High-Albedo Salt Crusts on the Tropical Ocean of Snowball Earth: Measurements and Modeling

Abstract:

During a Snowball Earth event, almost all of the ocean surface first freezes as sea ice. As in modern sea ice, trapped inclusions of liquid brine permeate the ice cover. As the ice grows and cools, salt crystals precipitate within the inclusions. At -23C, the most abundant salt in seawater, sodium chloride, begins to precipitate as the dihydrate mineral hydrohalite (NaCl·2H2O).

Crystals of hydrohalite within the sea ice scatter light. Measurements of cold, natural sea ice show a broadband albedo increase of 10-20% when salt precipitates. Such snow-free natural sea ice with a surface temperature below -23C is rare on modern Earth, but would have been common in tropical regions of a Snowball Earth where evaporation exceeded precipitation. The persistent cold and lack of summer melt on the Snowball ocean surface, combined with net evaporation, is hypothesized to yield lag deposits of hydrohalite crystals on the ice surface.

To investigate this process, we prepared laboratory-grown sea ice in a 1000 liter tank in a walk-in freezer laboratory. The ice was cooled below -23 C and the surface sprayed with a 23% NaCl solution to create a layer of hydrohalite-enriched ice, a proxy for lag deposits that would have formed over long periods of surface sublimation. We have developed a novel technique for measuring the spectral albedo of ice surfaces in the laboratory; this technique was used to monitor the evolution of the surface albedo of our salt crust as the ice matrix sublimated away leaving a layer of fine-grained hydrohalite crystals. Measurements of this hydrohalite surface crust show a very high albedo, comparable to fresh snow at visible wavelengths and significantly larger than fresh snow at near infrared wavelengths. Broadband albedos are 0.55 for bare artificial sea ice at -30C, 0.75 for ice containing 25% hydrohalite by volume, 0.84 after five days of desiccation and 0.93 after 47 days of desiccation.

Using our laboratory measurements, along with estimates of grain size and crust optical depth, as inputs to Mie scattering and radiative transfer models allowed us to infer the imaginary refractive index of hydrohalite. The model can calculate albedo for pure hydrohalite crusts of varying thickness and for mixtures of ice and hydrohalite. A parameterization is presented for albedo as a function of the thickness of the hydrohalite crust.