Mangroves are coastal evergreen, salt-tolerant forest ecosystems that play a vital role in biodiversity conservation. Effective mapping and monitoring of mangrove ecosystems are essential for informed management and scientific decision-making. However, the dense and often inaccessible nature of mangrove forests makes traditional field surveys and ground truth collection challenging. Hyperspectral remote sensing offers a valuable solution by enabling species-level discrimination in mangroves through the extraction of vegetation structure and composition from imagery using diverse classification approaches. This study focuses on the discrimination of mangrove species in the mangrove forest areas of Marine National Park (MNP), Jamnagar, Gulf of Kutch, Lothian Island, Sundarbans, West Bengal. The proposed approach utilizes the Modified Spectral Angle Mapper (Modified SAM) algorithm with AVIRIS-NG hyperspectral data. This algorithm addresses the limitations of the traditional Spectral Angle Mapper (SAM) algorithm, which fails to account for the spatial structure and information of mangrove species. Additionally, the proposed algorithm reduces the reliance on a large number of training samples and complex classifiers, which are insufficient for extracting mangrove characteristics due to computational inefficiencies and limited generalization capabilities. By incorporating both spatial and spectral information, the Modified SAM algorithm enhances performance and classification accuracy, even in mixed mangrove pixel regions. This is achieved through the utilization of various distance and similarity measures, including angle change resulting from shape alterations in mangrove spectra, angle change due to the magnitude of mangrove spectra changes, and distance change between two mangrove spectra. These measures facilitate the identification of subtle differences between mangrove spectra. The results demonstrate a significant improvement in classification accuracy compared to the traditional SAM algorithm, particularly in mixed mangrove pixel regions. The achieved overall accuracy (OA) is 93.48% for the Lothian Islands, Sundarbans, West Bengal, and 94.32% for the Marine National Park (MNP), Jamnagar, Gulf of Kutch. The proposed Modified SAM algorithm, in conjunction with AVIRIS-NG hyperspectral data, offers an effective approach for reliable and efficient classification and discrimination of mangrove species. The algorithm's enhanced performance and classification accuracy have significant implications for mapping, monitoring, and managing mangrove ecosystems, supporting conservation efforts, and facilitating informed decision-making for sustainable mangrove ecosystems. [ABSTRACT FROM AUTHOR]