Stray Light or Spectral Out-of-Band in MOS/MOBY Steve Brown & Carol Johnson (NIST) Michael Feinholz & Stephanie Flora (MLML) Jim Mueller (CHORS)
MOBY Workshop November MOBY Instruments MOS –MOS 202 MOS Profiler and MOS ROV –MOS 204 Even MOBY deployments –MOS205 Odd MOBY deployments Other instruments –SIS –FOS –MD5
MOBY Workshop November Marine Optical System (MOS) Blue SpectrographRed Spectrograph
MOBY Workshop November Imaging in Spectrographs Broad-band sources Narrow-band Sources
MOBY Workshop November MOS Imaging Specular Haze Diffuse Incident Ray Surface
MOBY Workshop November MOS Profiler—Measurements on SIRCUS Blue Spectrograph G CCD SM Slit Interreflection of second order causes side lobe (MOS designed to operate in first order) Spatial profile across the CCD MOS Spectral Characterization Complicated by presence of ‘reflection peaks’
MOBY Workshop November Manifestation of Stray Light in MOBY Circled Region: L u ( ) derived using the two spectrographs in MOBY or the MOS Profiler disagree in their region of overlap; degree of discrepancy is depth- dependent “Stray light” was suspected (a typical issue with single grating spectrographs used with sources of different spectral shapes) Red Spectrograph Blue Spectrograph
MOBY Workshop November Spectral Out-of-Band SeaWiFS Band 5
MOBY Workshop November MOS Stray light parameters Tuning the laser in small steps gives the single pixel in-band profile In-band Area Haze, Diffuse and Reflection Peak components In-Band Area
MOBY Workshop November Modeled stray light parameters Haze Component y=m 1 *exp(-2*abs(m-m 0 )/m 2 )+m 3
MOBY Workshop November Modeled stray light parameters Reflection peak component Corrected Uncorrected
MOBY Workshop November Modeled stray light parameters Reflection peak component
MOBY Workshop November Modeled stray light parameters BSG Movie
MOBY Workshop November Modeled stray light parameters Rsg movie
MOBY Workshop November MOS Images v Single Pixel Responsivity (Obs) (FSOB)
MOBY Workshop November Stray light characterizations possible Even buoys, each deployment –Es (1), Ed (3), Lu (3), LuMos (1) Odd buoys, each deployment –Es (1), Ed (3), Lu (3), LuMos (1) MOS Profiler MOS ROV Stray light characterizations needed Each MOS –MOS202, MOS204, MOS205 –Fibered and un-fibered configurations
MOBY Workshop November Stray Light Correction Algorithm V.1 Iterative Approach –using first order characterization V.2 Iterative Approach –More detailed characterization V.3 Iterative Approach –Generated Obs Lookup Table –Allows width to refl pk V.4 Matrix-based Approach –Allows width to in-band area –No separation of ‘in-band’ and ‘out-of-band’ regions
MOBY Workshop November Iterative SLC Algorithm Slit scatter function Relative spectral responsivity Separate into IB and OOB components
MOBY Workshop November Iterative SLC Algorithm In-band area of a pixel or column Slit scatter function 1.In-band area all in one pixel; usually +/- 10 pixels around center 2.Assume the ccd spectral responsivity spatially uniform (across the array) 3.Convergence to stable solution in 3 to 4 iterations usually Changes less than 0.1 % V.2 – put reflection peak area into single pixel Use discrete nature of spectrographs
MOBY Workshop November V.2 SLC Application Responsivity MOBY L u Top Better agreement in overlap Increase in UV
MOBY Workshop November V.2 SLC Algorithm
MOBY Workshop November Stray Light Correction to MOBY L w ’s 7.8 % 2.7 % 1.3 % 0.8 % -2.5 %
MOBY Workshop November Stray Light Correction to MOBY L w ’s Different Fiber
MOBY Workshop November Stray Light Correction Algorithm V.2 Validation and Uncertainties Lamp-ill. ISS ISS Stray light corrected Comparison with SIRCUS SIRCUS
MOBY Workshop November Algorithm Validation using Colored Sources Optronic Source with Colored Filters BG28 (blue) PER (green)
MOBY Workshop November Evaluation of the uncertainties Uncertainty in SLC for in-water upwelling radiance L u There are a number of parameters that go into the model; each has an uncertainty We doubled the uncertainty in the fits to those parameters to account for drift, etc. Then ran a Monte Carlo simulation: for each component we used a Gaussian probability distribution. Simulations run a minimum of 100 times & uncertainties calculated.
MOBY Workshop November V.3 Stray Light Correction Algorithm Change the way the SSF (RSR) is derived –Model obs for each pixel, combine into rsr Allow width to reflection peak –Generate lookup tables (faster data reduction)
MOBY Workshop November V.4 Stray light correction algorithm 512 equations and 512 unknowns (or 1024 by 1024) –Suggests straight forward application of matrix-based approaches – MATLAB is very well suite for. –No longer separate in-band and out-of-band; –Simpler, faster application; –May be more applicable for the general community. Change the algorithm Need to keep track of spatial v. spectral in
MOBY Workshop November Remaining Issues Finalize MOS ROV: 1 day L u MOS Parameters: ? Standardize (clean up) parameters used in V. 2 algorithm – consistent application of parameters Finish application of V.3 algorithm –Re-fit reflection peaks: 1 day per MOS Finish development of V.4 Algorithm –Develop Matlab program: ½ day –Compare with V. 2, V. 3 algorithms: ½ day –Evaluate robustness, uncertainties –Off-ccd terms? Documentation New ‘blue’ calibration source to reduce radiometric uncertainties (incl. stray light) in the blue
MOBY Workshop November Blue LED Calibration Source OL420, 425 suffer from low flux below 400 nm –Uncertainties in the repeatability grow rapidly – To 5 % below 380 nm (my memory – Mike?) Stray light correction in L u –Up to 40 % below 380 nm –While the uncertainty in the magnitude of the SL correction increases to 10 % Can we develop a stable source with increased flux in the blue that is a closer match to what MOBY measures – so the magnitude of the stray light correction is significantly reduced?
MOBY Workshop November Gamma Scientific LED Source RS-5 OL420 MOS In the MOBY tent along with OL420
MOBY Workshop November Gamma Scientific LED Source
MOBY Workshop November OL 420 v RS
MOBY Workshop November Use MOBY LEDs? Can we fill in the blue?