The Arctic has been warming three times as fast as the rest of the planet. The accelerated warming has caused a number of changes in rivers, including earlier timing of river ice break-up and later timing of river ice freeze-up. These changes have significant consequences for the local populations by increasing the potential for floods and threatening food security by rendering the traditional subsistence hunting and transportation routes inaccessible. Several studies have attempted to observe the changes in break-up/freeze-up patterns of river ice by leveraging optical remote sensors. However, the research is still limited as the optical sensors are affected by the atmospheric conditions and reduced winter daylight. To address this problem, We employ a supervised Random Forest (RF) classification method to the Sentinel-1 Synthetic Aperture Radar (SAR) backscatter intensity coupled with local gage-based temperature and wind velocity observations to classify river ice and open water pixels over 2020 to 2023 melt seasons. This analysis is applied to a 2 km reach of the Copper River below the confluence of the Tazlina River, where we have in-situ photographs available from a webcam for validation. Open water and river ice estimates are validated using oblique webcam photographs acquired from 2020-present that were processed for georeferencing and orthorectification. Our preliminary RF classification results yielded a testing accuracy of 78% using only the Sentinel-1 backscatter and 96% testing accuracy using the webcam images. Results from our analysis can inform changes in the freeze-up/break-up patterns of river ice on which the local population relies for transportation and subsistence fishing/hunting. Further, the estimates from our analysis can be used to improve and validate modeling capabilities.