C005-0017
Stereo2SWE: Snow depth and snow-covered area from commercial stereo satellite imagery

Monday, 7 December 2020
Poster
David E Shean1, J Michelle Hu2, Shashank Bhushan2, Oleg Alexandrov3, Christopher A Hiemstra4, Scott T Henderson5 and Lundquist Jessica1, (1)University of Washington, Civil and Environmental Engineering, Seattle, WA, United States, (2)University of Washington, Seattle, WA, United States, (3)NASA Ames Research Center, Intelligent Robotics Group, Moffett Field, CA, United States, (4)US Army Corps of Engineers Washington DC, Washington, DC, United States, (5)University of Washington Seattle Campus, Seattle, WA, United States
Abstract:
Seasonal snow is an important component of the cryosphere and larger hydrologic cycle. A comprehensive understanding of its spatiotemporal evolution is needed for effective water resource management. Traditionally, snow depth and snow water equivalent (SWE) has been estimated using sparse in situ measurements and/or parameterized models calibrated with low-resolution remote sensing data. The NASA Airborne Snow Observatory (ASO) now provides operational snow depth maps with high accuracy and spatial resolution from repeat LiDAR surveys, but coverage is relatively limited and logistical costs are relatively high.

To complement these efforts, we coordinated tasking campaigns for available very-high-resolution (~0.3 to ~0.9 m) commercial satellite optical stereo imagery from MAXAR/DigitalGlobe WorldView-1/2/3, Planet SkySat-C, and Airbus Pleiades. We developed automated, open-source workflows to generate co-registered time series of digital surface models (DSMs) from these data to measure snow depth evolution. Individual DSM products have <20-50 cm vertical accuracy and we are performing on-orbit calibration, developing sub-pixel image corrections, integrating satellite altimetry data for validation, and refining data collection strategies with the goal of achieving ~10-15 cm accuracy.

We are also developing machine-learning algorithms for landcover classification (e.g., snow, forest, bare ground) using the complementary 4/8-band multispectral imagery (and 8-band SWIR for WorldView-3) from these sensors. These products will enable improved stereo processing, DSM co-registration, and snow-covered area mapping.

We present results for NASA SnowEx sites and USGS glacier monitoring sites in the Western U.S. Preliminary comparisons show good agreement between the satellite stereo snow depth products and ASO, ICESat-2 altimetry, in situ GPR, and in situ probe measurements. We offer an evaluation of currently operational satellite stereo and altimetry resources and their potential to produce accurate, high-resolution maps of snow depth and SWE on a global scale, with recommendations for future mission concept design efforts.

<< Previous Abstract | Next Abstract >>