Pan-Antarctic Assessment of Sedimentary Basins and the Onset of Streaming Ice Flow from Machine Learning and Aerogravity Regression Analyses
Start Date:
2021-09-01
End Date:
2023-08-31
Description/Abstract
This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).
An important part of understanding future climate change is predicting changes in how fast the ice in Antarctica is moving. If ice flows more quickly towards the ocean, it will have a direct impact on sea level rise. One of the things that can influence the ice flow is the type of rock below the ice coverage in Antarctica. Sedimentary basins are large regions where sedimentary rocks accumulated in the past, often under ancient seas. It has been observed that where there are sediments below the ice, the ice can flow faster. This project seeks to understand what is below the ice and how the underlying rock influences the ice flow. Is it hard, crystalline rock? Is it a sedimentary basin? What is the relationship between sediments and ice flow? The answers to these questions will be addressed by using a combination of available data and geophysical methods. Information from well-known rock-types will be used to train the computer to recognize these features by using an application of artificial intelligence known as machine learning, which will help the characterization and identification of unknown sedimentary basins beneath the ice. The results of this project will be disseminated to a broad audience by holding workshops for teacher and students to explain our findings under the ice and to introduce the machine learning technique. Open-source codes used during this project will be made available for use in higher-level classrooms as well as in further studies.
To date, no comprehensive distribution of onshore and offshore sedimentary basins over Antarctica has been developed. A combination of large-scale datasets will be used to characterize known basins and identify new sedimentary basins to produce the first continent-wide mapping of sedimentary basins and provide improved basal parametrizations conditions that have the potential to support more realistic ice sheet models. Available geophysical compilations of data and the location of well-known sedimentary basins will be used to apply an ensemble machine learning algorithm. The machine learning algorithm will learn complex relationships by voting among a collection of randomized decision trees. The gravity signal related to sedimentary basins known from other (e.g. seismic) techniques will be evaluated and unknown basins from aerogravity data regression analyses will be proposed by calculating a gravity residual that reflects density inhomogeneities. The gravimetric sedimentary basins identified from the regression analyses will be compared with an independent method of identifying sedimentary distribution, the Werner deconvolution method of estimating depth to magnetic sources. The hypothesis, which is sedimentary basins are correlated to fast ice flow behavior, will be tested by comparing the location of the sedimentary basins with locations of high ice flow by using available ice velocity observations. A relationship between sedimentary basins and ice streams will be defined qualitatively and quantitatively, aiming to evaluate if there are ice streams where no sedimentary basins are reported, or sedimentary basins with no ice streams related. The findings of these project can confirm if the presence of abundant sediments is a pre-requisite for ice streaming. Analyzing previously known sedimentary basins and identifying new ones in Antarctica is central to evaluating the influence of subglacial sediments on the ice sheet flow.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
Personnel
Funding
AMD - DIF Record(s)
Data Management Plan
None in the Database
Product Level:
0 (raw data)
Keywords
|
This project has been viewed 24 times since May 2019 (based on unique date-IP combinations)