Search for: All records

Abstract Glacial isostatic adjustment produces crustal deformation capable of altering the slope of the landscape and diverting surface water drainage, thereby modulating the hydraulic conditions that govern river evolution. These effects can be especially important near the margins of ice sheets. In Maine, USA, post-glacial changes in sedimentation within major river systems have been interpreted as the result of regional tilting and drainage rerouting due to glacial isostatic adjustment. In this study, we model isostatic adjustment driven by retreat of the Laurentide Ice Sheet, quantify the associated tilting and drainage rerouting, and explore how these changes impacted sediment transport in Maine's rivers. Through an analysis of changes to river slope and drainage area produced by glacial isostatic adjustment, we show that ice sheet retreat altered the median sediment grain size that rivers could entrain. We also find support for previous estimates of the timing and direction of drainage reversal at Moosehead Lake, Maine's largest lake. Our results suggest that the history of sedimentation in Maine's rivers reflects time-dependent effects of glacial isostatic adjustment that are superimposed on any changes in runoff associated with deglaciation. Further, our case study demonstrates that isostatic adjustment affects alluvial channel evolution and sediment delivery to the coastline for several millennia after an ice sheet retreats. 

Abstract Classifying images using supervised machine learning (ML) relies on labeled training data—classes or text descriptions, for example, associated with each image. Data‐driven models are only as good as the data used for training, and this points to the importance of high‐quality labeled data for developing a ML model that has predictive skill. Labeling data is typically a time‐consuming, manual process. Here, we investigate the process of labeling data, with a specific focus on coastal aerial imagery captured in the wake of hurricanes that affected the Atlantic and Gulf Coasts of the United States. The imagery data set is a rich observational record of storm impacts and coastal change, but the imagery requires labeling to render that information accessible. We created an online interface that served labelers a stream of images and a fixed set of questions. A total of 1,600 images were labeled by at least two or as many as seven coastal scientists. We used the resulting data set to investigate interrater agreement: the extent to which labelers labeled each image similarly. Interrater agreement scores, assessed with percent agreement and Krippendorff's alpha, are higher when the questions posed to labelers are relatively simple, when the labelers are provided with a user manual, and when images are smaller. Experiments in interrater agreement point toward the benefit of multiple labelers for understanding the uncertainty in labeling data for machine learning research.