Anomaly detection in Remote Sensing imagery is a challenging problem due to the scarcity of anomalous samples, strong class imbalance, and the heterogeneity of sensors and acquisition conditions. Although recent geospatial foundation models provide powerful feature representations, their effective use for anomaly detection remains underexplored. This thesis investigates an embedding-based anomaly detection pipeline for multispectral and hyperspectral satellite imagery, combining foundation model backbones with PatchCore, a patch-level anomaly detection method originally developed for industrial inspection. Multiple foundation models and datasets are evaluated, highlighting the limitations of cross-sensor transfer and motivating a focused study on the FLOGA wildfire dataset, which mitigates sensor mismatch.