Maxwel F. Schwid Mackenzie K. Keith Brandon T. Overstreet 20250311 LAZ binary data data release doi.org/10.5066/P1Q5K657 https://doi.org/10.5066/P1Q5K657 In cooperation with the U.S. Army Corps of Engineers (USACE), the U.S. Geological Survey (USGS) deployed ground control points and coordinated aerial photograph acquisition of Fern Ridge Lake, a multi-purpose reservoir in western Oregon impounded by the 13-m tall Fern Ridge Dam. Aerial photographs were acquired by the Civil Air Patrol (CAP) during winter 2023 when water levels were at or near typical annual “low pool” or minimum pool, a target elevation (108 m NGVD 29) for flood-control operations. Photographs were acquired at two different altitudes with a WaldoAir XCAM Ultra 50 camera mounted on a Cessna aircraft and captured the entire reservoir area as defined by full pool (or maximum conservation pool elevation), including major tributaries entering the reservoir such as the Long Tom River and Coyote Creek, upstream of Fern Ridge Dam. Dam operations at the 3,700-hectare Fern Ridge Lake, located 39 kilometers upstream of the confluence of the Long Tom River and the Willamette River, along with other hydrogeomorphic conditions, result in a diverse array of geomorphic processes and landforms within the reservoir. To document reservoir floor geomorphology, the USGS applied structure-from-motion (SfM) techniques to these aerial photographs, following the workflow outlined in Over and others (2021), and generated three-dimensional xyz point clouds, digital surface models (DSMs), and orthomosaics of Fern Ridge Lake at low pool. This data release includes ground control points, dataset footprints, original aerial photographs, processed point clouds, DSMs, and orthomosaics of Fern Ridge Lake with varying aerial extents and resolutions that were all developed from imagery acquired in January and February of 2023: (1) the January 30 model (FernRideLake_10cm) covered the entire reservoir area with an average point density of 7.47 points per square meter, DSM resolution of 36.6 centimeters per pixel, and orthomosaic ground resolution of 9.15 centimeters per pixel; (2) the January 31 model (FernRidgeLake_5cmLongTomRiver) covered the Long Tom River within the reservoir with an average point density of 25 points per square meter, DSM resolution of 20 centimeters per pixel, and orthomosaic ground resolution of 5 centimeters per pixel; (3) the combined February 2 and February 8 model (FernRidgeLake_5cmCoyoteCreek) covered Coyote Creek within the reservoir with an average point density of 36.3 points per square meter, DSM resolution of 16.6 centimeters per pixel, and orthomosaic ground resolution of 4.15 centimeters per pixel. All DSMs and orthomosaics are formatted as Cloud Optimized GeoTIFFs (COGs) for enhanced web visualization (GDAL 2024). This documentation describes a high-resolution point cloud (LAZ file) of Fern Ridge Lake, Oregon, generated from SfM techniques using aerial photographs acquired on January 30, 2023. The original aerial photographs acquired by CAP, along with the point clouds, DSMs, and orthomosaics covering Fern Ridge Lake were created to provide high-resolution datasets depicting the reservoir floor that are exposed during low pool conditions. These datasets are useful for evaluating geomorphic characteristics of the reservoir floor and assessing other landscape conditions; when compared with future datasets, these datasets will also be useful for tracking changes in reservoir floor conditions. 20230130 ground condition Complete None planned -123.36270 -123.24320 44.12480 44.04150 ISO 19115 Topic Categories biota elevation environment inlandWaters geoscientificInformation imageryBaseMapsEarthCover USGS Thesaurus structure from motion geospatial analysis aerial photography remote sensing geomorphology geospatial datasets digital elevation models image collections GPS measurement None fluvial geomorphology aerial images sfm photogrammetry orthoimagery orthophotograph orthomosaic GeoTIFF Cloud Optimized GeoTIFF dam reservoir drawdown river processes Civil Air Patrol Cessna Waldo USGS Metadata Identifier USGS:65944acbd34e3265ab14f762 Geographic Names Information System (GNIS) Alvadore Amazon Creek City of Veneta Coyote Creek Elmira Eugene Fern Ridge Fern Ridge Lake Fern Ridge Shores Fir Butte Fisher Butte Gibson Island Hannavan Creek Inman Creek Job Swale Creek Lane County Long Tom River Middle Fork Coyote Creek Monroe Orchard Point Oregon Perkins Peninsula Richardson Butte Richardson Point Signal Island Veneta Warren Slough West Fork Coyote Creek Willamette River none The U.S. Geological Survey shall not be held liable for improper or incorrect use of the data retrieved from the system. Public domain data from the U.S. Government are freely redistributable with proper metadata and source attribution. The U.S. Geological Survey should be acknowledged as the data source in products derived from these data. Schwid, Maxwel F. U.S. Geological Survey Hydrologist Mailing and Physical

601 SW 2nd Avenue

Suite 1950

Portland OR 97204 United States 503-758-5589 gs-w-or_sciencebase@usgs.gov Funding for this study was provided by the U.S. Army Corps of Engineers. None Unclassified None Agisoft Metashape Professional Version 1.8.4; Esri ArcGIS Pro Version 3.2.49743; GDAL Version 3.6.2. One R8 and two R10 Trimble real-time kinematic global navigation satellite system (RTK-GNSS) receivers connected to the Oregon Department of Transportation Real-Time GNSS Network were used to collect the ground control points. Overall horizontal and vertical accuracy based on occupation of a USACE benchmark was within three centimeters. All data were acquired and handled in a consistent manner. Four crewed flights were contracted through the USACE to the Civil Air Patrol, Eugene, Oregon on January 30, January 31, February 2, and February 8, 2023. A total of 4,337 aerial photographs were captured at an average flight altitude of either 500-600 or 1,000 meters above ground level, collectively covering about 130 square kilometers. Photographs were acquired as three-band (RGB) images in JPEG format with a WaldoAir XCAM Ultra 50 camera, consisting of two oblique-mounted Canon EOS 5DS R cameras that were triggered simultaneously, mounted to a Cessna 182 aircraft. A Novatel OEMStar GPS recorded camera positions in the image file EXIF data in UTC time zone. All aerial photographs were contemporaneously aligned in Agisoft Metashape software. Agisoft software determined which aerial photographs were used in SfM product (point clouds, DSMs, and orthomosaics) generation based on photograph alignment and the validity of identified tie points, which represent common pixels in photographs as determined by the software. To provide check points for model accuracies, a portion of the ground control points were excluded from SfM product generation and used for model validation. Specifically, 29 checkpoints out of the 76 survey points for full Fern Ridge Lake datasets were excluded along with 3 out of the 14 for the Long Tom River coverage, and 8 out of the 27 for the Coyote Creek coverage. The tie points were visually inspected and filtered to exclude obvious anomalous points from the derivative products (dense point clouds, DSMs, and orthomosaics). Aligned, referenced aerial photographs and their corresponding filtered tie points were separated into the three domains (entire reservoir at 10 cm, Long Tom River at 5 cm, and Coyote Creek at 5 cm) for dense point cloud and derivative product generation. Comparison between overlapping DSMs revealed systematic elevation differences up to 1.5 m in unvegetated areas within the reservoir domain. The three orthomosaics contain compression-derived artifacts occurring as erroneous pixel values that border the outer extent of each rasters’ multiple overviews; these artifacts do not occur in, nor affect the underlying full-resolution images. The dataset is considered complete and consists of ground control points, coverage footprints, raw aerial photographs, point clouds, DSMs, orthomosaics and processing reports for the three separate acquisition efforts. All 78 surveyed ground control points are listed in the CSV text file and all 4,337 aerial photographs are included in each associated zipped flight file. All aerial photographs (excluding four that only captured water of the low reservoir pool and could not be aligned) were used to create SfM products (point clouds, DSMs and orthomosaics). Dense point cloud files were generated for each domain from tie points identified in the Agisoft Metashape software after a standardized filtering process that excludes low-certainty or anomalous points. Filtered dense point clouds were used to create DSMs and orthomosaics. The DSM generated for the entire reservoir domain (FernRidgeLake_10cm) excludes the full extent of the reservoir pool (water level); this area of the model contained false topography and high error propagated from the software’s identification of false tie points below the translucent, shallow water surface of the pool or in areas influenced by lake waves (Over and others, 2021). All DSMs and orthomosaics were clipped during export from Agisoft software to exclude areas of high uncertainty or distortion, particularly in densely vegetated or forested areas or near edges where minimal photograph overlap inhibited photogrammetric processing. No formal horizontal accuracy tests were performed. Sources of potential error that affect the horizontal accuracy include ground control point accuracy and error incurred during alignment, optimization, and ground control processing procedures within the Agisoft Metashape software. Although the aerial photograph locations (recorded by a Novatel OEMStar GPS located on the aircraft) are used by the photogrammetric software during initial alignment, these location data are not included in the generation of any derivative products and therefore the positional accuracy and potential errors do not contribute to the overall horizontal accuracy of the products (point clouds, DSMs, and orthomosaics). Horizontal accuracy likely decreases with distance from ground control. Ground control points surveyed with a network-based RTK-GNSS had a reported horizontal precision ranging from 0.01 to 0.06 meters. Though not an assessment of horizontal or vertical accuracy, horizontal positions of ground control points were used to calculate an RMS estimate of positional error at discrete locations by the photogrammetric software and are described here. Additionally, residuals between withheld ground control points (check points) and the model can serve as an indication of accuracy. Ground control and check point error is reported by the software with precision that is higher than provided by model inputs (RTK-GNSS precision) and thus has been rounded. For the January 30 model (FernRideLake_10cm), 47 ground control points were used to construct SfM datasets, and the model had a horizontal error of 14 centimeters, vertical error of 5 centimeters, and overall error of 15 centimeters. The 29 ground control points withheld as check points were compared to the model and had a horizontal error of 19 centimeters, vertical error of 27 centimeters, and overall error of 36 centimeters. For the January 31 model (FernRidgeLake_5cmLongTomRiver), 11 ground control points were used to construct SfM datasets, and the model had a horizontal error of 8 centimeters, vertical error of 6 centimeters, and total error of 10 centimeters. Three ground control point were withheld as check points and compared to the model and had a horizontal error of 13 centimeters, vertical error of 28 centimeters, and overall error of 31 centimeters. For the combined February 2 and February 8 model (FernRidgeLake_5cmCoyoteCreek), 19 ground control points were used to construct SfM datasets, and the model had a horizontal error of 16 centimeters, vertical error of 5 centimeters, and overall error of 16 centimeters. Eight ground control point were withheld as check points and compared to the model and had a horizontal error of 21 centimeters, vertical error of 30 centimeters, and overall error of 36 centimeters. Visual comparison of generated orthomosaics to publicly available aerial imagery indicated horizontal displacement was minimal for all datasets, although some obvious distortion is noticeable where vegetation is present or there was very little photograph overlap. Increasing noise at surface water locations occurs within each model; these areas were not edited, except for the aforementioned case of the low pool in the entire reservoir domain (FernRidgeLake_10cm). No formal vertical accuracy tests were performed. However, vertical accuracy likely decreases with distance from ground control. Ground control points surveyed with a network-based RTK-GPS had a reported vertical precision ranging from 0.02 to 0.10 meters. Though not an assessment of horizontal or vertical accuracy, horizontal positions of ground control points were used to calculate an RMS estimate of positional error at discrete locations by the photogrammetric software and are described here. Additionally, residuals between withheld ground control points and the model can serve as an indication of accuracy. Ground control and check point error is reported by the software with precision that is higher than provided by model inputs (RTK-GNSS precision) and thus has been rounded. For the January 30 model (FernRideLake_10cm), 47 ground control points were used to construct SfM datasets, and the model had a horizontal error of 14 centimeters, vertical error of 5 centimeters, and overall error of 15 centimeters. 29 ground control point were withheld as check points and compared to the model and had a horizontal error of 19 centimeters, vertical error of 27 centimeters, and overall error of 36 centimeters. For the January 31 model (FernRidgeLake_5cmLongTomRiver), 11 ground control points were used to construct SfM datasets, and the model had a horizontal error of 8 centimeters, vertical error of 6 centimeters, and total error of 10 centimeters. Three ground control point were withheld as check points and compared to the model and had a horizontal error of 13 centimeters, vertical error of 28 centimeters, and overall error of 31 centimeters. For the combined February 2 and February 8 model (FernRidgeLake_5cmCoyoteCreek), 19 ground control points were used to construct SfM datasets, and the model had a horizontal error of 16 centimeters, vertical error of 5 centimeters, and overall error of 16 centimeters. Eight ground control point were withheld as check points and compared to the model and had a horizontal error of 21 centimeters, vertical error of 30 centimeters, and overall error of 36 centimeters. Jin-Si R. Over Andrew C. Ritchie Christine J. Kranenburg Jenna A. Brown Daniel D. Buscombe Tom Noble Christopher R. Sherwood Jonathan A. Warrick Phillipe A. Wernette 2021 publication online U.S. Geological Survey https://doi.org/10.3133/ofr20211039 Digital and/or Hardcopy 20210614 publication date Over and others (2021) Software settings and processing procedures. Agisoft 2022 publication online Agisoft https://www.agisoft.com/pdf/metashape_1_8_en.pdf Digital and/or Hardcopy 20230306 publication date Agisoft (2022) Software settings and processing procedures. Geospatial Data Abstraction Library (GDAL) 2024 3.6.2 application/service online Geospatial Data Abstraction Library (GDAL) https://gdal.org/drivers/raster/cog.html#raster-cog https://gdal.org/api/python_bindings.html Digital and/or Hardcopy 20240218 publication date GDAL (2024) Raster conversion program for Cloud Optimized GeoTIFF creation. The USGS installed 39 temporary L-shaped control point targets throughout Fern Ridge Lake over January 17 and January 18, 2023. The photo target controls consisted of two, 1-foot wide by 3-feet long white garbage bags forming an L shape nailed to the ground with 6-inch-long nails. Additionally, 44 control points were selected from existing static infrastructure (for example, parking lot pavement arrows, retaining-wall corners) throughout the project area. 20230117 Control point targets were surveyed with a network-based RTK-GNSS at the inner point of the L on the target. 20230117 Aerial photographs were acquired by the Civil Air Patrol on January 30, between 14:46 to 16:26 Pacific Standard Time (PST) covering approximately 87.3 square kilometers of the reservoir area. Photographs were acquired as high-resolution JPG images with a WaldoAir XCAM Ultra 50 camera, consisting of two oblique-mounted Canon EOS 5DS R cameras that are triggered simultaneously, mounted to a Cessna 182 aircraft. A Novatel OEMStar GPS recorded camera positions in the image file EXIF data in UTC time zone. The pilot flew in longitudinal passes at about 1,000 meters above ground level over Fern Ridge Lake. A total of 1,741 photos were taken from takeoff to landing of flight. 20230130 Aerial photographs were acquired by the Civil Air Patrol on January 31, between 12:01 to 13:21 Pacific Standard Time (PST) covering approximately 17.6 square kilometers of the reservoir area. Photographs were acquired as high-resolution JPG images with a WaldoAir XCAM Ultra 50 camera, consisting of two oblique-mounted Canon EOS 5DS R cameras that are triggered simultaneously, mounted to a Cessna 182 aircraft. A Novatel OEMStar GPS recorded camera positions in the image file EXIF data in UTC time zone. The pilot flew a in longitudinal-oblique passes at about 600 meters above ground level over the Long Tom River arm of Fern Ridge Lake. A total of 1,212 photos were taken from takeoff to landing of flight. 20230131 Aerial photographs were acquired by the Civil Air Patrol on February 2, between 13:16 to 14:17 Pacific Standard Time (PST) and February 8, between 10:56 to 11:27 PST, collectively covering approximately 18 square kilometers of the reservoir area. Photographs were acquired as high-resolution JPG images with a WaldoAir XCAM Ultra 50 camera, consisting of two oblique-mounted Canon EOS 5DS R cameras that are triggered simultaneously, mounted to a Cessna 182 aircraft. A Novatel OEMStar GPS recorded camera positions in the image file EXIF data in UTC time zone. The pilot flew in near-longitudinal passes at about 500 meters above ground level over the Coyote Creek arm of Fern Ridge Lake. A total of 1,384 photos were taken from takeoff to landing of flight. 20230202 All 4,337 photographs from the January 30, January 31, February 2, and February 8 flights were added to Agisoft Metashape software (v. 1.8.4) for alignment to identify tie points (common pixels the software identifies between photos). All subsequent steps conducted in Agisoft Metashape software follow the workflow outlined in Over and others (2021). Camera groups were created based on the three domains (entire reservoir at 10 cm, Long Tom River at 5 cm, and Coyote Creek at 5 cm), but all photos remained in a single chunk. The coordinate system of the aerial photographs was converted from WGS84 (EPSG:4326) to NAD83(2011) UTM Zone 10 North (EPSG:6339) for the horizontal datum and projection and NAVD88 meters, GEOID 18 (EPSG:6318) for the vertical datum. The following reference settings were updated: measurement camera accuracy of 150 meters and marker accuracy of 0.1 meters; image coordinates marker accuracy of 1 pixel and tie point accuracy of 1 pixel. 2023 All aerial photographs were aligned together with an accuracy setting of "High", the generic preselection box checked, the reference preselection box checked and set to "Source," key point limit set to 60,000, and tie point limit set to 0 (meaning no tie point limit). The "Exclude stationary tie points," "Guided image matching," and "Adaptive camera model fitting" boxes were all left unchecked. 2023 The camera calibration model was optimized with the following coefficients: f, cx, cy, k1, k2, k3, p1, p2, and where f is focal length; cx and cy are the optimal point; k is radial distortion; and p translational distortion. 2023 A ground control point file (CSV) containing XYZ coordinates (NAD83(2011) UTM Zone 10 North (EPSG:6339) and NAVD88 meters, GEOID 18 (EPSG:6318)) was imported. The Agisoft software identified photos that contained each marker. Photographs were filtered by each marker and were inspected for marker placement relative to the ground control point visible in each photo. Markers were either left in their position and verified, moved to the correct position and verified, or removed. An even spatial distribution of ground control points were converted to check points for use in later model accuracy assessment (29 checkpoints out of the 76 survey points for full Fern Ridge Lake datasets; 3 out of the 14 for the Long Tom River coverage; and 8 out of the 27 for the Coyote Creek coverage). Ground control points 2, 16, 17, 36, 37, 49, 56, 57, 60, 63, 74, 79, and 82 were exclusively used as check points due to being duplicate field measurements with slight horizontal and vertical position variations or uncertain placement within aerial photographs. Camera GPS locations were excluded from further processing by unchecking (not disabling) all cameras in the reference-camera pane. 2023 The camera calibration model was optimized with the following coefficients: f, cx, cy, k1, k2, k3, p1, p2, where f is focal length; cx and cy are the optimal point; k is radial distortion; and p translational distortion. 2023 The sparse cloud of approximately 24 million photogrammetric tie points generated during alignment was gradually filtered based on several criteria of model fit to identify and remove points with high reprojection error, with camera calibration model optimization in between each filtering step. First, the point cloud was visually inspected for anomalous points which were selected then deleted. Second, points with a “Reconstruction uncertainty” greater than 10 were selected then deleted; this removed approximately 10 million points that were generated due to poor camera geometries. The camera calibration model was then optimized with the following coefficients: f, cx, cy, k1, k2, k3, p1, p2. Third, points with a “Projection accuracy” greater than 3.5 were selected then deleted; this removed approximately 7 million points that were generated due to pixel matching errors derived from variations in image scaling. The camera calibration model was then optimized with the following coefficients: f, cx, cy, k1, k2, k3, p1, p2. Last, points with a “Reprojection error” greater than 0.3 were iteratively selected and deleted, without exceeding 10% of all points with each selection; this removed approximately 2 million points in total that were incorrectly reprojected after alignment. The camera calibration model was then optimized with the following coefficients: f, cx, cy, k1, k2, k3, k4, p1, p2, b1, b2, with “Fit additional corrections” checked. The final unweighted Root Mean Square reprojection error for the tie point cloud was 0.34 pixels. 2023 Aligned, referenced aerial photographs and their filtered tie points were separated into the three domains (entire reservoir at 10 cm, Long Tom River at 5 cm, and Coyote Creek at 5 cm) by selecting the associated aerial photographs and separating them into individual chunks. 2023 Dense point clouds were built for each of the three domain chunks using the medium quality setting and mild depth filtering. Any anomalous or “floating” points were visually inspected and deleted. 2023 All SfM products were exported using their default resolutions. Dense point clouds and DSMs were exported for each of the three domain chunks using the default settings. Orthomosaics were exported without compression and saved as BigTIFF files. DSMs and orthomosaics were clipped using imported polygons to remove distorted areas or areas with missing raster values. 2023 Point Point 554482660 NAD 1983 UTM Zone 10N 0.9996 -123.0 0.0 500000.0 0.0 coordinate pair 0.000000002220024164500956 0.000000002220024164500956 meter D North American 1983 GRS 1980 6378137.0 298.257222101 North American Vertical Datum of 1988 (NAVD88) 0.001 meters Explicit elevation coordinate included with horizontal coordinates FernRidgeLake_10cm_2023_pointcloud.laz Elevation point cloud Other The attribute information associated with point clouds was output by the software in the LAZ file standard (LAS 1.2) with a point type of 2. Attributes include location (northing, easting, and elevation), color (red, blue, and green components), intensity (absent), and classification (all points unclassified). ASPRS 2013 U.S. Geological Survey - ScienceBase Mailing and Physical

Denver Federal Center

Building 810

Mail Stop 302

Denver CO 80225 United States 1-888-275-8747 sciencebase@usgs.gov Although these data have been used by the U.S. Geological Survey, U.S. Department of the Interior, no warranty expressed or implied is made by the U.S. Geological Survey as to the accuracy of the data. The act of distribution shall not constitute any such warranty, and no responsibility is assumed by the U.S. Geological Survey in the use of these data, software, or related materials. The use of firm, trade, or brand names in this report is for identification purposes only and does not constitute endorsement by the U.S. Geological Survey. The names mentioned in this document may be trademarks or registered trademarks of their respective trademark owners. LAZ The LAZ file contains the unclassified point cloud compressed with LASzip. Use LASzip, available from http://www.laszip.org or other software that can read laz files. 7600 https://doi.org/10.5066/P1Q5K657 None. This dataset is provided by USGS as a public service. 20260319 Schwid, Maxwel F. U.S. Geological Survey Hydrologist Mailing and Physical

601 SW 2nd Avenue

Suite 1950

Portland Oregon 97204 United States 503-758-5589 gs-w-or_sciencebase@usgs.gov FGDC Content Standard for Digital Geospatial Metadata FGDC-STD-001-1998 local time