Florida’s Coral Reef Water Quality Data

This project was motivated by the need to easily understand water quality patterns at different spatial and temporal scales, and to ultimately assess the effect of efforts to improve water quality locally. The need to aggregate and visualize data from different observing programs, and an analysis of water quality hotspots and data gaps, was identified among the management goals of the Florida Keys National Marine Sanctuary (FKNMS) and as a priority within the sanctuary’s Water Quality Protection Program (WQPP), co-chaired by the US Environmental Protection Agency (EPA) and the Florida Department of Environmental Protection (DEP). We proposed addressing these needs by aggregating various historical and current water quality data sets, analyzing spatial and temporal trends, and evaluating compatibility between monitoring protocols to establish a foundation from which existing monitoring programs can be adapted to better inform regulatory frameworks. 

Efforts to map and analyze water quality data collected over the past 20+ years will help in planning for coral restoration, tourism, and land discharge management. 

We initially identified over 80 potential water quality monitoring programs throughout South Florida by approaching partners within our professional network, examining past research permits within the FKNMS that mentioned water quality parameters, and systematically searching the WIN/STORET and SEACAR databases. We filtered out programs that did not conduct sufficient sampling (10 years minimum within the FKNMS and 5 years minimum north of the FKNMS to Martin County); that were substantially beyond the geographic extent of the area of interest (the Florida Reef Tract from Monroe to Martin County); and that did not sample most of the parameters of interest (Chlorophyll-a, Temperature, Salinity, Nitrate+Nitrite (NOx), Soluble Reactive Phosphorus (PO4), Silica (Si), Turbidity, Total Nitrogen (TN), and Total Phosphorus (TP)).

Ultimately, we identified four compatible programs for the purpose of examining hotspots and trends based on our initial criteria and an internal QA/QC process: 

1. the South Florida Ecosystem Restoration Cruises (Walton Smith), 
2. the Southeast Environmental Research Center Water Quality Monitoring Network (SERC),
3. the Miami-Dade County Department of Environmental Resources Management Water Quality Monitoring Program (DERM), and 
4. the Broward County Water Quality Monitoring Program.

This story map illustrates the water quality patterns found within each of these four programs and demonstrates how satellite imagery can complement these programs. To identify “hot spots” where water quality may be worsening over time (e.g., where Turbidity is “increasing faster” over time), time series were extracted from each sampling location and assessed using a seasonal Mann-Kendall test following the methods in Millette et al. 2019. This test estimates the Theil-Sen slope, or the rate of change of a water quality parameter over the period that data were collected. To ease interpretation, we categorized the Theil-Sen slope, or rate of change, as generally increasing or decreasing.

Turbidity Trends by Program

Total Nitrogen Trends by Program

Nitrates and Nitrites (NOX) Trends by Program

Silicate Trends by Program

Total Phosphorus Trends by Program

Chlorophyll-a Trends by Program

Satellite imagery

Using 18 years (2003-2020) of ocean color satellite data from the Moderate Resolution Imaging Spectroradiometer (MODIS-Aqua) sensor, a gridded set of monthly satellite images covering the Florida Keys was created using four MODIS-Aqua satellite products related to water quality: chlorophyll-a concentration (an estimate of phytoplankton biomass), the diffuse attenuation coefficient at 490nm (a proxy for water clarity), light absorption by colored dissolved organic matter (an estimate of blue light absorption by CDOM), and reflectance at 667nm (a proxy for suspended sediments).

Shallow water effects can be mitigated by looking at patterns over time as opposed to absolute magnitudes and accounting for seasonality during the Mann-Kendall analysis. The following maps illustrate the rate of change indicative of worsening water quality conditions (i.e., increasing trends in chlorophyll-a, nutrients, and turbidity) from satellite imagery products.

Generally, changes in water quality trends over time differed between inshore and offshore areas across several parameters and monitoring programs. For the most part, there were greater trends and changes in water quality calculated for nearshore stations/regions versus offshore. This held true both near the Florida Peninsula and near the Florida Keys. However, the waters connecting these two regions on the southwest Florida shelf, had smaller changes in water quality.  

Some of the trends observed in the satellite imagery in the Lower Keys and Biscayne Bay agree with those seen in the SERC and DERM. Furthermore, when we compare chlorophyll-a time series from the Walton Smith cruises to satellite derived estimates at the same coordinates, complementary trends emerge despite the increased variability from in situ observations. In general, chlorophyll-a concentrations are higher in winter months, particularly offshore of the reef tract. Closer to shore, patterns are less clear due to influences from land.

We could not combine several datasets simply because of the time required to reformat different databases for the purposes of bringing data into a common framework.

1) Databases use different naming conventions for basic information, and it greatly increases the time required to effectively work with similar datasets. The most common information that differed between monitoring programs, in terms of formatting, was site coordinates, sampling dates, program names, units, and nutrient names. 

• Solution: Agree to common naming conventions among existing programs, and/or create code that automatically renames datasets to a common framework moving forward. 

2) Station names between, and within, monitoring programs can be inconsistent and/or illogical. 

• Solution: Use unique names with a reference key and easily accessible metadata. For example, “year sampled” should not be in the station name. 

3) Stations are rarely sampled at the exact same location in repeated visits, but still need to be considered discrete locations for subsequent analyses. 

• Solution: Provide coordinates as the average of site coordinates or define a local fixed coordinate to represent the general sampling site in subsequent years. 

4) The time frame(s) of interest can differentially weight trends in subsequent analyses, and monitoring programs do not always overlap in temporal sampling frequency.

• Solution: Time periods of interest need to be clearly defined by management questions to compare and contrast regional water quality trends more accurately. 

5) Some programs sample year-round and others only sample in the summer. For example, the EPA’s Environmental Monitoring and Assessment Program (EMAP) was an ideal candidate for subsequent hotspot analyses but could not be aggregated because sampling only occurred in the summer. 

• Solution: Hotspot analyses and other pattern recognition techniques require quarterly sampling at minimum and more frequent sampling is preferred.