ICESat2 subsurface-scattering biases estimated based on the 2015 SIMPL/AVRIS campaign

Abstract:

One of the most important challenges facing the ICESat-2 mission is to identify and minimize the effects of potential biases in estimates of ice surface heights. The mission is designed to make very precise estimates of ice-sheet volume changes, and small, spatially correlated biases in the surface height measurements can produce large spurious volume-change signals. One potentially significant signal comes from the subsurface scattering of the emitted laser pulse. Because pure ice is nearly transparent at the instrument’s 532 nm (green) operating wavelength, a photon from ICESat2’s laser can be scattered by hundreds or thousands of ice grains within the snowpack before it returns to the surface and travels back to the satellite.. The time required for this leads to a volume scattering bias, The magnitude of the volume scattering bias depends on the relative fractions of photons returning from near the surface and those scattered many times at depth, which in turn depends on the effective grain size of the snow and ice, the impurity content, and the vertical distribution of each.

Here we present the results of the first ice-sheet survey designed to measure the magnitude of this potential problem. The NASA ICESat-2 project conducted an airborne survey in July, 2015 over the west coast of Greenland, using the Slope Imaging Multi-polarization Photon-counting Lidar (SIMPL). The SIMPL instrument makes photon-counting laser-altimetry measurements at two different wavelengths (532 and 1064 nm) using a common optical path to ensure accurate collocation. The infrared channel (1064 nm) was used in addition to the green because it is insensitive to volume scattering. These measurements were complimented by Airborne Visible/Infrared Imaging Spectrometer (AVRIS) measurements, intended to quantify variations in surface grain size and impurity content. Surveys were designed to measure a variety of surface conditions, from fine-grained snow at high elevations in the north to dirty, melting ice surfaces in the south. To derive maps of potential biases, we measure the spatial variation of the range difference between SIMPL’s green and infrared channels, and relate this to the surface conditions as measured by AVRIS.