MOBY
MOBY (Marine Optical BuoY) radiometry data are regularly used by the NASA OBPG as part of ocean color validation and vicarious calibration activities. MOBY is an autonomous buoy moored off of the island of Lanai in Hawaii. Each day it is deployed it collects several measurements of upwelling radiance from sensors on its underwater arms (at approximately 1, 5 and 9 m depth) and downwelling irradiance from sensors on its underwater arms as well as at the surface. More information about MOBY can be found on the MLML site and NOAA site.
Table of Contents
MOBY Validation Overview
NASA Ocean Ecology Laboratory (OEL) staff calculate Rrs values from MOBY measurements. The validation match-up results for various satellite ocean color sensors can be viewed through the SeaBASS Validation Search page. To view them, visit the Validation Search, select the "Rrs" product, then select the "MOBY only" radial button from the Data Sources listed at the bottom of the page. When you click the search button, the validation results displayed will be limited to data from MOBY. Note that data from MOBY are used in the vicarious calibration of ocean color satellites (Franz et al. 2007), therefore validation results using MOBY data are kept separate from other SeaBASS validation data by default.
How does the OBPG process MOBY data?
For consistency, the OEL performs analysis to generate water leaving values from MOBY measurements rather than using the calculated values provided in the MOBY Gold Directory or listing of Spectral Data. The results are similar, although the OEL has its own methods for processing the data and setting quality flags. A description of those methods follows.
For vicarious calibration and most validation, the OEL uses MOBY measurements that have had pre- and post- deployment calibrations applied, as noted in the MOBY reprocessing page. That restriction means there is a time-delay in SeaBASS MOBY match-ups since it takes months before the post-deployment calibrations can be measured and applied. Some exceptions apply when more recently collected data are used to increase the number of available results with recently launched missions, such as was performed for preliminary calibration and validation results shortly after SNPP-VIIRS became operational.
MOBY has 3 underwater arms (top, middle and bottom arms at approximately 1, 5 and 9 m depth respectively). Each arm has an upwelling radiance (Lu) sensor and a downwelling irradiance (Ed) sensor. Additionally, above-water downwelling irradiance (Es) is measured from sensors mounted on top of the buoy. During a sampling cycle*, the different sensors take measurements one at a time, with several Es measurements interspersed between the various underwater Lu and Ed measurements. Radiometric measurements from the different MOBY sensors include measurements from 2 spectrographs with some overlapping wavelengths. The OEL merges the spectra, preferentially using the higher wavelengths from the "blue" (or lower) spectra since those are of higher quality than the lower wavelengths of the red. Any slight offset between the two spectra is ignored. A cubic spline interpolation is used to resolve values at each 1 nm step between the full MOBY measurement range of approximately 340 nm to 960 nm.
*While true for historic data, the MOBY system has recently been updated and redesigned to collect multiple measurements simultaneously. The content described on this page focuses on describing the processing methods designed around the historic measurements.
Log-linear fits of the measurements versus depth are calculated and used to propagate Lu measurements to water-leaving radiances (Lw). For validation, paired Lw and Es measurements are used to calculate remote sensing reflectances (Rrs) with f/Q corrections applied. Only data that passed our quality criteria (described below) are used. For vicarious calibration, only Lw values are used, and even more stringent quality flagging is applied. Data are convolved to sensor-specific band patterns for vicarious calibration. Data used for SeaBASS validation are not convolved.
Lw (water leaving radiances)
Lw is calculated by propagating Lu measurements to just below and then across the surface.
First, Lu(0-) is calculated using a linear fit of log(Lu) versus depth at each wavelength using the top, middle and bottom arm measurements. During that calculation a small normalization correction is applied to Lu (and Ed) measurements to adjust for any changes in the light field during sampling (see the Es discription below for more details.) For quality control, two Lw values are computed and compared, one using measurements from all 3 arms and one using measurements from just the top and middle arms. Only the 3-arm values are used in output.
Lw(λ) are calculated by propagating Lu(λ,0-) across the sea-air threshold taking into account wavelength, temperature, and salinity (Voss and Flora, 2017; Huibers, 1997; Quan and Fry, 1995) rather than assuming a fixed constant value of ~0.543. Temperature measurements from the MOBY platform are used and a sea surface salinity of 34.85 PSU is assumed (Voss and Flora, 2017).
Rrs values used in validation are computed using calculated Lw and measured Es.
Rrs(λ) = ( Lw(λ) / Es(λ) ) * f/Q correction
Es (surface irradiance measurements)
MOBY takes several Es measurements throughout its sampling cycle. The Es measurement closest-in-time to the middle-arm's Lu measurement-time is used in the Rrs calculation. Additional Es measurements are used for quality control.
Ratios of Es measurements are used to generate a small scaling correction that is applied to normalize the Lu and Ed measurements. This calculation is performed because the underwater sensors do not measure simultaneously, and the entire sampling cycle can take a significant amount of time (e.g. 20 minutes), enough time for lighting conditions to fluctuate. Note that this correction is typically quite small and that if Es values change drastically (e.g. due to clouds or other significant changes) then the "Es_stability" quality flag described later will screen out the measurements.
Es measurements are matched with the closest-in-time Lu measurement from the 3 arms. For example, the following term "Es(bottom)" means the values of the surface irradiance measurement measured closest-in-time to when the bottom arm measured Lu.
MidScalingCoefficient = Es(top) / Es(mid)
BotScalingCoefficient = Es(top) / Es(bottom)
These coefficients are then multiplied against the middle and bottom log(Lu) and log(Ed) values to normalize them to the light field that existed when the top arm measured.
Es measurements are matched with the closest-in-time Lu measurement from the 3 arms. For example, the following term "Es(bottom)" means the values of the surface irradiance measurement measured closest-in-time to when the bottom arm measured Lu.
MidScalingCoefficient = Es(top) / Es(mid)
BotScalingCoefficient = Es(top) / Es(bottom)
These coefficients are then multiplied against the middle and bottom log(Lu) and log(Ed) values to normalize them to the light field that existed when the top arm measured.
Ed (i.e. at-depth downwelling irradiance) measurements are only used for QC; those values are not part of our calculation for Lw or Rrs. We calculate Kd using the Ed measurements.
Quality Criteria
Various parameters are calculated for quality control. If any of the following tests fail the listed limits then all output is flagged and not used for either validation or vicarious calibration. Measurements from unusual circumstances are also flagged, such as if damage is detected to one or more sensors/arms, or in rare cases if MOBY's tether is broken and it drifts off site.
Lw_RMS <= 5
The RMS (root mean square) of the percent error between Lw computed using all 3 arms and Lw computed using just the top 2 arms
Es_RMS <= 10
The RMS of the percent error between Es (surface irradiance measurements) and underwater Ed measurements extrapolated to Es
ES_stability <= 10
The percent error between the min and max measured Es
TheoryES_diff <= 15
The RMS difference between modeled clear sky Es and the measured Es between 425-575 nm.
Sensor Tilt and Roll <= 5 degrees
Last edited by Chris Proctor on 2020-09-22
Created by Chris Proctor on 2013-09-25