A sky camera network (SKYCAM) for the validation of the remote sensing of clouds from Sentinel 2/3

In this work, we describe a newly established network of fisheyes sky cameras (SKYCAM) at a dozen of locations worldwide for continuous cloud monitoring. Each location is equipped with two cameras at about a 100m distance from each other. This dual view enables the retrieval of the altitude of the cloud base.  The cameras acquire a picture (1000 x 2000) of the sky every minute in three different wavelength (Red, Green and Blue) and the data are directly sent to a central facility for processing.

The data are calibrated both for precise geometry (using a variety of techniques including systematic observation of the sun) and radiometry (using Radiative transfer and aerosol information). Using the cloud base information derived from stereo, calibrated radiances and radiative transfer, additional properties of the cloud can be derived (cloud thickness and top height) that can be used to re-construct observations from satellite data. We apply this technique to validate cloud observations from Sentinel 2 /3. This method enables an objective analysis of the remotely sensed cloud mask performances and possible improvements by providing  a large range of surface conditions (vegetation, snow, bright surfaces, urban area) and seasons as the system operates continuously.

At some locations, this system is complemented by surface reflectance measurements over a 100m x 200m area performed from a multispectral camera (CAMSIS) mounted a high tower and/or measurements from AERONET which enable the development/validation of more advanced products (aerosol spatialization, incoming shortwave and photosynthetically active radiation, satellite derived surface reflectance).