| Abstract: | Radiation from the Earth’s surface undergoes significant spectral alternation due to interaction with the atmosphere, before reaching the measuring device located at high altitude or space. Estimating the effect of the atmosphere on measurements of Earth is crucial for a variety of remote sensing tasks, atmospheric sciences, pollution monitoring and radiation budget estimation. During the recent years, several approaches have been proposed for estimation of the atmospheric optical characteristics. Usually, each approach is tailored to a specific measuring device and scene.
In this work, we present a parametric model for the interaction of radiation with the earth's surface and atmosphere. The estimation of the surface and atmosphere radiative properties is done by finding a set of parameters which minimize the difference between the model and the real data. The data we use is composed of multi-directional and multi-spectral measurements of Earth, taken by a space borne measuring device called MISR. MISR measures the top of the atmosphere radiance at four spectral bands and nine angles.
In the heart of the model, lays a property which we assume many natural surfaces posses. The property is low dimensionality dependence of the spectral and directional parts of the surface BRDF. It is possessed (to some extent) by any surface which does not change its apparent hue when viewed from different directions. This assumption poses strong constraints on the directional distribution of the surface-leaving radiance, at different wavelengths. Exploiting the relative transparency of the atmosphere at near infrared band, allows direct measurement of the surface leaving radiance at this band. Using the low dimensionality dependence property and the near infrared measurements, surface leaving radiance at all other measured bands is known up to a few (unknown) parameters. In addition, radiation exponential attenuation and path radiance are also inserted into the model. By minimizing the difference between the model and real data, the optimal parameters are found. Having these parameters enables the reconstruction of the surface leaving radiance (atmospheric correction) and quantification of some atmospheric properties. |
