The forum is locked.
The Ocean Color Forum has transitioned over to the Earthdata Forum (https://forum.earthdata.nasa.gov/). The information existing below will be retained for historical reference. Please sign into the Earthdata Forum for active user support.
Topic Frequently Asked Questions / Data Products & Algorithms FAQ / Why do I see negative radiances? (locked)
Why are do I see negative radiances?
Given the uncertainties in the atmospheric correction process,
instrument digitization, and geophysical values close to zero, some
negative radiances can be expected (e.g. 670nm retrievals in open
ocean frequently dip below zero). However, this does not explain all
negative nLw retrievals.
Most negative nLws, especially in the blue (i.e. 412nm), are the result of
errors in the atmospheric correction process. There are two primary reasons
nLw retrievals can go negative - particularly in coastal regions.
1) Failure of the black pixel assumption
The atmospheric correction algorithm assumes zero (or at least
negligible) radiance in the NIR bands (e.g. 765 and 865nm). This assumption
allows for the selection of an aerosol type. If there is a significant
water-leaving component in the NIR, the black pixel assumption is violated, and
an incorrect aerosol type will likely be selected. When this occurs, the tendency is
to select an aerosol model with a stronger spectral dependence than appropriate,
resulting in more of the signal in the blue to be erroneously attributed to aerosol
radiance. When this large aerosol radiance component is subtracted, the result can
be negative nLw.
The current atmospheric correction algorithm employs a 'NIR-correction'
method, which estimates water-leaving radiance in the NIR from reflectance in
the red (e.g. 670nm). This correction improves retrievals over turbid waters,
but is not perfect, so negative nLws can still result.
2) Absorbing aerosols are present
The atmospheric correction algorithm relies on aerosol models to
estimate the aerosol radiance contribution to the top-of-atmosphere (TOA)
signal. It also uses two fairly close NIR bands to determine which model is
appropriate. All of the existing models are non- or weakly-absorbing. If an
absorbing aerosol is present, the aerosol component will be incorrectly
estimated by our suite of models, resulting in a tendency for 'over-subtraction'
particularly in the blue portion of the spectrum.
Since absorbing aerosols are not easily distinguished from non-absorbing
aerosols in the NIR, we cannot simply include absorbing aerosols in the model
suite. Work is underway to address this limitation of the atmospheric correction.
Given the uncertainties in the atmospheric correction process,
instrument digitization, and geophysical values close to zero, some
negative radiances can be expected (e.g. 670nm retrievals in open
ocean frequently dip below zero). However, this does not explain all
negative nLw retrievals.
Most negative nLws, especially in the blue (i.e. 412nm), are the result of
errors in the atmospheric correction process. There are two primary reasons
nLw retrievals can go negative - particularly in coastal regions.
1) Failure of the black pixel assumption
The atmospheric correction algorithm assumes zero (or at least
negligible) radiance in the NIR bands (e.g. 765 and 865nm). This assumption
allows for the selection of an aerosol type. If there is a significant
water-leaving component in the NIR, the black pixel assumption is violated, and
an incorrect aerosol type will likely be selected. When this occurs, the tendency is
to select an aerosol model with a stronger spectral dependence than appropriate,
resulting in more of the signal in the blue to be erroneously attributed to aerosol
radiance. When this large aerosol radiance component is subtracted, the result can
be negative nLw.
The current atmospheric correction algorithm employs a 'NIR-correction'
method, which estimates water-leaving radiance in the NIR from reflectance in
the red (e.g. 670nm). This correction improves retrievals over turbid waters,
but is not perfect, so negative nLws can still result.
2) Absorbing aerosols are present
The atmospheric correction algorithm relies on aerosol models to
estimate the aerosol radiance contribution to the top-of-atmosphere (TOA)
signal. It also uses two fairly close NIR bands to determine which model is
appropriate. All of the existing models are non- or weakly-absorbing. If an
absorbing aerosol is present, the aerosol component will be incorrectly
estimated by our suite of models, resulting in a tendency for 'over-subtraction'
particularly in the blue portion of the spectrum.
Since absorbing aerosols are not easily distinguished from non-absorbing
aerosols in the NIR, we cannot simply include absorbing aerosols in the model
suite. Work is underway to address this limitation of the atmospheric correction.
Topic Frequently Asked Questions / Data Products & Algorithms FAQ / Why do I see negative radiances? (locked)