elib
DLR-Header
DLR-Logo -> http://www.dlr.de
DLR Portal Home | Imprint | Privacy Policy | Contact | Deutsch
Fontsize: [-] Text [+]

Deep learning for surrogate modelling of 2D mantle convection

Agarwal, Siddhant and Tosi, Nicola and Kessel, P and Breuer, Doris and Montavon, Grégoire (2021) Deep learning for surrogate modelling of 2D mantle convection. German-Swiss Geodynamics Workshop 2021, 29 Aug–1 Sep 2021, Bad Belzig, Deutschland.

Full text not available from this repository.

Abstract

The key parameters to mantle convection simulations are poorly constrained. Whereas the outputs can sometimes be observed directly or indirectly using geophysical and geochemical data obtained via planetary space missions. Hence, the “observables” can be used to constrain parameters governing mantle convection. Given the computational cost of running each forward model in 2D or 3D (on the scale of hours to days), it is often impractical to run several thousands of simulations to determine which parameters can satisfy a set of given observational constraints. Traditionally, scaling laws have been used as a low-fidelity alternative to overcome this computational bottleneck. However, they are limited in the amount of physics they can capture and only predict mean quantities such as surface heat flux and mantle temperature instead of spatio-temporally resolved flows. Using a dataset of 10,000 2D mantle convection simulations for a Mars-like planet, we show that deep learning can be used to reliably predict the entire 2D temperature field at any point in the evolution. We first use convolutional autoencoders to compress each temperature field by a factor of 140 to a latent space representation, which is easier to predict. We test feedforward neural networks (FNN) to predict the compressed temperature fields from five input parameters plus time. While the mean accuracy of the predicted temperature fields was high (99.30%), FNN was unable to capture the sharper downwelling structures and their advection. To address this, we use long short-term memory networks (LSTM), which have recently been shown to work in a variety of fluid dynamics problems. In comparison to the FNN, LSTM achieved a slightly lower mean relative accuracy, but captured the spatio-temporal dynamics more accurately.

Item URL in elib:https://elib.dlr.de/146295/
Document Type:Conference or Workshop Item (Speech)
Title:Deep learning for surrogate modelling of 2D mantle convection
Authors:
AuthorsInstitution or Email of AuthorsAuthor's ORCID iD
Agarwal, SiddhantSiddhant.Agarwal (at) dlr.dehttps://orcid.org/0000-0002-0840-2114
Tosi, Nicolanicola.tosi (at) dlr.dehttps://orcid.org/0000-0002-4912-2848
Kessel, PTechnical University BerlinUNSPECIFIED
Breuer, DorisDoris.Breuer (at) dlr.dehttps://orcid.org/0000-0001-9019-5304
Montavon, GrégoireInstitut für Softwaretechnik und Theoretische Informatik, Technische Universität BerlinUNSPECIFIED
Date:August 2021
Refereed publication:No
Open Access:No
Gold Open Access:No
In SCOPUS:No
In ISI Web of Science:No
Status:Published
Keywords:Mantle Convection, Machine Learning, Fluid Dynamics, Surrogate Modelling, Neural Networks
Event Title:German-Swiss Geodynamics Workshop 2021
Event Location:Bad Belzig, Deutschland
Event Type:Workshop
Event Dates:29 Aug–1 Sep 2021
Organizer:Deutsche Geophysikalische Gesellschaft
HGF - Research field:Aeronautics, Space and Transport
HGF - Program:Space
HGF - Program Themes:Space Exploration
DLR - Research area:Raumfahrt
DLR - Program:R EW - Space Exploration
DLR - Research theme (Project):R - Exploration of the Solar System
Location: Berlin-Adlershof
Institutes and Institutions:Institute of Planetary Research > Planetary Physics
Deposited By: Agarwal, Siddhant
Deposited On:26 Nov 2021 13:03
Last Modified:26 Nov 2021 13:03

Repository Staff Only: item control page

Browse
Search
Help & Contact
Information
electronic library is running on EPrints 3.3.12
Copyright © 2008-2017 German Aerospace Center (DLR). All rights reserved.