Search

Many studies of Earth surface processes and landscape evolution rely on having accurate and extensive data sets of surficial geologic units and landforms. Automated extraction of geomorphic features using deep learning provides an objective way to consistently map landforms over large spatial extents. However, there is no consensus on the optimal input feature space for such analyses. We explore the impact of input feature space for extracting geomorphic features from land surface parameters (LSPs) derived from digital terrain models (DTMs) using convolutional neural network (CNN)‐based semantic segmentation deep learning. We compare four input feature space configurations: (a) a three‐layer composite consisting of a topographic position index (TPI) calculated using a 50 m radius circular window, square root of topographic slope, and TPI calculated using an annulus with a 2 m inner radius and 10 m outer radius, (b) a single illuminating position hillshade, (c) a multidirectional hillshade, and (d) a slopeshade. We test each feature space input using three deep learning algorithms and four use cases: two with natural features and two with anthropogenic features. The three‐layer composite generally provided lower overall losses for the training samples, a higher F1‐score for the withheld validation data, and better performance for generalizing to withheld testing data from a new geographic extent. Results suggest that CNN‐based deep learning for mapping geomorphic features or landforms from LSPs is sensitive to input feature space. Given the large number of LSPs that can be derived from DTM data and the variety of geomorphic mapping tasks that can be undertaken using CNN‐based methods, we argue that additional research focused on feature space considerations is needed and suggest future research directions. We also suggest that the three‐layer composite implemented here can offer better performance in comparison to using hillshades or other common terrain visualization surfaces and is, thus, worth considering for different mapping and feature extraction tasks. Plain Language Summary: Characteristics of the land surface, such as steepness, relative slope position (e.g., ridge vs. valley), and roughness, can be digitally represented using a variety of methods. These digital representations, along with human‐annotated labels, can be provided to artificial intelligence algorithms to generate maps of landforms, such as those associated with river systems, glaciers, or human‐induced changes to the landscape. Given the large number of terrain derivatives that can be generated from digital elevation data, it is unclear how the chosen inputs impact the utility of the resulting maps. This study suggests that artificial intelligence mapping algorithms are sensitive to representations provided to them since different inputs resulted in varying map accuracies. We suggest a combination of features, which collectively describe relative slope position, steepness, and local terrain texture, as a means to represent the landscape and to serve as input to artificial intelligence algorithms. Our results can make it easier to train artificial intelligence algorithms to consistently and objectively find surficial features of interest across large swaths of Earth's surface. Key Points: We test the sensitivity of convolutional neural network geomorphic feature extraction approaches to the input feature spaceA three‐layer composite incorporating slope and topographic position index outperformed common inputs like hillshades and slopeshadesResults held true across four use cases including both natural and anthropogenic landforms [ABSTRACT FROM AUTHOR]