Application of Lidar Altimetry and Hyperspectral Imaging to Ice Sheet and Snow Monitoring

Abstract

The Greenland Ice Sheet (GrIS) is of tremendous importance for climate change projections. The GrIS has contributed an estimated 10.8 mm to sea level rise since 1992, and that contribution is expected to increase in the coming decades. It is therefore essential to make routine measurements of ice, meltwater, and snow over the GrIS using satellite and airborne observations. Two prominent methods for ice sheet monitoring include lidar altimetry and hyperspectral imaging. Lidar altimetry is typically used to make fine-scale estimates of ice sheet surface height, whereas hyperspectral imaging is commonly utilized to infer snow or ice surface composition. In this dissertation, I use data from the Ice, Clouds, and land Elevation Satellite-2 (ICESat-2) and the Next Generation Airborne Visible/Infrared Imaging Spectrometer (AVIRIS-NG) to examine light transmittance over the Greenland Ice Sheet. I first utilize ICESat-2 photon-counting data for the development of a retrieval algorithm for supraglacial lake depth, with validation from the Operation IceBridge airborne mission. This work was performed to support other depth retrieval efforts that struggle with attenuation in deep water. I then use hyperspectral radiative transfer models to perform a sensitivity analysis on snow grain size retrievals. Snow grain size is an important metric for snowpack evolution, but there are limited efforts to quantify potential errors in an existing inversion algorithm. Lastly, I used a combination of Operation IceBridge altimetry and AVIRIS-NG hyperspectral data to assess the impacts of snow grain size on surface heights derived from lidar altimetry.

Results from the three studies indicate that lidar signals and ice reflectance in the near-infrared are highly sensitive to changes in surface media. Because it operates at 532 nm, the ICESat-2 laser penetrates through liquid water with minimal signal loss, but volumetric scattering within a snowpack may induce significant errors in surface heights derived from Operation IceBridge, especially at large snow grain sizes. The ICESat-2 laser is susceptible to noise from clouds and rough surface topography, so additional work is needed to accurately identify supraglacial lake beds and volumetric scattering caused by snow. Also, the near-infrared spectrum of snow is highly sensitive to changes in solar geometry and to the presence of dust, therefore increasing uncertainties in snow grain size retrievals. Co-dependencies between snowpack perturbations were not considered, but I speculate that snow particle shape and snow impurities will impact the angular distribution of radiation reflected from a snowpack. I expect that the research presented here will motivate the development of improved algorithms for supraglacial lake depth, snow grain size, and lidar altimetry bias.