More Like this

1. Exploring the Influence of Input Feature Space on CNN‐Based Geomorphic Feature Extraction From Digital Terrain Data

   https://doi.org/10.1029/2023EA002845  

   Maxwell, Aaron E.; Odom, William E.; Shobe, Charles M.; Doctor, Daniel H.; Bester, Michelle S.; Ore, Tobi (May 2023, Earth and Space Science)

   Abstract 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. 

   more » « less
2. Reconstructing Historical Forest Cover and Land Use Dynamics in the Northeastern United States Using Geospatial Analysis and Airborne LiDAR

   https://doi.org/https://doi.org/10.1080/24694452.2020.1856640  

   Johnson, K.; Ouimet, W. (March 2021, Annals of the American Association of Geographers)

   null (Ed.)

   The northeastern United States experienced extensive deforestation during the seventeenth through twentieth centuries primarily for European agriculture, which peaked in the mid-nineteenth century, and followed by widespread farmstead abandonment and reforestation. Analysis of airborne light detection and ranging (LiDAR) data has revealed thousands of historical land-use features with topographic signatures across the landscape under the region’s now-dense forest canopy. This study investigates two different types of features—stone walls and relict charcoal hearths—both of which are associated with widespread deforestation in the region. Our results demonstrate that LiDAR is an effective tool in reconstructing and quantifying the distribution and magnitude of historical forest cover using these relict land use features as a reliable proxy. Furthermore, these methods allow for direct quantification of cumulative land clearing over time in each town, in addition to the extent, intensity, and spatial distribution of cleared land and forest cover. 

   more » « less
3. Extensibility of U-Net Neural Network Model for Hydrographic Feature Extraction and Implications for Hydrologic Modeling

   https://doi.org/10.3390/rs13122368  

   Stanislawski, Lawrence V.; Shavers, Ethan J.; Wang, Shaowen; Jiang, Zhe; Usery, E. Lynn; Moak, Evan; Duffy, Alexander; Schott, Joel (June 2021, Remote Sensing)

   null (Ed.)

   Accurate maps of regional surface water features are integral for advancing ecologic, atmospheric and land development studies. The only comprehensive surface water feature map of Alaska is the National Hydrography Dataset (NHD). NHD features are often digitized representations of historic topographic map blue lines and may be outdated. Here we test deep learning methods to automatically extract surface water features from airborne interferometric synthetic aperture radar (IfSAR) data to update and validate Alaska hydrographic databases. U-net artificial neural networks (ANN) and high-performance computing (HPC) are used for supervised hydrographic feature extraction within a study area comprised of 50 contiguous watersheds in Alaska. Surface water features derived from elevation through automated flow-routing and manual editing are used as training data. Model extensibility is tested with a series of 16 U-net models trained with increasing percentages of the study area, from about 3 to 35 percent. Hydrography is predicted by each of the models for all watersheds not used in training. Input raster layers are derived from digital terrain models, digital surface models, and intensity images from the IfSAR data. Results indicate about 15 percent of the study area is required to optimally train the ANN to extract hydrography when F1-scores for tested watersheds average between 66 and 68. Little benefit is gained by training beyond 15 percent of the study area. Fully connected hydrographic networks are generated for the U-net predictions using a novel approach that constrains a D-8 flow-routing approach to follow U-net predictions. This work demonstrates the ability of deep learning to derive surface water feature maps from complex terrain over a broad area. 

   more » « less
4. Assessment of terrain elevation estimates from ICESat-2 and GEDI spaceborne LiDAR missions across different land cover and forest types

   https://doi.org/10.1016/j.srs.2022.100067  

   Urbazaev, Mikhail; Hess, Laura L; Hancock, Steven; Sato, Luciane Yumie; Ometto, Jean Pierre; Thiel, Christian; Dubois, Clémence; Heckel, Kai; Urban, Marcel; Adam, Markus; et al (December 2022, Science of Remote Sensing)

   Accurate measurements of terrain elevation are crucial for many ecological applications. In this study, we sought to assess new global three-dimensional Earth observation data acquired by the spaceborne Light Detection and Ranging (LiDAR) missions Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) and Global Ecosystem Dynamics Investigation (GEDI). For this, we examined the “ATLAS/ICESat-2 L3A Land and Vegetation Height”, version 5 (20 × 14 m and 100 × 14 m segments) and the “GEDI Level 2A Footprint Elevation and Height Metrics”, version 2 (25 m circle). We conducted our analysis across four land cover classes (bare soil, herbaceous, forest, savanna), and six forest types (temperate broad-leaved, temperate needle-leaved, temperate mixed, tropical upland, tropical floodplain, and tropical secondary forest). For assessment of terrain elevation estimates from spaceborne LiDAR data we used high resolution airborne data. Our results indicate that both LiDAR missions provide accurate terrain elevation estimates across different land cover classes and forest types with mean error less than 1 m, except in tropical forests. However, using a GEDI algorithm with a lower signal end threshold (e.g., algorithm 5) can improve the accuracy of terrain elevation estimates for tropical upland forests. Specific environmental parameters (terrain slope, canopy height and canopy cover) and sensor parameters (GEDI degrade flags, terrain estimation algorithm; ICESat-2 number of terrain photons, terrain uncertainty) can be applied to improve the accuracy of ICESat-2 and GEDI-based terrain estimates. Although the goodness-of-fit statistics from the two spaceborne LiDARs are not directly comparable since they possess different footprint sizes (100 × 14 m segment or 20 × 14 m segment vs. 25 m circle), we observed similar trends on the impact of terrain slope, canopy cover and canopy height for both sensors. Terrain slope strongly impacts the accuracy of both ICESat-2 and GEDI terrain elevation estimates for both forested and non-forested areas. In the case of GEDI the impact of slope is, however, partly caused by horizontal geolocation error. Moreover, dense canopies (i.e., canopy cover higher than 90%) affect the accuracy of spaceborne LiDAR terrain estimates, while canopy height does not, when considering samples over flat terrains. Our analysis of the accuracy and precision of current versions of spaceborne LiDAR products for different vegetation types and environmental conditions provides insights on parameter selection and estimated uncertainty to inform users of these key global datasets. 

   more » « less
5. Evaluating Mask R‐CNN models to extract terracing across oceanic high islands: A case study from Sāmoa

   https://doi.org/10.1002/arp.1909  

   Quintus, Seth; Davis, Dylan_S; Cochrane, Ethan_E (July 2023, Archaeological Prospection)

   Abstract Lidar datasets have been crucial for documenting the scale and nature of human ecosystem engineering and land use. Automated analysis methods, which have been rising in popularity and efficiency, allow for systematic evaluations of vast landscapes. Here, we use a Mask R‐CNN deep learning model to evaluate terracing—artificially flattened areas surrounded by steeper slopes—on islands in American Sāmoa. Mask R‐CNN is notable for its ability to simultaneously perform detection and segmentation tasks related to object recognition, thereby providing robust datasets of both geographic locations of terracing features and their spatial morphometry. Using training datasets from across American Sāmoa, we train this model to recognize terracing features and then apply it to the island of Tutuila to undertake an island‐wide survey for terrace locations, distributions and morphologies. We demonstrate that this model is effective (F1 = 0.718), but limitations are also documented that relate to the quality of the lidar data and the size of terracing features. Our data show that the islands of American Sāmoa display shared patterns of terracing, but the nature of these patterns are distinct on Tutuila compared with the Manu'a island group. These patterns speak to the different interior configurations of the islands. This study demonstrates how deep learning provides a better understanding of landscape construction and behavioural patterning on Tutuila and has the capacity to expand our understanding of these processes on other islands beyond our case study. 

   more » « less