|
Many inter-consistency efforts force empirical agreement between sensors viewing a source nearly coincident in time
and geometry that ensures consistency between sensors rather than obtain an SI-traceable calibration with documented
error budgets. The method described here provides inter-consistency via absolute radiometric calibration with defensible
error budget avoiding systematic errors through prediction of at-sensor radiance for a site viewed by multiple sensors but
not necessarily viewed at coincident times. The method predicts spectral radiance over a given surface site for arbitrary
view and illumination angles and for any date dominated by clear-sky conditions. The foundation is a model-based, SItraceable
prediction of at-sensor radiance over selected sites based on physical understanding of the surface and
atmosphere. The calibration of the ground site will include spatial, spectral, and sun-view geometric effects based on
satellite and ground-based data. The result is an interconsistency of hyperspectral and multispectral sensors spanning
spatial resolutions from meters to kilometers all relative to the surface site rather than a single sensor. The sourcecentric
philosophy of calibrating the site inherently accounts for footprint size mismatch, spectral band mismatch, and
temporal and spatial sampling effects. The method for characterizing the test site allows its use for SI-traceable
calibration of any sensor that can view the calibrated test site. Interconsistency is obtained through the traceability and
error budget rather than coincident views. Such an approach to inter-consistency provides better understanding of biases
between sensors as well producing more accurate results with documented SI-traceability that reduces the need for
overlapping data sets.
|
