To date there exists no comprehensive salt marsh mapping in Canada, and global models are found to exhibit poor regional accuracy. The lack of consistent salt marsh extent mapping is needed to understand the ecological changes in habitat that occur over time, as well as for carbon accounting purposes, such as that required by the IPCC. Remote sensing imagery of dynamic coastal environments is challenging as the changes in tidal stage inhibit image mosaicking and compositing methods that are often used in large-area mapping. Further, low tide imagery is desirable because high tides result in flooded salt marsh vegetation and mudflats, making them more difficult to discriminate. The purpose of this research was to determine the optimal model inputs for classifying salt marsh vegetation extent in the Bay of Fundy, testing combinations of optical and radar imagery from different seasons and phenological stages, as well as different tidal conditions. From the optimal model (84%+) classification maps for both 2020 and 2023 were produced, concluding that a combination of low and high tide imagery from summer, spring, and fall were best for classifying high and low marsh vegetation. Low tide imagery was most important, and similar accuracy was found using low-tide only data. The inclusion of data from multiple stages of plant growth/seasons is perhaps most important across all models. Post-classification comparison was used to derive a change map of salt marsh vegetation, for which accuracy and uncertainty measures were also computed. This work establishes the beginning of what is needed to include salt marsh activity data in IPCC reporting.