Meeting on
Salinity Performance Measures in Florida Bay
Science
Program for Florida Bay and Adjacent Marine Systems
13 July 2000
Meeting Organizer - Bill Nuttle (PMC)
Purpose
The overall goal is to recommend to the RECOVER teams a preferred, empirical
salinity performance measure in time for the next round of alternative
evaluations, which will begin in October 2000. This meeting reviewed the
salinity performance measure currently in use for Florida Bay. The current
performance measure consists of a linear regression based on water levels at
P33. This is a purely empirical approach to linking salinity in Florida Bay with
Everglades hydrology. Discussion reviewed other similar approaches to this
problem, dating back to Durbin Tabb's groundwater relationship.
Objectives
The primary objective was to outline work to construct and evaluate a new
salinity performance measure for Florida Bay. Results of recent work suggest
that it is possible to improve on the P33 relationship by using different
sources of input data and alternative approaches to constructing the empirical
model. In addition to providing an alternative to the P33 relationship, the
salinity targets associated with the current measure must also be reviewed.
Results
Discussions produced guiding principles for formulating transfer functions and
two workplans for revising the Florida Bay salinity performance measure. A
short-term work plan will provide a revised transfer function similar to the P33
relation by October 2000. A long-term work plan (covering one year) will produce
a more robust transfer function that is consistent with guiding principles.
The strong correlation between salinity in the coastal bays of Florida Bay is best explained by the strong influence of regionally coherent climate (rainfall and evaporation) rather than by sheetflow and canal flows connecting P33 with the coast. A reanalysis of the existing salinity transfer function has demonstrated that the relationship between water levels at P33 and Joe Bay salinity is not preserved under different modes of operation of the water management system. Further, the highest correlation between salinity and local rainfall or canal discharge near the coast, e.g. S18C, is obtained with a one or two month lag, which can be attributed to residence time of water in the coastal wetlands and of salt in the coastal bays. The highest correlation between P33 water levels and salinity is obtained with less than one-month lag (0 lag when using monthly-averaged data). It is improbable that the hydraulic connection between P33, which is located 30 miles inland, and the coastal bays could be more direct than that between S18C and the coast.
Short-term work
plan
The short work plan builds on work that Frank Marshall has recently completed
evaluating the P33 relationship. The results of this work provide an up-to-date
(December 1999) data set and recommendations for alternative approaches, in
addition to the results discussed above.
An ad hoc committee (Steve Davis, Dave Rudnick, and Bill Nuttle) will evaluate the results of this work and recommend whether to accept the revised P33 relationship or to reanalyze the existing data, using simple regression techniques, to produce another relationship. This ad hoc committee will report their recommendations to the PMC, which will coordinate the additional work needed, if any, and forward the results to the RET.
In general, it is desirable to base the transfer function for northeast Florida Bay on hydrological conditions at a location nearer the coast. This group will consult with Ken Tarboton on the locations where the hydrological model reproduces hydrological conditions near the coast.
Long-term work
plan
The long-term work plan differs from the short-term plan in several respects.
Most significantly, the problem of connecting Everglades hydrology with Florida
Bay salinity is broken into two steps linked by estimated freshwater flow at the
coast, Figure 1. There are two advantages of this approach. First, it relies
explicitly on freshwater flow, which is the physical link between Everglades
hydrology and salinity. Second, by breaking the problem in half, this approach
provides for the application of more sophisticated, physically based models of
Everglades hydrology and Florida Bay salinity, as these become available.
Further, this approach benefits from efforts to assemble a standard data set for
Florida Bay; this data set will include a reconstruction of historical flows
into the Bay for the period 1995 through 2000.
Figure 1:
Model form - To make the connection between freshwater flow and salinity, this work will make use of high-frequency (hourly and daily) salinity and hydrological data and it will explore the application of seasonal, autoregressive, integrated moving average models (SARIMA). Past work has used salinity data collected as monthly grab samples, which may or may not fairly characterize the central tendency of salinity in that month. The use of higher frequency data minimizes this concern, and it holds the potential of characterizing better the nature of short-term salinity variations that affect the biota in the Bay. Use of high frequency data also reduces the variation/uncertainty in the statistics derived from the data, such as monthly mean and variance.
This work will review prior work on estimating flows based on hydrological conditions in the Everglades. Examples exist of several different approaches that have been developed and applied independently.
Bill Walker divided South Florida into several wetland "watersheds" and calibrated runoff models for each one. His objective was to estimate discharge to the coast from available hydrology data. Walker's watershed models might prove useful as physically based runoff transfer functions when calibrated against observed flows.
Water budget calculations have been performed as part of the nutrient studies focused on the mangrove swamps north of the Bay.
Eric Swain has formulated linear transfer functions to relate flows measured at canal structures to flow at Taylor Slough Bridge and Taylor Slough flow to flows at the coast. These relationships were formulated using data from a limited period of time, and it remains to be seen how they perform over longer periods.
Application of
FATHOM to analyze 30 years of hydrology and salinity data used the default
transfer function that monthly inflow to Florida Bay is equal to the monthly
flows measured at Taylor Slough Bridge plus S18C minus S197. This basically
assumes no net gain or loss from precipitation and evaporation in the mangrove
swamps. Application on shorter timescales would have to account for the time
taken for flow measured at these inland points to reach the Bay.
Verification - Verification of the model will be obtained by splitting the data set into two parts. One part of the data set will be used to calibrate the model, and the second part of the data set will be used to verify the predictive power of the model.
Input variable - Flows in creeks discharging into northeast Florida Bay, monitored by USGS, provide the connection between Everglades hydrology and salinity in northeast Florida Bay. Flows measured crossing Tamiami Trail might be used in the same way for building transfer functions for salinity in the estuaries along the southwest coast.
Output locations - This work will attempt to construct transfer functions for predicting salinity variation at the following locations:
Trout Cove
Taylor river
Terrapin Bay
Garfield Bight
Little Madeira Bay
Highway Creek
Long Sound
Duck Key
Joe Bay
Manatee Bay
North River
Whipray Basin
West Lake
Guiding
Principles
Focus on flows into Florida Bay at the boundary. Discharge from the mangrove
swamp through small creeks accounts for probably 90% of total freshwater inflow.
Overland, sheetflow accounts for the remaining 10%, and groundwater accounts for
little or none of the freshwater entering the Bay.
Transfer functions should be physically based, cover the range of variation observed over the long-term (especially periods of drought), and avoid relying on a single parameter to characterize the hydrology of the Everglades (be flexible).
Transfer function must be able to capture the effects of anticipated changes in management structures - WCA III, Tamiami Trail, L31, and C111.
Global change is occurring and may have significant effects in this area. Sea level is rising, and that will continue. However, the effects of long-term trend in sea level are mitigated by substantial variation on seasonal and interannual time scales and by the effects of sediment accretion and subsidence. There is some evidence that changes in land use have decreased the annual amount of rainfall in South Florida and shifted its distribution.
Empirical models used as transfer functions can take several forms:
SARIMA (seasonal autoregressive integrated moving average) can deal explicitly with seasonality and correlation among the independent variables, characteristics found in the hydrology and salinity data that cause problems with application of regression techniques.
Regression models have the advantage of being easy to explain. Problems in the data can be dealt with through use of conditional regression, lagged variables, and seasonal variables.
Lag-1 autocorrelation models are a subset of the more general SARIMA models that are directly related to the form of time-dependent box model calculations of salinity.
SWIFT2-D and other physically based models might be useful either directly or in the development of empirical transfer functions.
Readings
Davis, S.M. 19xx. Florida Bay Performance Measure: Salinity in coastal basins estimated from upstream water stages.
Klein, J.C. et al. 1996. Salinity Characteristics of Florida Bay: A Review of the Archived Data Set. draft report (unpublished) National Ocean Service, NOAA, Silver Spring, MD.
Marshall, F.E. 2000. Florida Bay Salinity Transfer Function Analysis. final report submitted to South Florida Water Management District. Cetacean Logic Foundation, Inc.
Nuttle, W.K. 1997. Salinity Transfer Functions for Florida Bay and West Coast Estuaries. final report submitted to Everglades National Park and the South Florida Water Management District.
Nuttle, W.K. et al. 2000. Influence of net freshwater supply on salinity in Florida Bay. Water Resources Research 36:1805-1822.
Sculley, S.P. 1986. Florida Bay Salinity Concentration and Groundwater Stage Correlation and Regression. report to Jacksonville District of USACE from Water Resource Division of South Florida Water Management District.
Tabb, D.C. 1967.
Prediction of Estuarine Salinities in Everglades National Park, Florida, by the
Use of Ground Water Records. dissertation submitted to University of Miami.
Agenda
9:30 | Introduction - Bill Nuttle |
9:50 | Update on Performance Measures - Steve Davis |
10:10
|
Review of
Existing Salinity Relationships
|
11:00 |
Break |
11:15 |
Discussion - What characteristics are desired and what should we avoid? |
12:00 |
LUNCH |
1:00 |
Discussion - How to build the ultimate salinity performance measure? |
2:00 |
Discussion - How will the salinity performance measure be applied? |
3:00 | Summary Remarks |
Attendees
Name | Affiliation | |
Joe Pica | Noaa/AOML | Joseph.A.Pica@noaa.gov |
Steve Davis | SFWMD | sdavis@sfwmd.gov |
Libby Johns | Noaa/AOML | johns@aoml.noaa.gov |
Frank Marshall | CLF | femarsha@digital.net |
Bob Evans | COE | Robert.A.Evans@saj02.usace.army.mil |
Tom Armentano | NPS | tom_armentano@nps.gov |
Tom Schmidt | NPS | tom_schmidt@nps.gov |
DeWitt Smith | NPS | dewitt_smith@nps.gov |
Eric Swain | USGS | edswain@usgs.gov |
Tim McIntosh | Miami-Dade DERM | mcintt@co.miami-dade.fl.us |
David Rudnick | SFWMD | drudnick@sfwmd.gov |