Harmful Algal Bloom Forecast Project

Application of Remote Sensing to Red Tide Forecasts in the Gulf of Mexico: Proceedings of a Workshop

Workshop Findings

With the eastern Gulf of Mexico as a model system, a variety of remote sensing platforms and instruments were evaluated to determine the advantages and disadvantages of each for use in an operational HAB forecast system. An operational system would require continuous coverage and would be developed to provide data to predict, verify, and improve the parameterization of the biological and physical processes in a forecast model.

Remote Sensing Platforms

Satellite Platform:

Satellite imagery provides spatial data at scales of hundreds to thousands of kilometers with resolution varying from meters to kilometers; a specific area will have repeat coverage that varies from less than one to greater than six days, depending on the satellite orbit. This spatial scale of information is warranted because oceanographic features that initiate G. breve blooms and the blooms themselves occur on kilometer scales. Satellite monitoring may be most useful to detect the conditions suitable for bloom development and to track the progress of a bloom as it moves in from offshore. This technique could allow prediction of when and where the coastal region would be affected by an HAB.

Ocean Color: The Coastal Zone Color Scanner (CZCS), which was active from 1978 to 1986, proved that discoloration of the water caused by chlorophyll a present in phytoplankton is detectable by satellite. The substantial CZCS and Ocean Color and Temperature Scanner (OCTS: Sept. 1996 to June 1997) archives serve as data sources for historical information on the frequency, duration, and spatial extent of phytoplankton blooms on the Florida shelf. The ocean color satellite, Sea-viewing Wide Field-of-view Sensor (SeaWiFS), launched in August 1997, is essential for current and future forecasting capabilities. Satellite coverage would be vital as an early warning for bloom initiation and development in offshore areas. SeaWiFS provides normalized water-leaving radiance at five key wavelengths in the visible region of the spectrum. Derived products include chlorophyll a concentration, as a phytoplankton biomass parameter (Figure 3), and attenuation coefficients, as indicators of water clarity. Preparations are being made by the U.S. and foreign space agencies to launch additional ocean color sensors during the next several years.

Advantages: SeaWiFS images provide data with a horizontal scale on the order of 10³ km with resolution of 1 km; repeat coverage of an area occurs approximately every 1 to 2 days. Future ocean color satellites will have limited information available at a resolution of 250 m (Moderate Resolution Imaging Spectroradiometer: MODIS) or less (Coastal Ocean Imaging Spectrometer: COIS).

Disadvantages: Data in the vicinity of clouds are questionable.

There currently is no species-specific algorithm to determine G. breve concentrations from water-leaving radiance. G. breve is not the only species found in the area; therefore, high chlorophyll a concentrations do not necessarily indicate a harmful bloom.

Only surface characteristics are visible to a satellite ocean color sensor. The current limitation of 1-km resolution affects the ability to accurately monitor nearshore environments and to discriminate phytoplankton patches that are less than 1 km in size.

Currently, SeaWiFS has a two-week embargo on data availability that limits real-time access to data unless commercial prices are paid for this access. This situation is not critical to development efforts but access to data and the ability to distribute products in real-time would be needed to support an operational system.

Proposed Solutions: Several approaches to solve the species-specific algorithm problem are possible. These include approaches that are based on the optical signature of G. breve (cf. Millie and others 1997) and the use of future ocean color sensors that are hyperspectral or detect at lower wavelengths (e.g. Global Imager).

Sea Surface Temperature: The Advanced Very High Resolution Radiometer (AVHRR) has been used to identify physical oceanographic features by detecting differences in surface water temperature (Figure 2). For example, northern intrusions of warm Loop Current water onto the Florida shelf and nearshore upwelling can be detected in AVHRR imagery. The ability to identify features that concentrate and transport phytoplankton can be used as an early warning of conditions conducive to bloom development. The mechanism of transport of HAB species to atypical regions has been traced using AVHRR imagery (Tester and others 1991; Keafer and Anderson 1993).

Advantages: Data processing is automated and occurs in real-time. Data access occurs four times daily and is relatively inexpensive. In the near future, there will be SST products available from the Geostationary Operational Environmental Satellites (GOES) which will yield SST information at 4-km resolution every hour.

Disadvantages: The uniformity of warm waters throughout the Gulf of Mexico in the summer months prevents clear identification of oceanographic features. Data are invalid under cloudy conditions. Spatial resolution is 1 km and nearshore data may be questionable due to pixels affected by the signal from land.

Proposed Solutions: Physical oceanographic features are detectable using sensors with other wavelengths. Ocean color provides information on the location of fronts (Muller-Karger and others 1991); altimeters provide information in the presence of clouds.

Sea Surface Height: Imagery from microwave altimeters aboard the TOPEX and European Remote Sensing (ERS 2) satellites can be used to identify physical oceanographic features by detecting differences in surface water height. The ability to identify features such as geostrophic currents that concentrate and transport phytoplankton can be used as an early warning of conditions conducive to bloom development.

Advantages: Data processing is automated and occurs in real-time. Imagery is available regardless of cloud conditions. Accuracy is high (error in height is less than 13 cm).

Disadvantages: Repeat coverage is limited (3 to10 days). Spatial resolution is 3 to 5 km, depending on frequency, and is less useful in coastal areas.

Proposed Solutions: Images can be analyzed with SST to increase temporal resolution. Spatial resolution can be augmented using SAR imagery.

Surface Waves and Fronts: Ocean currents and fronts can be identified by synthetic aperture radar (SAR) that detects surface roughness. This imagery would allow identification of physical oceanographic features that concentrate and transport phytoplankton and could aid in detection of bloom initiation and transport. This type of satellite could identify the Loop Current in summer months when the temperature signal is obscured by surrounding warm waters. These microwave sensors are available on ERS 1 and 2 and the Canadian RADARSAT satellite.

Advantages: Resolution is high (10 to 100 m), and the data are unaffected by clouds.

Disadvantages: Data are expensive to acquire and may not be available in near real-time. Interpretation of imagery is still in the research phase. Atmospheric wind signatures often obscure oceanographic phenomena. Rapid repeat coverage (3 to 5 days) can be maintained for only short periods without incurring large data costs because of the data policy agreement.

Proposed Solutions: Programs can apply to have repeat coverage of an area for a specified amount of time. An example of this type of program is StormWatch, a program coordinated by Johns Hopkins University, Applied Physics Laboratory, to monitor nearshore coastal wind fields using SAR imagery during winter months on the northeast coast.

Wind Fields: Wind fields may be most useful in determining the direction and velocity of bloom transport. Large-scale wind patterns can provide information critical to determining the movements of surface ocean features. Winds are suspected to be a major factor in the demise of blooms by mixing the water column, thereby disrupting the physical environment. Data are available from the Advanced Microwave Instrument (AMI) aboard ERS 1 and 2 and from the Special Sensor Microwave/Imager (SSM/I).

Advantages: Precision of satellite-derived wind speeds is better than 2 m s^-1. The distribution of SSM/I winds is operational and therefore readily available and inexpensive; global coverage is available within 2 days. AMI global coverage occurs within 4 days.

Disadvantages: Resolution for imagery is 50 km and is invalid for areas less than 50 km from the shoreline. SSM/I provides wind speed only; AMI provides both speed and direction.

Proposed Solutions: Nearshore wind fields with 1-km resolution may be derived using SAR imagery for limited areas and duration. A network of Coastal-Marine Automated Network (C-MAN) buoys and other meteorological stations maintained along the coast of Florida can be used to supplement satellite-derived wind fields.

Aircraft Platform:

Aircraft instruments provide spatial data at the meter to kilometer scale with resolution on the scale of a meter or less; temporal resolution is variable depending on flight schedules. Aircraft monitoring may be most useful to detect large-scale features nearshore where satellite data are often invalid because of interference by continental aerosols and terrestrial features. Because they are flown at low altitude, aircraft-based instruments are relatively unaffected by cloud cover; although some atmospheric correction would be required.

Aircraft can be equipped with several instruments for simultaneous measurement of several variables. If the systems are automated and user friendly, they could be mounted to aircraft-of-opportunity such as the Coast Guard helicopters that routinely fly over coastal regions of the Gulf of Mexico. Targeted, local areas could be monitored routinely using instruments mounted on aircraft.

Ocean Color: Ocean color can be measured using passive optical systems such as the Ocean Data Acquisition System (ODAS) and SeaWiFS Aircraft Simulator (SAS) that measure radiance at selected visible wavelengths. These systems have been used previously in nearshore areas such as the Chesapeake Bay (Harding and Itsweire 1990). Imaging instruments are available that collect swath data and have been used in, for example, Barkley Sound (Gower and others 1988). Examples are the Airborne Visible Infra-Red Imaging Spectrometer (AVIRIS) and the Compact Airborne Spectrographic Imager (CASI).

Advantages: Radiometers are relatively inexpensive and provide quantitative data. Spatial resolution is on the order of 5 m.

Disadvantages: As with satellite-derived ocean color data, currently there is no species-specific algorithm for G. breve. Similarly, only surface characteristics are visible to an aircraft ocean color sensor; therefore, sub-surface populations of cells would not be detected. Imaging systems are expensive and require complex atmospheric and geometric correction algorithms.

Proposed Solutions: Progress for species-specific algorithms may be improved with hyperspectral or lower wavelength sensors.

Laser Fluorosensing: An alternative to passive ocean color measurements, active laser fluorescence (Light Detection and Ranging [LIDAR]) has been used to detect phytoplankton by measuring chlorophyll pigment fluorescence (Hoge and Swift 1981). LIDAR can be modified to detect fluorescence of colored dissolved organic material (Vodacek and others 1995) and phycoerythrin-containing plankton such as cyanobacteria (Sathyendranath and others 1994), which could prove useful in the detection of Trichodesmium spp. Some reports indicate that toxic dinoflagellate blooms are often preceded by or occur with blooms of cyanobacteria (Steidinger and Baden 1984; Paerl and Bebout 1988) Laser fluorosensing also has been used to measure the beam attenuation coefficient (Bristow and others 1981; Hoge and Swift 1981). This method has been used successfully by the remote sensing group at NASA, Wallops Island.

Advantages: This method is useful as an additional or alternative parameter in areas where ocean color algorithms for phytoplankton biomass are affected by variable amounts of colored dissolved material (i.e., in optical case II waters). Recent improvements that make the instrument smaller and lighter will improve its usability.

Salinity: The Scanning Low Frequency Microwave Radiometer is a remote sensing system (Gooberlet and others 1997) that measures salinity to within 1 ppt and sea surface temperature to within 0.5 ºC. Temperature is required to correct microwave measurements and obtain accurate salinity data; Spatial resolution depends on aircraft altitude and is typically on the order of 100 m. A salinity map could address bloom demise due to freshwater influx.

Advantages: This is the only remote sensing instrument that is capable of providing salinity data on large spatial scales. It is useful for validating and refining large-scale circulation models in and near estuaries.

Disadvantages: Fog prevents accurate measurement of sea surface temperature that is required to correct measurements for salinity calculation.

Mooring, Buoy and Ground Platforms:

Instruments on moorings and buoys provide spatial data at the meter to kilometer scale with resolution on the scale of 1 m or less; temporal data can be obtained on scales that can vary from hours to months with resolution of seconds or less. Moored instruments can provide vertical profile data; these data are not available readily from any other remote sensing platform. Moorings have the potential of being equipped with several instruments for simultaneous measurement of several variables. Instruments mounted on drifters are essential for the development and verification of area circulation models.

Forecast models will require continuous monitoring for timely data acquisition. Data collection is relatively impervious to weather conditions, in particular to cloud cover. Moorings may be the most efficient means for continuous data collection from targeted local areas. The National Data Buoy Center (NDBC) maintains a system of buoys that monitor wind speed and direction, air temperature, and atmospheric pressure above the surface and water temperature at the surface for locations along the coast and at selected offshore sites. The University of South Florida also maintains buoys in the area. The NOAA Coastal Ocean Program’s Ecology and Oceanography of Harmful Algal Blooms (ECOHAB) project has plans to establish 12 moorings in strategic locations in the Gulf of Mexico.

Ocean Color: (see summary by Cullen and others 1997) Instruments moored at depth can provide information on the sub-surface distribution of phytoplankton populations by measuring irradiance and absorption coefficients. A chain of irradiance sensors allows calculation of spectral diffuse attenuation coefficients. Information on irradiance propagation below surface and on spectral absorption and beam attenuation coefficients is useful for the interpretation of surface signals that are measured by aircraft and satellites.

Advantages: Irradiance, absorption, and attenuation sensors have the advantage of quantitative measurement. These sensors are all available commercially at prices ranging from thousands to tens of thousands of dollars depending on the number of sensors required and the equipment options specified. Absorption and beam attenuation coefficient sensors are valid for night-time data collection. Data quality is not subject to atmospheric conditions.

Disadvantages: Data are not synoptic. Moored instruments will require periodic servicing to analyze instrument performance, and reduce contamination of data due to biofouling. All moorings must be clearly marked to avoid unintentional damage by passing ships. These platforms are subject to manipulation by passers-by.

Winds, Temperature, Salinity, and Currents: Conductivity, temperature, and depth sensors provide information allowing the detection of water masses. Depth profiles provide information on the depth of the mixed layer and allow detection of subsurface water masses. Current meters provide information on the speed and direction of water mass movement. This information is necessary for validation and refinement of circulation models.

Advantages: Sensors are widely used and readily available from commercial sources. Measurements are direct and quantifiable.

Disadvantages: Moored instrumentation will require periodic servicing to download data, analyze instrument performance, and reduce contamination of data due to biofouling. All moorings must be clearly marked to avoid unintentional damage by passing ships. These platforms are subject to manipulation by passers-by.

Waves, Fronts, and Currents: The Ocean Surface Current Radar (OSCR) system is a series of high frequency (25.4 MHz) radar stations that has been developed by the Rosenstiel School of Marine and Atmospheric Science to resolve surface current fields (Shay and others 1995; 1998). Spatial coverage is approximately 45 km with a resolution of 1 km for the High Frequency (HF) mode of OSCR (Figure 4). Temporal scale is variable; surface currents can be mapped approximately every 20 min for use in understanding particle accumulation and dispersal in surface current convergent and divergent regimes. The measurements also can be used to analyze the importance of tidal and subinertial flows in the transport of blooms. This instrument may be most useful in monitoring the currents transporting blooms along shore.

Advantages: OSCR provides measurements of currents nearshore where satellite imagery is inaccurate or invalid.

Disadvantages: Instruments require a secure location to prevent damage during operation.

Proposed Solutions: Use of the OSCR system can be targeted to strategic dates and areas. State parks or federal lands provide a secure location for instruments.

Ship Platform:

Point-and-profile data are of limited spatial and temporal coverage; however, these data are critical for the validation of remote sensing algorithms. Ships are essential for rapid response to an HAB event and can be directed to an area of interest indicated by other remote sensing data. Historical data primarily are point-and-profile data. Attempts to statistically analyze remote sensing data in combination with historical data to determine long-term changes in bloom frequency and intensity will require that remote sensing data be validated by methods comparable to those used for historical data collection.

Access to a ship and water samples allows for the development of techniques that are considered to be in the research phase. These techniques often require a discrete water sample and intensive analysis, but may prove useful to an automated remote sensing-based forecast program in the future. Many of these techniques involve the identification and enumeration of G. breve cells as a constituent of the phytoplankton population (Van Dolah and others 1994). Research has begun on the discrimination of the optical signature of G. breve from the optical signature of a mixed phytoplankton assemblage (Millie and others 1997). Additionally, the carotenoid pigment gyroxanthin-diester is being investigated as a quantitative indicator of G. breve. This pigment is consistent within G. breve and has not been detected in other species in the eastern Gulf of Mexico (Millie and others 1995). Absorption in the ultraviolet region of the spectrum also may be useful as an identifying feature for toxic dinoflagellates (Kahru and Mitchell submitted).

Shipboard measurements also would be required to measure carbon and oxygen production by G. breve. Productivity analyses require discrete water samples; however research using active (e.g., Falkowski, Greene, and Kolber 1994) and passive (Kiefer, Chamberlin, and Booth 1989; Chamberlin and others 1990) fluorescence measurements may prove useful to a remote sensing program in the future. Other studies that would benefit from access to a ship platform are those that investigate the nutrient dynamics of G. breve blooms (e.g., Vargo and Howard-Shamblott 1990).

Advantages: Ships may be the most flexible platform to schedule in a timely manner for rapid response to an event. Ship platforms allow multiple instruments to be deployed.

Disadvantages: Spatial and temporal scales for sampling are small. Operational sampling on a scale that would be useful for daily predictions is unlikely.

Proposed Solutions: Ships of opportunity may be available. These ships provide the ability to sample at low cost as well as serve as an opportunity for public education and involvement. Institutes in the west Florida area (e.g. Mote Marine Laboratory) already conduct routine survey transects.

Data Integration and Modeling Techniques

The "Marine Biotoxins and Harmful Algae: A National Plan" (Anderson, Galloway, and Joseph 1993) states as a weakness that there are no predictive models for population development, transport, and toxin accumulation for any of the major harmful algal species in the US. Predictive models, which would be necessary to implement a forecasting capability, require integration of basic research and long-term monitoring measurements to validate parameterization.

A method for data storage and integration suggested in the National Plan is a geographic information system (GIS). This type of system links spatially-related oceanographic data collected by remote sensing methods with in-situ measurements and with other spatial data including geology, municipal land-use patterns, and plant and animal habitat classifications. This type of data storage and analysis system will be useful for both research purposes and for management purposes such as prioritizing mitigation and control responses to predicted HAB events.

Numerical models are one mechanism to integrate data for physical and biological components for the purposes of prediction. Numerical models have been designed to describe oceanic circulation in limited spatial areas and have been used in the prediction of oil spill trajectories (Galt 1994). These models would be used based on the premise that tracking and predicting the movements of the water mass would track and predict the movements of an HAB. They are complex models in nearshore areas that must account for variation in wind, tides, temperature and salinity and topography.