Development and Validation of a Brdf Model for Ice Mapping for the Future Goes-r Advanced Baseline Imager (abi) Using Artificial Neural Network

Information on ice cover extent over seas is crucial for ship navigation. Ice cover can also show interannual fluctuations and reflects climate variations. Ability of satellites to provide global observations at high temporal frequency has made them the primary tool for the ice cover monitoring. This study is a part of GOES-R Cryosphere application group effort to develop new, and improve existing, applications for the future GOES-R Advanced Baseline Imager (ABI). In this paper, a new approach was developed to minimize the effect of both observation and illumination angles on the ice mapping accuracy. A Bidirectional Reflectance Distribution Function (BRDF) was developed to simulate the reflectance of ice and water over the Caspian Sea. The ultimate objective of this research is to develop a daily ice concentration map. The estimation of the reflectance of water and ice is a step toward the achievement of this goal. The Northern region of the Caspian Sea has been selected for algorithm development and calibration. Artificial Neural Networks (ANN) have been used to simulate reflectance values for both water and ice from solar, azimuth and satellite angles. Data collected by SEVIRI instrument onboard of Meteosat Second Generation (MSG) satellite have been used as a prototype. The approach used in the algorithm development includes daily cloud-clear image compositing. The simulated reflectances were compared to observed values and have shown a satisfactory agreement. This implies that the BRDF model coupled with ANN technique can be used to simulate reflectance values.