Kevin Summers

U.S. Environmental Protection Agency

Gulf Ecology Division, Gulf Breeze, FL

Rick Copeland, Tom Singleton

Florida Department of Environmental Protection, Tallahassee, FL

Sam Upchurch

Environmental Resource Management, Inc., Tampa, FL

Anthony Janicki

PBS&J Engineering, Tampa, FL

Abstract

In late 1996, Florida DEP initiated an effort to design a multi-tiered monitoring and assessment program that integrated the monitoring of multiple natural resources (e.g., streams, groundwater, lakes, estuaries) with the execution of multiple programs (e.g., 305(b) reporting, TMDL establishment, ecosystem management, permitting, bio-criteria). The program is being designed in a manner that maximally provides information to other important state needs such as basin-wide assessments, development of TMDLs, and the provision of information for permitting. The design, at present, consists of three monitoring tiers focused on three spatial levels of data collection and resource assessment. Tier 1 (Status and Trends) will establish the condition of all aquatic resources in the state by broad geographic divisions (i.e., USGS Accounting Units, Water Management Districts, and Florida DEP Districts) using a probabilistic design to report 305(b) results. Tier 2 (Basin Assessments) will examine individual basins to establish environmental condition and to set TMDLs by water body using a variety of statistical designs and incorporating significant levels of "found" data. Tier 3 focuses on local conditions within a single water body to provide the information necessary to examine issues associated with re-permitting. All tiers will focus on the utilization of biological and ecological data compared to the previous reliance on physical and chemical variables. Tier 1 is scheduled to be initiated in 1999. The process by which a multi-faceted, multi-objective comprehensive monitoring plan for Florida waters is being developed and implemented is described.

Introduction

In 1996, the Florida Department of Environmental Protection initiated an effort to re-design their environmental monitoring programs in order to create a multi-resource comprehensive, integrated monitoring program (IWRM—Integrated Water Resources Monitoring) that would fulfill many of the department’s needs. These needs included 305(b) reporting, TMDL establishment, ecosystem management needs, permitting, and the development and testing of biocriteria. The design of a comprehensive, integrated approach has resulted in the adoption of a three-tiered monitoring framework that has an integrated sampling design and provides information for many of the department’s issues outlined above.

The three tiered monitoring framework (Figure 1) focuses on three spatial levels of data collection and resource assessment. Tier 1 (Status and Trends) establishes the condition of all aquatic resources in the state by broad geographic divisions (i.e., USGS Hydrologic Code Units, Water Management Districts, and Florida DEP Districts). This scale is important for the assessment of the condition of the State’s resources (e..g., to address 305(b) issues). Tier 2 (Basin Assessments) examines individual basins to establish their specific environmental condition and to set basin-level TMDLs (Total Maximum Daily Loads) by water body using a variety of statistical designs and incorporating significant levels of "found" data (historical information collected under a variety of designs and approaches). Tier 3 focuses on local conditions within a single water to provide the information necessary to examine issues associated with permitting.

To aid in the transition from earlier monitoring programs in FDEP to the proposed approach, a four-level of monitoring has been adopted that continues the collection of environmental data from a subset of historic locations at relatively short time scales (i.e., monthly). This collection level, termed the Temporal Variation Network, will ease the early interpretations (i.e., first 6-10 years) of the Status and Trends Tier particularly with regard to annual trend and seasonal variation.

The design of the Status and Trends Tier of IWRM will, for the first time, permit FDEP to answer many water-resource related questions with an unbiased, rigorous data set. The collection of information has been designed to place statistically sound confidence limits on the survey results. This design is, in part, dictated by the questions it must address.

Formulation of the questions to be addressed by IWRM Network and the Status and Trends Tier was initiated at a meeting of over 50 representatives from throughout FDEP in November 1996. A list of over 200 issues and desired outcomes of a comprehensive, state-wide monitoring plan was formulated by this group. These assessment questions ranged from site-specific or issue-based questions to broad questions related to water quality of the state as a whole. This list of assessment question was condensed into a set of 28 topics, from which the final assessment questions were drawn. These questions comprise the "roadmap" by which the success of the Status and Trends Tier will be determined.

The Status and Trends Tier monitoring design is structured to address these questions at two different scales: (1) the state as a whole and (2) large drainage basins, or drainage basin complexes within the state. The state has been subdivided into 20 of these large drainage basins (Figure 2) which are termed reporting units. The questions that the Status and Trends Tier is designed to address, therefore, relate to the status of water quality on a regional and state-wide basis. They do not address smaller drainage basins, ecosystem management areas, counties, or localities. These smaller areas are addressed by other monitoring tiers within the IWRM Network.

Addressing the assessment questions is a three-step process. First, monitoring must be accomplished following standardized protocols for data acquisition. Second, the larger "parent" population (e.g., State of Florida, Water Management District, etc.) From which the sample data were collected must be characterized in order to statistically describe the magnitude and variability of the distributions of indicators used to evaluate the water resource. Finally, the distributions are used to draw inferences about the overall status of the resource in question.

The Status Network of the IWRM Program is designed to ultimately monitor and report on all waters of the state of Florida. In order to systematically sample the many different occurrences of water, they have been subdivided into "resources". Each resource constitutes a readily identifiable occurrence of water of interest for the purposes of management. The resources that will be monitored as part of the Status Network include

· Ground water,

Scale of a water body has an effect on sampling strategy and, in many cases, management of these resources. As a result, some of the resources have been subdivided to facilitate sampling and resource evaluation. The resources and their subdivisions are discussed in the following subsection.

Ground water, as a resource, includes those portions of Florida aquifers that have the potential for supplying potable water or affecting the quality of currently potable water. Florida has three aquifer systems (Southeastern Geological Society, 1986), all of which will be sampled. These aquifers include the Surficial aquifer system (SAS), Intermediate aquifer system (IAS), and Floridan aquifer system (FAS). The ground-water resource is subdivided into two target populations for the purposes of sampling and resource characterization. These subdivisions are unconfined aquifer and confined aquifer. Typically, the SAS, which is unconfined and near the land surface, can be readily affected by human activities. Because of this vulnerability to contamination, the SAS will be randomly sampled where present. In areas where the SAS is not present and either the IAS or FAS is unconfined, these aquifers will be sampled as part of the unconfined-aquifer target population. The confined-aquifer target population includes confined portions of either the IAS and FAS, depending on which is most heavily utilized as a source of public-water supply. The rationale for sampling a confined-aquifer target population is that pumpage for municipal supply typically involves high volumes of water which may induce lateral or upward movement of saline-water intrusion. Since the effects of salt-water intrusion take many years to reverse and the resulting degradation of water quality may result in significant and costly changes in water-supply systems, the Department feels that the confined IAS and FAS should be monitored as part of the Status Network. The design for sampling groundwater resources is described below. Networks to be used include the Ambient Program’s Background Network and wells located at facilities that have been permitted by the Department. The Background Network was created to monitor background ground-water quality throughout the state, so wells were not placed in areas known, or strongly suspected, to have contaminated ground water. The areas that were excluded included coastal areas where salt-water intrusion was suspected and many heavily industrialized areas. Agricultural, residential, and local, isolated industrialized areas were not avoided.

Lakes have also been subdivided into two groups: (1) small lakes, which are 10 hectares or less in size, and (2) large lakes, which are over 10 hectares in area. This differentiation on the basis of area is intended to accommodate differing sampling strategies and methods. Small lakes will be randomly sampled from a list frame, while large lakes will be randomly selected for sampling from a grid. The details of this sampling plan are given in Section IV.

Only perennial rivers, streams, and canals will be sampled. These have been subdivided into two categories based on stream order (Horton, 1945). Wadeable streams are perennial streams of orders 1-4. Non-wadeable streams and canals include higher order streams (order >4) that are expected to require different sampling strategies than the smaller streams. Canals predominate in many areas of the state where former streams and rivers have been modified to enhance drainage. Because they require similar sampling strategies and represent master drainage systems, they are included in the non-wadeable stream category.

Florida’s estuaries and nearshore marine environments will be sampled as part of the Status Network. Estuaries are defined as coastal water bodies that are bounded upstream by head of tide and downstream by articulation with the Gulf of Mexico or the Atlantic Ocean. Nearshore waters include all coastal waters surrounding the shoreline of Florida to a point three miles from the shoreline. Sampling of estuarine and nearshore waters will be timed to coincide with the Status and Trends Tier’s rotation through reporting units (to the extent possible) so that the terrestrial and marine/estuarine sampling will be synchronous. The specific aspects of the estuarine sampling design will not be described in this manuscript.

There is a great need in Florida to include wetlands in the IWRM Program. The health of wetlands, including areas, hydrologic regimes, water quality, and biological integrity are changing from year to year. While physical and chemical criteria for wetlands exist, the Department has not adopted methods for biological assessment of wetlands. Resources do not exist to develop these criteria and include wetlands in the monitoring program at this time. Consequently, it is premature to include wetlands in the Status Network. Recognizing the need for this type of monitoring, however, the IWRM has included wetlands as a resource to be monitored.

The candidate lists of indicators to be measured as part of the IWRM Ambient Tier are presented below by water resource in Tables 1-3. These lists are considered candidate since final choice of indicators will depend in great part not only on applicability and utility of the data generated but also the feasibility. It should be recognized that meeting the published holding times ortho-phosphate, total coliform, and fecal coliform has routinely been a problem in past monitoring efforts and that such problems will likely continue into Tier I. However, a number of reviewers of the IWRM design have supported including these indicators since there is some utility in the information provided by these indicators, despite the short-coming of not meeting the holding times. The habitat assessments to be conducted will be based on those protocols accepted as part of the FDEP BioRecon procedures.

The Temporal Variability (TV) Network will include wadable and non-wadable streams as well as small and large lakes. The indicator lists for these resources will be identical to those shown above for the Ambient Tier, with two exceptions. First, some indicators such as the habitat assessment will not be repeated monthly, but likely will be limited to two time periods within the year. Secondly, discharge and/or stage data from those sites where discharge and/or stage is being monitored by USGS or the Water Management Districts will also be included.

The assessment question workshop resulted in three primary approaches: (1) Site-Specific, (2) Basin, and (3) State. The site-specific approach requires a delineation of the hypotheses to be tested by the monitoring activity (e.g., a comparison of selected areas receiving anthropogenic impacts to reference or unaffected sites). The basin and state approaches, if applicable to all waters, require the probabilistic approach. These approaches require that the boundaries of the monitored population (e.g., waters of the State, waters of a District, waters of a reporting area) be determined, acceptable uncertainty criteria are ascertained, and the appropriate design and reporting strata be determined.

The spatial and temporal aspects of a monitoring design are derived from the assessment questions and the variation associated with the selected indicators. The state-level assessment questions (and some basin-level questions) tended to call for monitoring results that apply to "all" Florida waters. A probabilistic design is required to meet this need although thousands of probabilistic options are available. The site-specific group’s questions called for results that would differentiate among selected sites or test working hypotheses. As a result, a set of judgmental sites would be required to address each hypothesis. Because both forms of questions were posed then a multi-tier design should be incorporated to include aspects of these approaches.

Uncertainty criteria must be defined and agreed upon to select a monitoring design that has the appropriate power to address the assessment questions. For example, one assessment criteria might be that all status or "health" assessments have 95% confidence intervals of ±10% such that an assessment of estuarine sediments with contaminant concentrations greater than criteria A would be X%±10% (e.g., 35±10% of all Florida estuarine sediments). This type of uncertainty pertains to probabilistic statements. Site-specific assessments also would require uncertainty criteria at primarily the level of discrimination often referred to as a p-level. For example, the uncertainty level for discrimination between affected and reference sites might be a 95% chance of discerning a difference if a difference exists between the sites.

Appropriate design strata can include many approaches. Basically, a rule of thumb is that if you wish to answer an assessment question with regard to a strata with the desired level of certainty then that strata should be designed into the overall monitoring plan. However, if the strata simply represents a geographic unit that you want information on (e.g., by habitat unit, use type, etc.) but do not care whether the design certainty level is met, then the strata should not be incorporated into the monitoring design.

In general, the use of strata within a sampling design enhances the power to detect differences because it optimizes the design based on the natural variability characteristics of what is being measured. However, in broad scale monitoring designs where many indicators are being utilized, what is optimal for one indicator is often not optimal for another. In addition, to design a monitoring plan based on strata that represents the entire resource (i.e., "all" Alabama coastal waters) requires that the physical distribution of the selected strata be known and the variability of the indicators in question be known. Often this is not the case. While much information is known concerning potential strata in Alabama waters, rarely can a known distribution be determined for all strata variables without preliminary sampling.

Several options were discussed by the working group with regard to stratification for both state-wide and Water Management District-wide monitoring. Final strata included: (1)Base geography (i.e., the State of Florida), (2) Water Management District (WMD) boundaries, and (3) Four reporting area within each WMD comprised of a single or multiple hydrologic units (HUCs).

These strata represent reasonable approaches to developing a spatial sampling design. The key to selecting the appropriate strata is a determination of the needs of our monitoring program, the availability of data on the distribution of the strata, the availability of data on the spatial variability of indicators of interest within the strata, and the ramifications of multiple strata on sampling size (i.e., reduces sampling size for site-specific monitoring and increases sampling size for ecosystem-wide sampling). Because the selected strata represent a graduated subdivision of the base stratum (State of Florida), the design needs only to be determined for the reporting units of the WMDs. Figure 2 shows the reporting units for each WMD.

The actual placement of sites and the total number of sites is also based on the assessment questions. Since many of these questions require assessments for "all" Alabama coastal waters then an element of the sampling design must be extrapolable and thus probabilistic in nature. This does not necessarily mean that the sites are randomly placed, although that type of placement is one possibility. Probabilistic simply infers that the sites are representative and not biased. If the sites can be placed judgmentally (i.e., based on experience and knowledge) so that they are representative of selected strata (e.g., habitats, use zones), then the requirement for a probabilistic nature for the design will be met. The specific protocol for the selection of sample sites for each resource type (e.g., small lakes, wadeable streams, etc.) can be somewhat different. Specific protocols by resource type are listed below under the heading, Sample Selection Protocols.

Designing the temporal aspects of the sampling plan also relate directly to the initial set of assessment questions. If the desire of the planning group is to fully characterize short-term status and trends, then monthly sampling are required. For these types of questions, the Temporal Variation Network (TV Network) was created from a subset of SWAMP sites. If the question lends itself to longer-term assessment of status and trends, then seasonal or annual samples are required. The issue with the choice of a temporal framework is not that the values of the indicators change all the time, but rather, what is the time scale of interest. If the desire of the planning groups is to "understand" coastal phenomena then shorter-time scale sampling is appropriate that includes monthly and seasonal variation. However, if the desire of the planning group is to ascertain changes in overall status and long-term trends, then annual or semi-annual time scales tend to be more useful.

Many of the proposed indicators exhibit large intra-annual variability (i.e., they are seasonal)(Oviatt and Nixon 1973, Jefferies and Terceiro 1985, Grassle et al. 1985, Holland et al. 1987). Generally, monitoring programs do not have the monetary resources to characterize this variability or to assess status in all seasons for "all" resources (i.e., all Alabama coastal waters). Therefore, sampling has often been limited to a confined portion of the year (i.e., an index period) when indicators are expected to show the greatest response to anthropogenic and climatic stress. The annual sampling sites for the Ambient Network utilize an index period for 4-8 weeks for sampling for each resource type. For example, most coastal ecosystems in the Northern hemisphere, mid-summer (July-August) is the period when ecological responses to pollution exposure are likely to be most severe. During this period, dissolved oxygen concentrations are most likely to approach stressful, low values (USEPA 1984, Officer et al. 1984, Oviatt 1981). Moreover, the cycling and adverse effects of sediment contaminant exposure are generally greatest at the low dilution flows and high temperatures that occur in mid-summer (Connell and Miller 1984, Sprague 1985, Mayer et al. 1989). The index periods for each resource type are shown in Table 4.

The assessment questions raised by the majority of the workshop participants suggest the a generalized probabilistic design for the Ambient Tier with nested designs supplementing the remaining Basin and Site Tiers. The overall design must include both ecosystem-wide annual elements based on reporting strata and collected over a five-year period and site-specific monthly elements to characterize intra-annual or seasonal trends. In addition, the design must permit an estimate of the condition of Florida’s resources each year with an enhanced estimate every five years. The designs for six of the eight resource types are described below. The two coastal designs will be determined during the next 30 day period and will be compatible with the designs for the remaining resources.

The protocol for site selection for groundwater water for two strata—confined and unconfined—is based on available information relating to established wells. The 8 step protocol is listed below.

As an example, using the Year 1 selection for Northwest Florida WMD (NorthwestC) for unconfined wells, 5 of the 30 "wells" selected were urban and are represented by a latitude/longitude that will be coupled with a permitted well, 17 of the 30 wells are known confined aquifer wells, and 8 of the 30 wells are unknown "confined aquifer" wells. Ten alternate wells have been provided in case a well is unsampleable, no permitted well exists, or an unknown well really represents the opposite stratum (in this case is really "unconfined).

The protocol for site selection for stream surface waters for two strata—wadeable (Orders 1-4) and Non-Wadeable (Orders >4)—is based on available DLGs (Digital Line Graphs) for Streams provided by USGS and from the RiverReach3 File provided by EPA.

Twenty additional random samples were selected for each stratum to be used for potentially unsampleable segments (as replacements).

The protocol for site selection for small lake surface waters—<10 hectares in surface area—is based on available DLGs (Digital Line Graphs) for Surface Waters provided by USGS and from the RiverReach3 File provided by EPA.

Twenty additional random samples (small lakes) were selected for each Reporting Unit to be used for potentially unsampleable lakes (as replacements).

The protocol for site selection for large lake surface waters—>10 hectares in surface area—is based on available DLGs (Digital Line Graphs) for Surface Waters provided by USGS and from the RiverReach3 File provided by EPA.

An additional random sample was selected for hexagonal space and coupled to the original sampling site for each Reporting Unit to be used for potentially unsampleable locations (as replacements). As this design is spatially dependent, only the coupled alternative site can be used in the event of a unsampleable location.

The overall state design provides for collection for all Reporting Units comprising the Florida within a five-year period (1999-2003, 2004-2008, etc.). One Reporting Unit from each Water Management District is selected randomly to be sampled twice in the five-year period—resulting in five Reporting Units for each WMD (three units once and one unit twice). One of the "five" Reporting Units from each Water Management District is selected randomly for each sampling year. The only constraint on the random selection is that the same Reporting Unit cannot be sampled two years in a row. The distribution of Reporting Units throughout the first five-year cycle in shown in Table 5. The total number of samples by resource type of each Water Management District over the five year cycle is shown in Table 6.

The ecosystem-wide surveys are supplemented by intensive surveys conducted monthly at 80 locations throughout Florida and represent all resource types. The purpose of these sites is to fully characterize the hydrograph at the location. This characterization will include all seasonal and short-term variation that could make interpretation of population trends difficult in early years. This difficulty is often due to changes in indicators that are related to climatic shifts and variation in seasonal lengths. In general, these issues are easily ascertained in population-level data after 6-10 years. In the interim, the temporal variation network will provide additional information on short-term variability for each resource.

Florida’s IWRM Network represents an early adaptation by a state of probabislistic approaches to collect the information necessary to meeting its 305(b) requirements. In addition, IWRM will use its sampling networks to address other important issues to Florida including basin assessments, the determination of total maximum daily loads, the allocation of those loads, and long-term permitting issues. IWRM is presently expanding its resource base to include estuarine resources and expects that this monitoring activity will initiate through the Florida Marine Research Institute in 1999. In 1998, the Gulf of Mexico program initiated an effort to help the Gulf States adopt a core, comprehensive, integrated coastal monitoring approach that includes probabilistic sampling as one of its elements. Early discussions by this group have praised Florida’s efforts and described their program as a framework upon which a Gulf-wide consistent surface-water monitoring program could be based.

In 1996, EPA’s 305(b) Working Group described probabilistic sampling approaches as an acceptable approach for collecting environmental condition data and reporting 305(b) results. At present, four states are adapting their overall water resources monitoring programs to utilize probabilistic surveys as an element of their overall programs.

Literature Cited