This question was submitted on behalf of the User.
I have a doubt related with collocation methodology (CERES - ES8_Terra-FM1 vs MTSAT1). In order to do the CERES FOV I just apply a circle whose centroid is located by CERES Lat/Lon (Colatitude of CERES FOV at TOA/ Latitude of CERES of FOV at TOA) with radius - (20km/cos(VZAceres))/(4km/cos(VZAmtsat)) - or I have to apply the CERES PSF (CERES Collection Guide) to get the true ellipse?
(20km/cos(VZAceres))/(4km/cos(VZAmtsat)) - because I am comparing with the MTSAT.
Subject expert answer:
One thing to be careful about is the ES8 provides geolocation information at TOA using a geocentric Earth. The SSF provides geolocation information at the surface on a geodetic Earth. Most imagers use geodetic surface geolocation.
The lag in the electronics is accounted for in the centroid provided for the CERES footprints. As the Collection Guide shows, the CERES Point Spread Function (PSF) is quite peaked near the centroid and drops off quickly and is slightly larger (30 km at nadir) than the nominal 20 km optical size we use as a reference. When assigning imager pixels to the footprint, only the center is considered. I’m unaware of any studies here that compared a simple average of imager pixels within the footprint to the value that is obtained using the PSF.
Dividing the size of a CERES footprint at nadir by the view zenith angle will determine the growth in size with view zenith.
A study using geostationary satellite imager radiance and cloud properties on a 1 degree by 1 degree grid then apply the Angular Distribution Model (ADMs) had about 1 percent bias and 8 percent RMS over oceans in a monthly average when compared to gridded CERES fluxes. They concluded the quality of the geostationary clouds has the largest impact. https://ceres.larc.nasa.gov/documents/STM/2016-10/3rdSession_Tue18OctWed19Oct_CEREStechnical/22_Doelling_TISA.pdf p43-46.
The accuracy you need in your results should guide your decision on the approach to take.