Sample SRF for common sensors (incl AVRIS and LANDSAT 7): https://oceancolor.gsfc.nasa.gov/resources/docs/rsr_tables/
https://sci-hub.st/http://dx.doi.org/10.1109/TGRS.2012.2198828
SWIR differences limited to ~2.9% → ~1.6%; more effective for NDVI (7.1 →1.8%).
They used radiative transfer models called “PROSPECT” and “SAIL” (supposedly widely used) literature review they used →
34] S. Jacquemoud, W. Verhoef, F. Baret, C. Bacour, P. J. Zarco-Tejada, G. P. Asner, C. François, and S. L. Ustin, “PROSPECT plus SAIL models: A review of use for vegetation characterization,” Remote Sens. Environ., vol. 113, pp. S56–S66, Sep. 2009.
top-of-atmosphere (TOA) reflectance
where T_g is total gaseous transmittance, and the other values involve zenith angles, rayleigh/aerosol reflactanec, albedo, and some other factor (not super clear).
To couple/combine models, they literally passed the output from one model into another model (not much literature, but:[35] W. Verhoef and H. Bach, “Coupled soil-leaf-canopy and atmosphere ra- diative transfer modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance data,” Remote Sens. Environ., vol. 109, no. 2, pp. 166–182, Jul. 30, 2007.)
some studies show that SRF cross sensor correction is land cover dependent; they mentioned that it’s impractical due to seasonal changes in vegetation, but we might be able to correct for that.
“best SRF cross sensor fit for SWIR is simple linear regression”
^disparity between measurements
need to consider (not considered in paper): combined effects of radiometric calibration accu- racy, sensor degradation, detector-specific SRF, quality assurance, differences in spatial resolution with view angle, atmospheric uncertainty and variability, topography, and sampling directions on SRF of spectral band reflectances and NDVItarget +/- 3%
in SWIR reflectance, common satilites (AVHRR, landsat, etc) were pretty accurate.
