projects > monitoring sub-aquatic vegetation through remote sensing > work plan
Project Work Plan
Greater Everglades Science Program: Place-Based Studies
Project Work Plan FY 2003
A. GENERAL INFORMATION:
Project Title: Monitoring Sub-Aquatic Vegetation through Remote Sensing: A pilot study in Florida Bay
Other Investigator(s): Charles Holmes
Project Summary: Seagrass beds are essential components of any marine ecosystem because they provide feeding grounds, nurseries, and habitats for many forms of marine life, including commercially valuable species; they are important foraging grounds for migratory birds; and they anchor sediments and impede resuspension and coastal erosion during storms. This valuable natural resource has been suffering die-offs around the world in recent years, yet the causes of these die-offs are undetermined. The purpose of this project is to use a number of tools - geographic, geologic, and biologic - to investigate the causes of seagrass die-offs and to develop methods that can be used to monitor the health of seagrass meadows. If we understand the causes of the die-offs and can easily monitor the health of seagrass beds, then resource managers have a tool for forecasting areas of potential die-offs. This pilot study will focus on Florida Bay, a region that suffered the loss of 40,000 ha of turtle grass in a die-off event that began in 1987, and a small, localized die-off in 1999. These events were well documented and provide a baseline for testing methods of monitoring grass beds remotely. Remote sensing data, including aerial photos and satellite imagery data, and data extracted from sediment cores will be used to examine the long-term sequences of events leading up to seagrass die-off events. An understanding of the sequence of events that precede die-offs is critical for resource managers, so further losses can be effectively prevented, and damaged systems can be restored.
Project Objectives and Strategy: The objectives of this pilot study are to develop a methodology for monitoring spatial and temporal changes in sub-aquatic vegetation using remote sensing data, including satellite imagery and aerial photography, and to analyze potential causes of seagrass die-off using geographic, geologic and biologic tools. The ultimate goal is to develop a method for forecasting potential sea-grass die-offs and to determine if remediation efforts would be cost-effective.
Florida Bay is selected for the pilot study because the thorough documentation of the 1987-1988 die-off event provides a baseline for examining data preceding and succeeding the event. In addition, a small well-studied die-off occurred in 1999-2000 at Barnes Key in Florida Bay. A 10-15 km2 portion of Florida Bay that encompasses areas affected by the 1987 and 1999 die-offs will be analyzed for this pilot study. Current remote sensing data acquired for this area will be used to test different platforms, to determine detection limits, and to isolate distinct signals for different types of vegetation. Once ground-truthing and remote sensing data analysis is completed, archived remote sensing data can then be used to examine the sequences of events leading up to the die-offs. The remote sensing data can be compared and compiled with the data collected by seagrass biologists in 1987 and 1999, and to sediment core data collected at the sites of seagrass die-off. Sediment cores provide a long-term perspective on changes in nutrient geochemistry, substrate, water chemistry (salinity, temperature, oxygen), and changes in the biota. The geologic, biologic and remote sensing data will be integrated and analyzed to determine the patterns of change and sequences of events that occur in healthy seagrass beds and in beds undergoing a die-off.
Historically, several different remote sensing techniques have been used to map the location of seagrass habitats in multiple geographic locations throughout the world. These methods have produced varying results, but have demonstrated the successful use of visible and near-infrared wavelengths for mapping seagrass habitat. Few studies have employed a multiple sensor approach and no known studies have utilized the newest high-resolution (0.6m to 4m ground sample distance) panchromatic and multispectral satellite systems which are currently available. Analysis of remote sensing data from multiple systems will be applied in this study to determine the ideal wavelengths and spatial resolution necessary to detect and monitor the health of seagrass beds. The sensors to be tested will include Landsat 7 Enhanced Thematic Mapper, the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), Quickbird high resolution sensor, and large-scale aerial photography. If available, experimental data from NASA's EO-1 HYPERION hyperspectral sensor and the multispectral Advanced Land Imager (ALI) will also be acquired and compared.
Potential Impacts and Major Products: By integrating remote sensing data, biological data and core data we can examine the long-term (decadal-scale) sequences of events leading up to die-off events. These data can be contrasted to normal seasonal changes that occur in healthy grass beds to establish criteria for identifying areas that may be on the threshold of experiencing a decline. This provides a very powerful predictive tool for resource managers. By examining the causes of die-off and the natural patterns of change in seagrass meadows over biologically significant periods of time we can determine the components of change that may be related to anthropogenic activities versus natural cycles of change. This information would allow resource managers to make informed decisions about the cost-effectiveness of and mechanisms for remediation, if an area of decline was identified via the predictive tool. Once the predictive tools and potential remediation tools have been developed in this pilot study, in well-studied seagrass meadows, the tools can be applied to threatened coastal ecosystems around the country and worldwide.
Primary products will include a GIS map product compiling all the data from the project for the study area, a journal article compiling the results from the tasks, and a general interest publication for resource managers and the public explaining the methodology and findings. Successful development of a repeatable methodology for mapping the location and health of seagrass beds and an increased understanding of the long-term factors affecting seagrass health will be the anticipated products of the research.
B. WORK PLANTitle of Task 1: Evaluation of Remote Sensing Methods for Seagrass Detection
Task Funding: : Land Remote Sensing Program
The subaquatic vegetation (SAV) in Florida Bay includes four species of true grasses and dozens of species of macrobenthic algae. Water depths range from near 0 on top of the shallow mud banks, where SAV can be emergent during low tides, to 3-4 meters in the basins and channels. The primary focus of the study is the seagrass Thalassia (Turtle Grass), which has suffered the extensive die-offs, and the effect that large mats of macro-benthic algae (primarily Chondria, Laurencia and Polysiphonia) may have on the health of Thalassia. We will investigate different methods for discriminating types of SAV from remote sensing data.
There are several challenges to discriminating types of SAV through the use of remote sensing. The main difficulties include the spatial size of the seagrass beds, which range from several square meters to hundreds of square meters, the depth of the seagrass below water and other vegetation such as the floating algal mats. Remote sensing data in the blue and green wavelengths will penetrate water and near-infrared bands may assist in resolving different types of vegetation. It is expected that through a combination of blue, green, near-infrared bands, and higher spatial resolution, it will be possible to conduct detailed mapping of seagrass beds. Remote sensing data from sensors with lower spatial resolution, but a higher spectral resolution may assist in characterizing the health and environmental conditions of the seagrass.
While other studies have successfully used Landsat data to map seagrass habitat, the course spatial resolution of the data (30m-multispectral) has limited the results to mapping only large meadows and not distinguishing seagrass from other types of SAV. Landsat ETM+ will be used in this study to develop small-scale seagrass habitat maps and to develop an index of chlorophyll values for these areas, which may be translated into a measure of seagrass health. The ASTER sensor, which lacks a band in the blue wavelength, will be analyzed to determine if the higher spatial resolution (15m) data assists in the delineation of smaller seagrass beds. In addition, both ASTER and Landsat have thermal infrared capabilities, which will be tested to examine surface temperature of habitat areas. If available, HYPERION data would represent the highest spectral resolution (with over 200 bands from 0.4- 2.5 micrometers) and would provide a unique opportunity to conduct a detailed spectral analysis of seagrass habitat. The Quickbird sensor captures 2.5m visible and near-infrared data and 0.61m panchromatic data. This sensor will be used for high resolution mapping of seagrass beds and will be tested to determine if SAV types can be discriminated from one another using spectral and spatial techniques. Natural color aerial photography will be interpreted to map seagrass through the classification of habitat types and their density. The comparison of satellite sensors to aerial photography and ground truth data will provide the basis upon which remote sensing data is evaluated. A combination of sensor types to maximize both spatial resolution and spectral signatures may provide the best solution for mapping and monitoring seagrass.
Work to be undertaken during the proposal year and a description of the methods and procedures: The proposed work consists of the acquisition of remote sensing imagery, aerial photographic interpretation, imagery classification, ground truth data collection (task 2), and the comparison and evaluation of results.
Data from the selected remote sensing systems will be acquired for a specific time period in FY03. Preferably, all imagery will be collected on the same date/time to ensure environmental variables are constant for all images and will be closely correlated with field mapping (task 2). If this is not possible, due to tasking and weather limitations, the best possible combination of imagery and image dates will be acquired. These data sets include Landsat 7 ETM+, ASTER, Quickbird, Natural color, 1:12,000 scale aerial photography, historical photography and, if available, a test image of HYPERION hyperspectral data.
Digital image processing will be performed on the Landsat, ASTER, and Quickbird imagery using portable spectrometers and field observation data (task 2). Several processing and enhancement methods will be applied including, conversion from sensor radiance to absolute reflectance prior to processing. Images will be created based on spectral responses of seagrass, water and other environmental materials as well as statistical significance tests, where applicable. The goal of the image processing is to develop the best possible band combination and processing technique for each sensor for the purpose of mapping seagrass habitat in Florida Bay. Once processed, image classes, which correspond to those used in the aerial photography interpretation step, will be developed for each sensor used. Ancillary information will be compiled for the study area that will include the bathymetry, tidal range, water temperature, and salinity. These data layers will assist in the processing and evaluation of the satellite.
The aerial photography mapping protocol consists of stereoscopic photo interpretation, cartographic transfer, and digitization of seagrass habitat in accordance with previously used seagrass classification systems. Other important aspects of the protocol include the development of a classification system, validation with ground control data, quality control, and peer review. The information derived from the photography will subsequently be transferred using a Zoom Transfer Scope onto a stable medium overlaying USGS 1:12,000 scale quadrangle basemaps. Habitat classification done through the aerial photography interpretation will be stored as a geographic information system (GIS) data layer to be used for remote sensing correlation analysis. Historical photography will also be analyzed to evaluate spatial size and location related to historical die-off events. We will begin gathering these historical photographs in FY03, but interpretation will probably take place in FY04.
The resultant image classifications of each sensor system will be correlated against both the independent ground truth data (task 2) and the aerial photography interpretation habitat classes for an assessment of accuracy. The different methodologies and image processing techniques will be evaluated based on their correlation with the habitat maps and ground truth data. The procedures will be documented and explained to show the comparison of sensor systems, band combinations, spatial resolution and correlation with ground truth.
Planned Outreach: Primary product generation will be geographically registered and geometrically corrected satellite and aerial photography of Barnes Key and Rabbit Key. From these sources GIS map layers will be developed showing locations and spectral signatures of seagrass beds. A remote sensing and aerial photography seagrass habitat map, bathymetric map, spectral profiles of seagrass habitat, GPS coordinates of seagrass site locations, and correlation tables for sensors will be developed. These map products will be used in field assessments and tested for accuracy. Successful development of a repeatable methodology for mapping the location and health of seagrass beds in Florida Bay will be the anticipated product of the research. This method will aid researchers in monitoring and analyzing environmental variables, which may affect seagrass health. Information and data will be presented to clients and colleagues at meetings and workshops, and will eventually be compiled for a journal article and a publication aimed at resource managers.
Task Summary and Objectives: The primary objective of this task is to provide the ground-truthing data for interpretation of the remote sensing data and aerial photographs. Ground truth data collection will be performed to coincide as closely as possible with imagery collection. Ground truth data consists of date/time, water temperature, water depth, tidal fluctuations, type of subaquatic vegetation, size and density of seagrass beds and substrate for individual seagrass beds within the study area. Turbidity and chlorophyll levels in the water will be measured. Ground control data will include spectral signatures of seagrass habitats acquired through a portable spectrometer. All ground truth data will be linked to geographic reference coordinates developed from global positioning systems (GPS) receivers. The secondary objective of this task is to collect information on seasonal changes in SAV and responses to environmental variables, such as salinity, water temperature, nutrients, etc. These data will provide the background information for interpreting the changes seen in the historical data (remote sensing, aerial photographs, and core data) by providing insights into the causes of changes in SAV distribution.
Work to be undertaken during the proposal year and a description of the methods and procedures: Field mapping in the study area will commence in FY03. Three to six test sites (at scales ranging from 1m up to a maximum of 90 m2) will be selected within the study area (Figure 1) that represent different environments, habitats and water depths typical of the area. These sites will be selected based on preliminary field observations and examinations of historical and current aerial photographs.
Investigations at each test site will include the following work. 1) All submerged aquatic vegetation (SAV) within the test areas will be mapped using GPS units. The perimeter of the grass beds and any associated barren sections will be measured and documented. Over the course of this project these boundaries will serve as a baseline to determine any expansion or contraction of vegetation that may occur. 2) The SAV within the test area will be identified to species level and a permanent reference collection of samples established. After the initial identification and sampling process is complete, the plant biomass density will be measured and documented by using a grid system and counting plant stems and/ or shoots within a random sampling of the grid layout. 3) Portable spectrometer readings will be made over areas of different types of SAV to determine the spectral signatures associated with each type. These data will be critical in interpreting the remote sensing data. 4) Water quality measurements will be taken including; salinity, temperature, turbidity, chlorophyll content, PH, dissolved oxygen, ammonia, nitrates, nitrites, etc. at each test site. 5) Water depth, sediment depth, and substrate type will be documented at each measured site. 6) Selected test sites will be monitored on a regular basis (4-8 week intervals) in order to establish seasonal variability within the submerged plant communities. The increases and decreases in abundance of various groups of SAV and the phytoplankton densities within the water column can then be established.
Planned Outreach: This task is part of James Murray's graduate student thesis at George Mason University (Dr. Ted Bradley, Advisor); data and results will be presented in talks and posters prepared for scientific meetings and workshops. A seasonal SAV species list will be developed and made available on the Ecosystem History of Florida Bay web site (Wingard) (http://sofia.usgs.gov/flaecohist/). Collections that are created will be archived and become part of the George Mason University Herbarium collection, which currently houses more than fifty thousand specimens and is used by the USGS, the Smithsonian Institution and many other universities. Results from this task will be compiled and incorporated with data from tasks 1 & 2 to prepare a journal article, posters, and general interest publications, some of which will be aimed at resource managers.
Task Summary and Objectives: The objective of this task is to examine the relationship between seagrass and environmental factors over a time-scale of years and decades by using sediment cores from two known and well-documented areas of seagrass die-off, Barnes Key and Rabbit Key. These cores preserve a record of the substrate, salinity, nutrient supply, and the fauna and flora present at a site. By examining the link between environmental factors and seagrass over an extended period, we can test hypotheses about the causes of seagrass die-off. These hypotheses include 1) changes in salinity; 2) changes in light availability; 3) changes in water temperature; 4) nutrient availability; 5) disease; and 6) increases in atmospheric dust (aerosols). These data will be compiled with the analysis of historical aerial photos and remote sensing data (task 1) to determine 1) if any distinctive patterns in the sequence of events can be detected in the cores, the aerial photos, and or the remote sensing data, and 2) in hindsight, did the aerial photographs and/or remote sensing data pick up some signal to the beginning of the stressors that could be used as a predictive tool to monitor seagrass beds for future problems.
Work to be undertaken during the proposal year and a description of the methods and procedures: In FY03 we will process two cores collected in June 2001 from Barnes Key and Rabbit Key. Sampling and analyses of cores will follow methods established by the Florida Bay Ecosystem History Project (Brewster-Wingard, USGS, GD; Orem, USGS, GD). Cores will be dated using 210-Pb geochronology (Holmes, USGS, GD). Faunal and floral assemblage analyses will be conducted on 2-cm segments of the cores from Barnes and Rabbit Key; these data will be interpreted using a large dataset on the biology and ecology of modern Florida Bay fauna and flora (Wingard, Ecosystem History of Florida Bay Project; http://sofia.usgs.gov/flaecohist/). The presence of seagrass-indicator species (primarily molluscs in FY03) will be quantified in down-core samples. In FY04 additional analyses can be done including 1) downcore geochemical analyses to evaluate past changes in nutrients (C,N,P,S); 2) the occurrence of lignin phenols (seagrass biomarkers); 3) shell geochemistry (salinity and temperature); 4) stable isotopes (salinity and sources of water); and 5) atmospheric dust (aerosols). The analyses of the indicator-species and the shell geochemistry work will be an extension of ongoing and completed ecosystem history work in Florida Bay that has focused on historical salinity patterns (Brewster-Wingard and others).
Planned Outreach: Data from cores will be compiled, analyzed, and presented as a poster or talk at conferences and meetings with clients. Additionally, all data will be made available on line at http://sofia.usgs.gov/flaecohist/. An Open File Report will be produced on the cores and distributed to clients and cooperators. When the core analyses have been completed and the data compared and compiled with the historical aerial photograph analyses and remote sensing (task 1), a journal article for a scientific publication will be produced.