The Multi-angle Imaging SpectroRadiometer (MISR) onboard the EOS-Terra satellite uses several unique approaches to global, automated cloud detection, including a stereoscopic approach, a band-differenced angular signature approach, and spectral and spatial approaches that use dynamic thresholds. The combination of these approaches results in good cloud detection over the entire globe, including traditionally difficult surfaces like snow, ice, desert, and sunglint. We present an overview of the MISR cloud detection algorithm, examples of performance, and early validation results.

The MISR instrument carries nine separate cameras with viewing zenith angles at the surface of 0°, ±26.1°, ± 45.6°, ± 70.5°. Each camera is designed to give a cross-track resolution of about 275 m and to provide images in a push-broom fashion in four narrow spectral bands (446, 558, 672, 866 nm), over a swath of 360 km. In its sun-synchronous orbit, this results in global coverage every 9 days.