Search for: All records

Abstract The world’s coastlines are spatially highly variable, coupled-human-natural systems that comprise a nested hierarchy of component landforms, ecosystems, and human interventions, each interacting over a range of space and time scales. Understanding and predicting coastline dynamics necessitates frequent observation from imaging sensors on remote sensing platforms. Machine Learning models that carry out supervised (i.e., human-guided) pixel-based classification, or image segmentation, have transformative applications in spatio-temporal mapping of dynamic environments, including transient coastal landforms, sediments, habitats, waterbodies, and water flows. However, these models require large and well-documented training and testing datasets consisting of labeled imagery. We describe “Coast Train,” a multi-labeler dataset of orthomosaic and satellite images of coastal environments and corresponding labels. These data include imagery that are diverse in space and time, and contain 1.2 billion labeled pixels, representing over 3.6 million hectares. We use a human-in-the-loop tool especially designed for rapid and reproducible Earth surface image segmentation. Our approach permits image labeling by multiple labelers, in turn enabling quantification of pixel-level agreement over individual and collections of images. 

Coastal resiliency is the ability of a beach–dune system to recover to a previous state after a storm, and this resiliency is affected by prestorm beach and dune morphology and storm climate (i.e. storm frequency and intensity). Improvements in remote sensing technology such as LIDAR and structure from motion have enabled rapid collection and production of digital elevation models used to assess storm impact and recovery. Although rapid poststorm assessment requires a consistent approach for extracting dune morphology, relatively little attention has focused on defining the different parts of a dune. The goals of this paper are to examine how the definition of a dune feature drives the methodology used to extract dunes and to synthesize a comprehensive definition of dune features. An analysis of existing approaches for extracting beach and dune morphology demonstrates that there is considerable variation in how the beach–dune transition (i.e. dune toe) is defined. Many definitions are recursive or include ambiguous terminology, resulting in a dune toe or crest line position dependent on user interpretation of the definition. Other definitions rely heavily on user interpretation of dune features at varying stages in the feature extraction process. Reliance on visual interpretation can result in substantially different feature locations across different interpreters. Given the impact of varying definitions on dune resiliency assessments and legal implications for dune features location, we propose a series of semantic models for dune features. Semantic modelling of coastal morphology is vital for consistently and accurately assessing coastal recovery and predicting future coastal assessments on the basis of a consistent set of criteria. 

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.