ESTIMATION OF INHERENT OPTICAL PROPERTIES AND WATER CONSTITUENT by daw95820

VIEWS: 9 PAGES: 30

									   ESTIMATION OF INHERENT OPTICAL
PROPERTIES AND WATER CONSTITUENT
 CONCENTRATIONS FROM THE REMOTE-
SENSING REFLECTANCE SPECTRA IN THE
  ALBEMARLE-PAMLICO ESTUARY, USA
       Leonid Sokoletsky and Ross Lunetta


       U.S. Environmental Protection Agency
      National Exposure Research Laboratory
        Landscape Characterization Branch
    Research Triangle Park, North Carolina, USA

        2nd MERIS – (A)ATSR Workshop
             Frascati (Rome), Italy
             September 22-26, 2008
                     Project Participants
                      and Consultants

    • Leonid Sokoletsky, Ross Lunetta, Joseph Knight, Yang Shao (U.S.
        Environmental Protection Agency)
    •   Alexander Kokhanovsky (Institute of Environmental Physics,
        Iniversity of Bremen, Germany)
    •   Jayantha Ediriwickrema (SRA International)
    •   Darryl Keith (Atlantic Ecology Division, U.S. Environmental
        Protection Agency)
    •   Hans Paerl, Michael Wetz, Benjamin Peierls (Institute of Marine
        Sciences, University of North Carolina at Chapel Hill)
    •   Anatoly Gitelson (Center for Advanced Land Management
        Information Technologies, University of Nebraska-Lincoln)
1
                   Research Objectives

    • Develop simple underwater algorithm for estimation of water
     quality components (CDOM, Chl a, VSS, FSS, and TSS) from
     measured underwater remote-sensing reflectance spectra based on
     radiative transfer theory

    • Develop simple atmospheric correction algorithm based on radiative
     transfer theory

    • Parameterize and integrate both algorithms for water quality
     satellite monitoring above the Albemarle-Pamlico Estuary, USA



2
    Fig. 1. The Neuse River-Pamlico Sound Estuarine
    System with ModMon stations and FerryMon routes.

3
                         In situ measurements
            ModMon Project:
• Radiometric quantities [spectral downwelling
 irradiance just above the surface, Ed(λ, 0+);
 underwater spectral upwelling radiance, Lu(λ, z)]
 achieved from the ship measurements (Satlantic
 HyperOCR Hyperspectral Radiometer, Fig. 2)


• In-water quality parameters: Chl a, CDOM,
 volatile (organic) and fixed (inorganic) particles
 estimated from the laboratory analysis               Fig. 2. Satlantic Hyperspectral
                                                      Ocean Colour Radiometer
                                                      (HyperOCR).
• Auxiliary parameters: T, S, Secchi depth, PAR
 attenuation coefficient, turbidity etc.


   4
                   In situ measurements (cont.)
   FerryMon Project:
  Chl a and auxiliary parameters (T, S, pH, dissolved oxygen, turbidity)
 measured from two ferries (one in Neuse River, one in Pamlico Sound), Fig. 3


Fig. 3. Ferry work. Taken from:
http://www.unc.edu/ims/paerllab/
research/ferrymon/how.htm




  5
6
    Fig. 4. In situ remote-sensing reflectance spectra in Neuse River.
                Remote sensing measurements




    Fig. 5. Imaging Spectrometer MERIS on board ENVISAT. Taken from
    http://www.mumm.ac.be/Assets/OceanColour/Pages/MERIS_sensor.gif
7
    Table 1. The band center wavelengths λ (in nm) and calibration coefficients
     k(λ) = LTOA(λ)/DN(λ) (in μW cm-2 sr-1 DN-1) for the MERIS instrument
     (after Barnes and Zalewski, 2003; Govaerts and Clerici, 2004)



          Channel           λ            k       Channel     λ         k
             1            412.5        0.01683      8      681.25    0.00404
             2            442.5        0.01436      9       709      0.00359
             3             490         0.01111     10      753.75    0.00305
             4             510         0.00995     11       760      0.00299
             5             560         0.00753     12       779      0.00282
             6             620         0.00545     13       870      0.00223
             7             665         0.00434

8
                                  LTOA(λ) θv = 0º




                                               θ0
                                                    Ed(λ, 0+)
                                  Lu(λ, 0-)

                                       θrefr

    Fig. 6. A schematic geometry for in-water and atmosphere processes and
    measurements.



9
10   Fig. 7. Relationships between the water quality parameters in Neuse River.
                        Main assumptions of the study

     aTSS, red = a1Chl = a2VSS = a3FSS = a4TSS; TSS = VSS+FSS
      TSS,

     avis = aw, vis + aTSS, vis + aCDOM, vis; aNIR = aw, NIR + aCDOM, NIR; bb, red ≈ bb, NIR = bb;
                       TSS,        CDOM,                        CDOM,

     aCDOM, blue = a5(Ggreen/Gred)a6; aCDOM(λ) = aCDOM(412.5)(412.5/λ)a7,
      CDOM,

     G(λ) ≡ bb(λ)/[a(λ)+bb(λ)], λ = 412.5 nm (blue), 560 nm (green), 665 nm (red),
             709 nm (NIR);
     Rrs(λ, 0-) = Lu(λ, 0-)/Ed(λ, 0-) = (1/π)[dESA + (1 - dE)RFcos(θrefr)] (current
      study)
     SA = F1(IOPs), RF = F2[IOPs, cos(θrefr)] by
           Kokhanovsky and Sokoletsky (2006);
     dE ≡ Ed, dif/Ed = F(θ0, λ) by Højerslev (2001, 2004)
     Rrs(λ, 0-) = Rrs(λ, 0+)/[0.52+1.7Rrs(λ, 0+)] by Craig et al. (2006)

11
                    Water quality in situ algorithm

Chl = aTSS, red/a1; VSS = aTSS, red/a2; FSS = aTSS, red/a3 ; TSS = aTSS, red/a4;
      a3 = a2a4/(a2 - a4), where
aTSS, red = (aw, NIR + aCDOM, NIR + bb)(GNIR/Gred) - (aw, red + aCDOM, red + bb), where
aCDOM, blue = a5(Ggreen/Gred)a6, aCDOM(λ) = aCDOM(412.5)(412.5/λ)a7,
G(λ) = 8.945Rrs(0-) – 37.98[Rrs(0-)]2 + 140.2[Rrs(0-)]3 – 286.7[Rrs(0-)]4 ,
bb = aNIRGNIR/(1-GNIR)




 12
                             Atmospheric correction algorithm
     According to Gordon and Voss (1999):
                 Lw, TOA (λ )         LTOA (λ ) − Lpath, norm (λ ) μ0                          ⎧ ⎡τ (λ )           ⎤       ⎫
     Lw (λ ) =                    =                                     , T dif , up (λ ) = exp⎨− ⎢ R    + τ a (λ )⎥ / μ v ⎬, μ v ≡ cos θ v
                 Tdif , up (λ )                T dif , up (λ )                                 ⎩ ⎣ 2               ⎦       ⎭

      After Hansen and Travis (1974):

                 τ R (λ ) = 0.008569λ−4 (1 + 0.0113λ−2 + 0.00013λ−4 ), λ in μm.
      After Haltrin (1998):
                                        0.08
                     ⎛ 0.745 ⎞ τ 0
      τ a (λ ) = τ 0 ⎜       ⎟ , where τ 0 ≡ τ a (0.745 μm).
                     ⎝ λ ⎠

        After Sekera (1970) and Yang and Gordon (1997) (reciprocity principle):

                                                                                        [
        E s , dif ( λ ) = E d , TOA ( λ )Tdif , down ( λ ), Tdif , down (λ ) = Tdif , up (λ )        ]μv / μ0
                                                                                                                .
        After Højerslev (2001, 2004):
13      E s (λ ) = E s , dif (λ ) / d E (λ ).
Fig. 8. Relationships between Gordon’s parameter and Rrs(0-).
Fig. 9. Absolute errors of TSS absorption at the red spectral range under
using two approximations: aCDOM = 0 and QSSA: in-water algorithm.
                                        QSSA           algorithm
Fig. 10. Absolute errors of TSS absorption at the red spectral range under
using two approximations: aCDOM = 0 and QSSA: remote-sensing algorithm.
                                        QSSA                 algorithm
17   Fig. 11. Predicted vs. in situ aCDOM(412.5) in Neuse River (2006-2008).
18
     Fig. 12. Predicted vs. in situ Chl a in Neuse River (2006-2008).
19
     Fig. 13. Predicted vs. in situ VSS in Neuse River. 2006-2008.
20
     Fig. 14. Predicted vs. in situ FSS in Neuse River. 2006-2008.
21
     Fig. 15. Predicted vs. in situ TSS in Neuse River. 2006-2008.
22   Fig. 16. MERIS vs. FerryMon Chl a in APES ( 2007-2008).
Fig. 17. Absolute errors for MERIS vs. FerryMon Chl a estimations.
24
     Fig. 18. Relative errors for MERIS vs. FerryMon Chl a estimations.
25
     Fig. 19. Observed annual dynamics of Chl a in APES (2007-2008 ).
26
     Fig. 20. Estimated annual dynamics of Chl a in APES (2007-2008 ).
               <5    <10   <15   <20   <25   <30    <35    <40      <45




               <50   <60   <70   <80   <90   <100   <200    >200    Back
                                                                   ground




Fig. 21. Examples of Chl a imagery: bloom (left) & non-bloom (right) periods.
               Summary and future work:
     • Collected in situ optical and water quality components (WQC) data in the
     Neuse River and Pamlico Sound region (Chl a, CDOM, FSS, VSS, and TSS)
     • Developed in situ and remote-sensing WQC algorithms based on applying
     radiative transfer and observation data
     • Further improvement of methods developed by using more rigorous in-water
     and atmospheric algorithms and codes
     • Develop and implement a cloud mask algorithm to increase the temporal
     frequency of MERIS observation data.




28
     Thank you!!!



29

								
To top