Historical Fairbanks lake shorelines, 1949-2009

• Main Authors: Walter-Anthony, Katey, Lindgren, Prajna
• Format: Dataset
• Language: English
• Published: NSF Arctic Data Center 2021
• Subjects: Alaska; lake; perimeter; Fairbanks; permafrost
• Online Access: https://dx.doi.org/10.18739/a2x63b62j; https://arcticdata.io/catalog/view/doi:10.18739/A2X63B62J

This data set consists of 4 shapefiles of historical lake shorelines in the 214 square kilometers (km2) Pleistocene-aged yedoma permafrost study area in Goldstream Valley and the surrounding landscape just north of Fairbanks in interior Alaska (Fig. 1). Our work focused on the southern extent of the valley in order to exclude anthropogenic lakes formed by mining peat, gravel, and gold that are concentrated in the northwestern extent of Goldstream Valley. We mapped lake perimeters from four observation times spanning 60 years (1949 to 2009). We mapped historic lake margins using high-resolution (1-meter (m) spatial resolution) single-band airborne optical images acquired in 1949, 1967 and 1985 available through U.S. Geological Survey (USGS) Earth Explorer (https://pubs.usgs.gov/gip/AerialPhotos_SatImages/aerial.html). We used high-resolution orthomosaiked (2.5 m spatial resolution) multi-band Satellite Pour l’Observation de la Terre (SPOT; visible and infrared) satellite optical images acquired in 2009 to map recent lake boundaries. Images were multispectral, except 1949, which was panchromatic. We applied a semiautomated adaptable mapping solution based on the Object-based Image Analysis (OBIA) technique in the eCognition (version 8) image processing software to map historic and recent lake margins on both aerial and SPOT images (Navulur 2007, Blaschke and Strobl 2001, Lindgren et al 2019). The initial step of the OBIA technique involved creation of highly homogeneous image objects or segments that met certain spatial and spectral homogeneity criteria of various scales depending on the regional lake properties. For historic lake margin delineation, we segmented the airborne historic images into image objects followed by a simple threshold classification technique based on brightness of the segmented image objects. To map modern lake margins, after segmentation of SPOT image bands, we applied thresholds on the bands transformed using Principal Component Analysis (PCA), an image enhancement technique applied for spectral separation of water bodies from the surrounding landscape features (Mather and Koch 2011). In addition, we applied the Normalized Difference Water Index (NDWI) (McFeeters 1996) to further classify waterbodies. Finally, historic and recent water body segments that were not lakes, such as streams and rivers and other spectrally similar misclassified objects, such as shadows, were removed through manual editing. More details on our data processing workflows can be found in Lindgren et al (2019).