Mine waste, or tailings, often contain heavy metals and contaminants that can harm plants, wildlife, and human health. Historic and abandoned mine tailings may be left exposed at surface, increasing risk and allowing transport of materials by wind or streams. Cleanup of these sites is costly and time-consuming. We propose a classification method to identify impacted regions, helping prioritize further on-the-ground research. A remote sensing model was developed using Google Earth Engine to classify mine tailings at historic gold mines in Nova Scotia. Over 300 mines in 64 districts were active from the 1860s to the 1940s, many containing arsenic and mercury above safe limits according to soil guidelines. These tailings mostly consist of sulphide minerals, which form secondary minerals such as iron oxides and arsenates when exposed to the air. These secondary minerals have reflectance spectra that can be detected in the visible to near-infrared EM range. The model used multispectral Sentinel-2 data and was trained with the Nova Scotia Mine Tailings Database (NSMTD). The classifier was applied to identify tailings or non-tailings areas using two methods: pixel-wise and object-based. The object-based method, which segments regions into "superpixels," yielded better results, with fewer false positives. This study demonstrates that multispectral data can accurately distinguish tailings from non-tailings, and object-based classification reduces false positives. These methods could assist in identifying and mapping tailings more efficiently, reducing the need for manual delineation.