This paper presents a study on texture analysis of SAR sea ice imagery using gray-level co-occurrence matrices (GLCM). The goal is to determine the best parameters for representing sea ice texture in SAR imagery. The study evaluates different GLCM implementations and their effectiveness in classifying sea ice textures. The key findings include the importance of displacement values over orientation, the effectiveness of using a range of displacement values to capture both microtextures and macrotextures, and the conclusion that an eight-level quantization is undesirable. The study also shows that the best GLCM implementation is one that uses a range of displacement values to adequately represent sea ice texture. The study uses a supervised Bayesian classifier to evaluate the effectiveness of different GLCM implementations. The results indicate that the MDMO implementation outperforms the ODMO and ODOO implementations in terms of classification accuracy. The study also highlights the importance of using a range of displacement values and the negative impact of cluster-type features on classification accuracy. The study concludes that the MDMO implementation is the best for representing sea ice texture in SAR imagery.This paper presents a study on texture analysis of SAR sea ice imagery using gray-level co-occurrence matrices (GLCM). The goal is to determine the best parameters for representing sea ice texture in SAR imagery. The study evaluates different GLCM implementations and their effectiveness in classifying sea ice textures. The key findings include the importance of displacement values over orientation, the effectiveness of using a range of displacement values to capture both microtextures and macrotextures, and the conclusion that an eight-level quantization is undesirable. The study also shows that the best GLCM implementation is one that uses a range of displacement values to adequately represent sea ice texture. The study uses a supervised Bayesian classifier to evaluate the effectiveness of different GLCM implementations. The results indicate that the MDMO implementation outperforms the ODMO and ODOO implementations in terms of classification accuracy. The study also highlights the importance of using a range of displacement values and the negative impact of cluster-type features on classification accuracy. The study concludes that the MDMO implementation is the best for representing sea ice texture in SAR imagery.