0202 agu kerfoot

Slide1: 

Vertical Migration of a Toxic Karena brevis Red-tide and the Impact on Ocean Color Remote Sensing Reflectance John Kerfoot1, Kevin Mahoney2, Gary Kirkpatrick3, Steve Lohrenz2, Mark A. Moline4, Oscar Schofield1 1Inst. Of Marine and Coastal Sciences, Rutgers University, New Brunswick, N.J. 08901, 2Dept. Of Marine Science, University of Southern Mississippi, Stennis Space Center, MS 39529, 3Mote Marine Laboratory, Sarasota, FL 34236, 4Biological Sciences Department, California Polytechnic State University, San Luis Obispo, CA 93407 PLATE 2: Karenia brevis (Gymnodinium breve) 1) Increase the accuracy of methods to deconvolve bulk inherent optical properties (IOPs) into individual constituents, with an emphasis on improving methods to derive phytoplankton spectra. 2) Combine different optical measurements to increase the utility of multivariate pattern recognition methods. 3) Determine the proportion of the population visible by a satellite based on the vertical migration behavior of cells. METHODS: The physical and optical dataset was collected on the R/V Suncoaster during the October 2001 EcoHAB Florida Process cruise on the West Florida Shelf. A bloom patch was located on October 22 and a radio telemetry drifter was deployed to follow the position of the patch. Beginning at 05:00 on October 23, an intensive profiling schedule was undertaken, using an optical package (Plate 3) fitted with a WetLabs ac-9, Satlantic downwelling irradiance (Ed) sensors, Satlantic upwelling radiance (Lu) sensors, a HOBILabs HS-6 backscattering sensor, a HOBILabs a-beta absorption meter, a HOBILabs c-beta attenuation meter, and a Wetlabs HiSTAR meter. A HyperTethered Spectral Radiometric Buoy (Satlantic, Inc.) was deployed at every station during the day as long as the sun was high enough above the horizon to allow for accurate measurements of the light field. The optical package was profiled once every hour for the 24 – hr period (25 casts total), while CTD 10/19/01 - 10/26/01 FIGURE 1: EcoHAB Cruise Area a (m-1) 0.03 0.09 0.15 0.18 b (m-1) 0.35 0.55 0.75 1.15 0.95 0.25 0.50 0.75 1.00 c (m-1) 05:00 08:00 11:00 14:00 17:00 20:00 23:00 02:00 05:00 10/24 Time of Day Given this, there is a critical need to develop effective sampling approaches that provide HAB-specific information over relevant temporal and spatial scales. During the October 2001 Ecology of Harmful Algal Blooms (EcoHAB) process cruise, we collected a full suite of IOP measurements for a natural population of Karenia brevis over a diel (24 hour) cycle. This poster outlines the potential for remote sensing of Karenia brevis blooms, utilizing this suite of measurements. Harmful Algal Blooms (HABs) (plate 1) occur worldwide and result in adverse economic and health effects for coastal societies. Blooms of the toxic dinoflagellate Karenia brevis (Gymnodinium breve) (Plate 2) occur in the Gulf of Florida every year. Estimates of economic losses due to these bloom events vary from millions to tens of millions of dollars per episode (Steidinger and Vargo, 1988; Tester et al., 1991), and the fishing and tourism industries are the most severely affected. Plate 1: Harmful Algal Bloom (HAB) 10/24 bb/b (m-1) 0.009 0.012 0.015 0.018 PLATE 3: The optics package used to measure changes in the inherent optical properties of the K. brevis bloom over the course of the 24 hour period. casts were performed every two hours. In addition, a prototype benchtop in situ phytoplankton detector (Gary Kirkpatrick, Mote Marine Lab, Sarasota, FL) was used to track K. brevis spectral absorption as well as colored dissolved organic matter (CDOM) absorption. Water Mass Characteristics: The CTD data from the profiles taken during the EcoHAB October 2001 cruise are displayed the T-S diagram in Figure 2. There appeared to be 3 separate water masses: 1) inshore water with low salinity and temperature, 2) offshore water with high salinity and high temperature and, 3) an intermediate water mass. FIGURE 3:  Karenia brevis is positively phototactic and negatively geotactic, so cells migrate to the surface during daylight hours (Evans, 2001). This behavior is evident in coulter counts of cell numbers (Figure 3A) in the upper 2 meters of the water column. Time course measurements of the IOP’s (Figure 3B, 3C, 3D) inside the bloom patch show K. brevis cell numbers increasing at the surface during the day, and also vertical migration to depth at night. This figure displays B) the attenuation coefficient c(676), (m-1), C) the absorption coefficient, a(676), (m-1), and D) the scattering coefficient, b(676) (m-1), measured in the water column over the 24-hour cycle. Maximum surface values of c, a, and b occured in the late afternoon (about 16:00 EDT).  Figure 3A also displays a time-course plot of the absorption coefficient, a(l), at 440nm. Absorption at this wavelength is proportional to CDOM concentration. As would be expected, there is a lag in the buildup of CDOM in surface waters which follows the migration of the bloom population to the surface waters. In addition, even as the population begins to migrate down into the water column just before sunset, a(440) remains high. Cells L-1 (x105) 0.0 0.4 0.8 1.2 1.6 BULK INHERENT OPTICAL PROPERTIES 10/23 FIGURE 2 FIGURE 4 Spectral backscatter, bb(l) (m-1), is an important quantity for inverse modeling of in-water constituents and in determining the nature of particle assemblages (Cullen et al. 1997). Figure 4 is a time course plot of the ratio of bb(676) to overall scattering, b(676). Ratios of bb/b can reflect particle size as well as the refractive index of the suspended particles. Low values of bb/b can be characteristic of larger particles. In this case, it reveals an increase in size of K. brevis cells during the day as the population migrates to the surface. Another significant feature of Figure 4 is the increase in bb(676)/b(676) values in the early morning hours of 10/24 at depth. Larger values are characteristic of smaller particles and, in fact, studies of the cell cycle of this dinoflagellate performed concurrently on this cruise reveal that cell division took place around 12 a.m. local time, plus or minus about one hour (Tod Leighfield, NOAA, personal communication). Optical Backscatter, bb(l) Optical approaches show much promise in mapping the distribution of phytoplankton and will be useful in monitoring HABs (Cullen et al. 1997). While promising, these approaches have been criticized because they provide only bulk composite signals for a water mass and the signatures for distinct phytoplankton species are difficult to detect (Garver et al. 1994). To allow these new optical tools to reach full potential, it is necessary to: Temperature (° C) Salinity (P.S.U.) 24 25 27 28 26 34 35 36 24-Hour Station 0.20 0.25 0.30 0.35 a(440) (m-1) PARTICLE CHARACTERISTICS 10/23 (1) inshore (2) offshore (3) A) Karenia brevis cell abundance CDOM absorption @ 440nm Attenuation (m-1) @ 676nm 1 3 5 7 9 B) Depth (m) Absorption (m-1) @ 676nm 1 3 5 7 9 C) Depth (m) Scattering (m-1) @ 676nm 1 3 5 7 9 11 D) Depth (m) 0.06 0.12 1.25 05:00 08:00 11:00 14:00 17:00 20:00 23:00 02:00 05:00 bb(676)/b(676) Time of Day Depth (m) 1 3 5 7 9 REMOTE SENSING REFLECTANCE 05:00 08:00 11:00 14:00 17:00 20:00 23:00 02:00 05:00 Time of Day 10/24 10/23 Maximum bb(676)/b(676) inside the bloom 0.02 0.00 0.01 What Does an Ocean Color Satellite See? FIGURE 7 This figure displays the calculated ratio of Rrs(490)/Rrs(555) measured with the HyperTSRB. This quantity was chosen because standard SeaWIFS processing algorithms (O’Reilly 1998) use this ratio to estimate surface chlorophyll a. The vertical movement of the K. brevis bloom from depth to the surface over the day resulted in increased blue absorption relative to green, which led to a decrease in the ratio. The population had reached the upper meter of the water column by 4 p.m. (EDT), thus measured Rrs(l) at this time is inaccurate due to the position of the bloom relative to the Lu sensor on the buoy (Figure 7C). By 4 p.m., the solar zenith angle was too high for an accurate downwelling irradiance spectra to be measured. However, we want to look at Rrs(l) as the bloom moved back down the water column after sunset .…. So, is it possible to model Rrs(l) using the measured IOPs a(l), b(l), c(l), and bb(l) to solve the radiative transfer equation? One model used to investigate this question is Hydrolight® (Sequoia Scientific, Inc. Redmond, WA). FIGURE 8 is a simple schematic on using Hydrolight ®. FIGURE 9 plots the measured (Figure 7B) vs. modeled Rrs(l) using a(l), b(l), c(l), and bb(l) as inputs. There are significant differences in the green and yellow wavebands. The red peak in the TSRB spectra (inset) is solar stimulated fluorescence picked up by the Lu sensor, but not resolved by Hydrolight. Modeled AOP’s using IOP datasets in Lake Michigan and the LEO-15 site provide excellent agreement between measured and modeled values. With this in mind, we used Hydrolight to predict Rrs(l) as the bloom patch migrated to depth at night. Modeling Apparent Optical Properties (AOPs) a(l) b(l) c(l) (z) IOPs AOPs Rrs(l) Time of Day & Location MEASURED Rrs(l) VS. MODELED Rrs(l) FIGURE 8 Cell division FIGURE 10 Displayed above are the Hydrolight modeled Rrs(l) spectra. The most apparent feature of these spectra is the “greening”, or shifting of the peak towards the green/yellow portion of the spectrum, of Rrs(l) as the K. brevis population migrates to the surface. The peak in Rrs(l) then shifts back towards the blue wavebands as the population swims back down into the water column in the evening. The effect is more effectively demonstrated in the time course plots of Rrs(l) (inset). ACKNOWLEDGEMENTS: Many thanks to the crew of the R/V Suncoaster for assisting in gathering of this dataset. Thanks also to Tod Leighfield (NOAA) for his insights into the Karenia brevis cell cycle and Trisha Bergman for help with processing the TSRB data and discussions of Hydrolight. Funds for this project were provided by The Office of Naval Research Hyperspectral Coupled Ocean Dynamics Experiments (HyCODE) and the Ecology of Harmful Algal Blooms (EcoHAB) project. REFERENCES: Garver, S.A., D.A. Siegel, and G.B. Mitchell. 1994. Variability in near-surface particulate absorption spectra: What can an ocean color imager see? Limnology and Oceanography, 39: 1349-1367. O’Reilly, J., S. Maritorena, B. Mitchell, D. Siegel, K. Carder, S. Garver, M. Kahru, C. McClain. 1998. Ocean color chlorophyll algorithms for SeaWIFS. JGR, 103:24937-954. Steidinger, K.A. and G.A. Vargo. 1988. Marine dinoflagellate blooms: dynamics and impacts. In C.A. Lembi and J. R. Waaland (eds.), Algae and Human Affairs. Cambridge University Press, New York. pp. 373-401. Tester, P.A., R.P. Stumpf, F.M. Vukovich, P.K. Fowler, and J.T. Turner. 1991. An expatriate red tide bloom: transport, distribution, and persistence. Limnology and Oceanography, 36:1053-1061. Evans, T.J., G. Kirkpatrick, D. Millie, D. Chapman, and O. Schofield. 2001. Photophysiological responses of the toxic red-tide dinoflagellate Gymnodinium breve (Dinophyceae) under natural sunlight. Journal of Plankton Research, 23(11): 1177-1193. Cullen, J.J., A.M. Ciotti, R.F. Davis, M.R. Lewis. 1997. Optical detection and assessment of algal blooms. Limnology and Oceanography, 42: 1223-1239 FIGURE 5 This figure plots the maximum values of bb(676)/b(676) within the migrating population as determined from the measured IOPs (Figure 3). A significant increase in the bb(676)/b(676) ratio was observed within this population in the late night hours of 10/23 and the early Morning hours of 10/24. This is consistent with the time of cell division obtained from concurrent experiments performed on the cruise. FIGURE 6 This figure displays median cell diameter (mm) of K. brevis cells before and after the observed cell division event at ca. 0:00 on 10/24. A decrease in median cell diameter of ca. 20% was observed. These measurements, while not easily resolveable (except perhaps for the bb(676)/b(676), do provide some insight into cell size within the bloom; however, more work is needed to provide definitive proof. 19 20 21 18 Median Cell Diameter (mm) 23:00 22:00 24:00 1:00 10/24 21:00 20:00 19:00 Cell division INTRODUCTION 1) Inherent optical properties (IOPs) such as a(l), b(l), c(l) and bb(l) can be used to track the development and movement of a vertically migrating phytoplankton population. Applications include longterm deployment of drifters, moorings, AUV’s, etc., for real time and pre-emptive monitoring of HABs. CONCLUSIONS 2) Satellite derived chlorophyll a values rely, in part, on remote sensing reflectance in the green wavebands. The vertical migration of Karenia brevis resulted in “greening” of the modeled Rrs(l) spectrum to varying degrees over the course of the migration cycle. Since Rrs(l) is used to estimate surface chlorophyll a, the timing of the satellite overpass is critical to accurate estimations of surface chlorophyll a. This poster can be viewed online at http://marine.rutgers.edu/cool/coolresults/agu2002 4mm 5mm 6mm 7mm 0.002 0.003 0.001 0 0.2 0.4 0.6 0.8 9am 10am 11am 12pm 1pm 2pm 3pm Rrs(490)/Rrs(555) Measured Rrs(490)/Rrs(555) SeaWIFS Chl a (O’Reilly et al., 1998) Chl a = -0.04 + 10[0.341 – 3.001x + 2.811x2 – 2.041x3] x = log10[Rrs(490)/Rrs(555)] Ed Lu Rrs ~ Lu/Ed 1 m Rrs(l) (sr-1) 0.001 0.002 0.003 0.004 4mm 5mm 6mm 7mm Measured Modeled October 23 @ 12p.m. EDT y = 0.53 + 0.0006 R2 = 0.82 y = x 0.001 0.002 0.003 0.004 0.001 0.002 0.003 0.004 FIGURE 9 Modeled Rrs(l) (sr-1) Measured Rrs(l) (sr-1) 0 0.001 0.005 0.004 0.003 0.002 450 500 550 600 650 10/23 @ 09:00 10/23 @ 12:00 10/23 @ 16:00 10/23 @ 17:00 10/23 @ 18:00 10/23 @ 20:00 10/24 @ 00:00 Modeling Remote Sensing Reflectance A) B) C) Measured vs. Modeled Rrs(l) Rrs(545) Examination of the absorption spectrum for K. brevis (Figure 11) reveals the reason for this effect. Pigments in K. brevis (and other dinoflagellates) cells absorb highly in the violet and blue, but absorption drops off considerably in the green and yellow region of the spectrum. When cells are concentrated at the surface, more blue light is absorbed relative to green, resulting in an Rrs(l) spectrum dominated by green wavelengths. 3) Optical closure is very promising; however, further progress must be made in order to resolve the impact of surface slicks on in situ measurements. K. brevis absorption, a(l) (m-1) FIGURE 11 9 12 16 18 20 0 3 17 Rrs(505) Rrs(525) Rrs(535) Rrs(555) Rrs(545) Rrs(l) (sr-1)