Revisiting NASA's Operation IceBridge Snow on Sea Ice Radar Measurements in the Arctic

Snow on sea ice plays a critical role in modulating ice mass changes in response to anthropogenic warming, with significant implications for ocean mixed layer processes, the surface energy budget, and marine ecosystems. Most importantly, accurate snow depth measurements are essential for deriving reliable sea ice thickness estimates from all altimetry satellites. Operation IceBridge (OIB), which collected snow depth data using the airborne CReSIS FMCW C/S-band snow radar for a decade, remains a pivotal reference for understanding pan-Arctic snow depth changes and validating remote sensing snow retrievals. Despite its importance, significant concerns persist regarding snow retrieval algorithms from snow radar, particularly around algorithm performance and the representation of snow properties.

In this study, we revisit OIB snow depth retrieval algorithms by comparing them with underutilized in-situ snow depth measurements from MagnaProbe surveys conducted near Eureka, Canada, in 2016. To enhance the spatial representation of the in-situ data, we employ Kriging interpolation methods. Additionally, we make use of the co-collected conical laser scanner data. A detailed comparison of retrieval algorithms - focusing on the detection of the air-snow and snow-ice interfaces as well as the derived snow depth - reveals that the Continuous Wavelet Transform (CWT) algorithm performs best for the 2-8 GHz snow radar version, yielding a correlation of R=0.72 over undeformed sea ice. However, the CWT algorithm predominantly detects snow depths within the 80-90% quantile of the in-situ distribution within the radar footprint. This bias is attributed to the air-snow interface being identified as the first rise above the radar noise floor, which typically corresponds to the highest snow elevations within the footprint. Finally, we compare a subset of newly derived snow depth data from OIB  including highly-valuable uncertainties with existing datasets, highlighting potential improvements.

Looking ahead, we propose a framework to enhance snow depth retrieval algorithms, offering robust pathways for validating and improving satellite-based snow datasets. This approach holds significant promise for advancing the accuracy of snow depth measurements critical to polar science in the future campaigns.