The Allan Hill region of Antarctica has produced the oldest ice core samples ever recovered, which provide insights into Earth’s climate history prior to the existing 800,000 year ice core record. However, the highly disturbed nature of this ice complicates straightforward dating and interpretation. Understanding the scales of preserved climate records in this old ice will enable deeper insights into the variability of climate over the last 2 million years. Here I study a section of ice from ALHIC1901, an ice core recovered from the Allan Hills in 2019. This section has three parallel sets of water isotope measurements, and they all show a small but significant dip. However, the cause of this dip remains unclear. The goal of this study is to test whether this isotope change could be a glacial-interglacial transition preserved for 1.3 million years, or whether diffusion should have eliminated any climate signal. To investigate this question, I apply a simple water isotope diffusion model that takes as inputs temperature, an initial water isotope profile, and a thinning history, and provides as an output the resulting water isotope profile after a given number of years. I identify the range of possible temperature, thinning, and initial water isotope signals for this ice. I use these as inputs for the diffusion model, and compare the results to the ice core record to evaluate if the observed water isotope signal can be climatically driven. Constraining the cause of this water isotope signal will improve our understanding of fine scale paleoclimate proxy changes in the extremely old Allan Hills ice cores, enabling new insights into past climate variability.

Old (> 4 million years ago) ice drilled at the Allan Hills, Antarctica, can help us understand how Earth’s atmosphere has changed in the past. The bubbles trap bits of the atmosphere when they form, which can be analyzed to see what the climate was like. However, the preservation of this old ice depends on ice flow dynamics, possibly including localized shearing (one side of the ice is getting pulled faster than the other), that are difficult to observe. Bubbles in the ice become elongated when the ice around them deforms from strain. Over time, surface tension processes tend to restore bubbles to spherical. Thus, they can indicate the directions of recent or ongoing strain in the ice. We analyze thin/thick section images taken from four samples of Allan Hills ice. The images include information on grain size (size of individual ice crystals) and bubble size, shape, and distribution. We use the Segment Every Grain (SEG) model, a Python package based on the Segment Anything Model developed by Meta, to automatically calculate the sizes and shapes of bubbles in an image. We validate this method by comparing the values it returns with those obtained using another segmenting software, ImageJ, and manually calculated measurements. We can see if automated calculations are reliable enough to use regularly. So far, the SEG model has analyzed one image and it has made mostly correct bubble identification. The data shows that most bubbles are either elongated and small area, or round and large area. It is expected that the SEG and ImageJ models are close to humans in accuracy. The bubble orientations that we measure show the predominant directions of strain in the ice. Future work will use these data along with models of bubble elongation to estimate the strain rates at the Allan Hills.

The Mid-Pleistocene Transition (MPT) was a major climatic shift in Earth’s history occurring between 1.2 and 0.7 million years ago. During the MPT, Earth’s glacial cycles shifted from a high frequency (~40 kyr), low amplitude cadence to a low frequency (~100 kyr), high amplitude cadence which has dominated since the MPT. While we are able to observe the MPT in benthic 𝛿¹⁸O records, our current ice core record only extends back 800 kyr and does not include a preservation of the entire MPT. COLDEX, a multi-institution collaboration, is seeking to find a region in Antarctica where a continuous deep ice core may preserve the MPT in order to better understand the underlying mechanisms that caused it. Ice at this depth, however, is subject to much different conditions than the ice cores that comprise our current record. My study aims to analyze how atmospheric gases, namely CO₂ and the 𝛿O₂/N₂ ratio (used to identify precessional cycles for dating ice cores), diffuse in Antarctic ice of 1 to 1.5 million years old. I focused on the COLDEX survey region between the South Pole and Dome A. I employed two models: 1) a one-dimensional steady state model which calculates the temperature and age of the ice with respect to depth, and 2) a gas-diffusion model which uses the temperature- and age-depth relations to calculate the amplitude of the gas signals in the ice through time. The input parameters for these models are measured using aerial radar, provided by COLDEX, or interpolated accordingly. So far, I have found that CO₂ is relatively well preserved in the region, while the 𝛿O₂/N₂ ratio is much less well preserved. This suggests that finding an ideal region for a deep ice core drill site, on the basis of gas diffusion, may be difficult.