The U.S. Navy seeks a partner to license and commercialize a method to detect cloud shadows using only visible channels in remotely sensed imagery
The US Naval Research Laboratory detachment at Stennis Space Center has developed a method to detect cloud shadows in images received from airborne and space-borne ocean color sensors. Cloud shadows can produce errors of 30-40% in the observed reflectance of individual pixels in an image, resulting in corruption of biophysical parameters derived from the pixel values, such as phytoplankton biomass. In the method, a Cloud Shadow Detection Index (CSDI) is generated by comparing radiance from pixels illuminated by direct and diffuse sunlight to cloud shadowed pixels lit only by diffuse light. The CSDI is generated entirely from visible light channels, alleviating the need for thermal and shortwave infrared bands, and does not require inputs of viewing angle nor estimations of cloud vertical heights. The Navy’s technology is less computationally intense than geometry-based cloud shadow approaches, allowing rapid image processing. An innovative Adaptive Sliding Box feature teases out the subtle differences between cloud shadowed and sunlit pixels in a manner that uses raw top-of-atmosphere radiance digital counts. As a result, there is no need for data calibration or atmospheric correction.
Cloud shadow detection can improve ocean color imagery products such as chlorophyll and bathymetry estimates, enable the automation of sensor calibration and atmospheric correction without the need for resource intensive ground-truthing, and improve local weather forecasting by fully accounting for cloud shadow effects. The technology can also be used to estimate cloud base and cloud top heights.
Defense Technology Information Center (DTIC) Report: Optical Algorithm for Cloud Shadow Detection Over Water
- Simple: The only input required is raw top-of-atmosphere radiance from visible bands in the 400-720 nm range, eliminating the need for angular information, cloud vertical height estimation, or ground-truthing
- Fast: Less computationally intense than geometry based systems
- Customizable: Users can adjust the size of the Adaptive Sliding Box in order to most accurately capture cloud shadow based on the characteristics of a given scene
- US Patent 8,509,476 is available for license
- Potential for collaboration with NRL Stennis Space Center researchers