped may constitute a compromise in the degree of characterization that is possible. Landuses and landuse practices are distributed throughout the basins, reside on various soil types, occur at various elevations and distances from canals, etc. These factors contribute to the amount of phosphorus that leaves the land and makes its eventual way to the estuaries. In the case of the Indian River Lagoon basin, the lumped citrus grove landuse category might have been better served by division into two subcategories based on irrigation type, namely grove-flood and grove-other. Similarly, division of the pasture category into improved pasture and unimproved pasture might have resolved some conundrums (e. g., in the South Fork basin), since runoff quality from these two types of pasture would be expected to differ. On the other hand, division of the urban landuse type into pervious and impervious subcategories seemed inappropriate since the total amount of urban acreage in the Indian River Lagoon basin was relatively minor. Better results might have been possible if the urban landuse was considered a single landuse type. However, it is important to note that regrouping landuse types is not a trivial exercise. In whatever way landuse groups are defined, it would then be necessary to determine not only the acreage for each but also the mean runoff potential for each landuse group in each basin. How landuse groups are initially defined for a basin should not be done haphazardly, but should be intended to reflect how the various basin landuses may affect overall conditions. During the course of this process, it became clear that drawing conclusions based on monitoring data in the absence of detailed local knowledge can lead to misleading assumptions. Although monitoring data surely tells a story, exactly what that story is can only be reliably deciphered when coupled with an understanding of how the local system operates (e.g., some canals that appear on a map to clearly flow toward the ocean in truth behaved as expected less than half of the time). This highlights the inadvisability of water quality analysts and managers drawing conclusions, with potentially far-reaching policy and environmental consequences, without having such conclusions thoroughly reviewed by the ubiquitous, yet often overlooked, local water quality experts. 13

14 REFERENCES: Bahk, B An investigation of a wet-detention pond used to treat stormwater and irrigation runoff from an agricultural basin. Southwest Florida Water Management District, Tampa, Florida. Bahk, B. and M. Kehoe A survey of outflow water quality from detention ponds in agriculture. Southwest Florida Water Management District, Tampa, Florida. CDM Consultants, Inc Boggy Creek basin non-point source water quality study. Report for South Florida Water Management District. South Florida Water Management District, West Palm Beach, Florida. Davis, S Phosphorus inputs and vegetation sensitivity in the Everglades. In: Davis, S. and J. Ogden (editors) Everglades, the Ecosystem and Its Restoration pp St. Lucie Press, Delray Beach, Florida. Fall, C Characterization of agricultural pump discharge quality in the upper St. Johns River basin. St. Johns River Water Management District, Palatka, Florida. Fall, C Characterization of agricultural pump discharge quality in the upper St. Johns river basin. St. Johns River Water Management District, Palatka, Florida. Fall, C. and J. Hendrickson An investigation of the St. Johns water control district: reservoir water quality and farm practices. St. Johns River Water Management District, Palatka, Florida. Fall, C., P. Jennings, and M. Von Canal The effectiveness of agricultural permitting in the upper St. Johns river basin. St. Johns River Water Management District, Palatka, Florida. Graves, G., D. Strom and B. Robson Stormwater impact to the freshwater Savannas Preserve marsh, Florida, USA. Hydrobiologia 379: Harper, H. and J. Herr Alum treatment of stormwater - the first ten years: what we have learned and where do we go from here? Environmental Research and Design Nov Harper, H. and D. Miracle Treatment efficiencies of detention with filtration systems. Proceedings of the 3 rd Biennial Stormwater Conference, Tampa, Florida, pp Hendrickson, J Surface water quality in areas of intensive agriculture: Putnam county. Putnam Soil and Water Conservation District. Hendrickson, J Effect of the Willowbrook Farms detention basin on the quality of agricultural runoff. St. Johns River Water Management District, Palatka, Florida. Janicki, A., D. Wade, J. Pribble, and A. Squires St. Lucie watershed assessment. PBS&J, 5300 W. Cypress Street, Suite 300, Tampa, Florida Kehoe, M, C. Dye and B. Ruston A survey of the water quality of wetlands-treatment stormwater ponds. Southwest Florida Water Management District, Tampa. 14

15 St. Johns River Water Management District and South Florida Water Management District Indian River Lagoon Joint Reconnaissance report. Contract No. CM-137, Palatka. Sculley, S Frequency analysis of SFWMD rainfall. Tech Pub DRE-226, South Florida Water Management District, West Palm Beach, Florida. Trimble, P Frequency Analysis of One- and Three-Day Rainfall Maxima for Central and Southern Florida. Tech Memo DRE-291, South Florida Water Management District, West Palm Beach, Florida. U. S. Army Corps of Engineers and South Florida Water Management District Central and Southern Florida Project Comprehensive Review Study: Final Integrated Feasibility Report and Programmatic Environmental Impact Statement. U. S. Army Corps of Engineers, Jacksonville, Florida. U. S. Army Corps of Engineers and South Florida Water Management District Central and Southern Florida Project Indian River Lagoon South Feasibility Study: Engineering Design and Modeling Appendix B. U. S. Army Corps of Engineers, Jacksonville, Florida. White, K Stormwater treatment wetlands: a functional and aesthetically pleasing management alternative for urban/recreational developments. University of South Alabama. Woodward-Clyde Consultants, Inc Physical Features of the Indian River Lagoon. Prepared for Indian River Lagoon National Estuary Program, Project Number 92F274C, Melbourne, Florida. 15