Simple Method to Extract Lake Ice Condition From Landsat Images

Ice plays key roles in regulating hydrological, ecological, biogeochemical, and socioeconomic functions of lakes. Long-term in situ lake ice phenological records indicate that is trending toward later freeze-up, earlier breakup, a shorter duration. Parallel to study using process-based models, satellite remote sensing can expand our understanding change over large spatial scales. However, most studies have focused on lakes or short periods time, which may not robustly represent changes multidecadal time the much more numerous small Here, we present random forest model, Sensitive Lake Detection (SLIDE), accurately extract conditions from Landsat TM, ETM+, OLI images. We trained model manually labeled dataset (1089 areas 995 globally). Our results show achieves accurate classification between ice/snow water (accuracy: 97.8%, kappa coefficient: 95.5%). Comparing Landsat-derived cover with conditions, produces less bias, lower RMSE, higher than does snow/ice flag quality assessment band. This especially true during transitional period surrounding off dates reported (mean bias ?7.3% ?17.3% band). demonstrate feasibility mining rich archive dynamics better flagging ice-affected observations.