What is the Rate-Limiting Step In furthering our Understanding of Subduction Dynamics?

What is the Rate-Limiting Step In furthering our Understanding of Subduction Dynamics? Magali Billen - Katrina Arredondo, - Juan Rodriguez, - Erin Burkett, --Margerete Jadamec

Processes in Series are Rate-Limited by Slowest Sub-Process? Visco-elastic Deformation of a Maxwell Solid Rate-limited by the slow response of the viscous dashpot Diffusion of Vacancies Through a Grain Rate-limited by slowest diffusing species Leaving for Work in the Morning Rate-limited by my son s willingness to cooperate

Sub-Processes in Subduction Studies COMPILING OBSERVATIONS Two sets of observations: 1. Defining Initial State of Models geometry, thermal structure, age, material properties from experiments 2. Testing/Analyzing Model Results plate/trench motion, mantle flow, stress-state, topography, gravity, magmatism, sedimentary record, seismicity, tomography

Sub-Processes in Subduction Studies COMPILING OBSERVATIONS DESIGNING NUMERICAL MODEL Equations, model size, mesh, boundary conditions, initial conditions, instantaneous, timedependent, material properties, Necessary simplifications??

Sub-Processes in Subduction Studies COMPILING OBSERVATIONS DESIGNING NUMERICAL MODEL RUNNING SERIES OF NUMERICAL EXPERIMENTS Which parameters will be held fixed, or varied? What is range of uncertainty? Which observables are sensitive to specific parameters?

Sub-Processes in Subduction Studies COMPILING OBSERVATIONS DESIGNING NUMERICAL MODEL RUNNING SERIES OF NUMERICAL EXPERIMENTS Two Types of Analysis 1. Generalized fluid dynamics, mechanics What happened? 2. Comparison to observations Is this applicable to the Earth? What does it tell us about the parameters and processes? ANALYSIS OF MODELING RESULTS NEW UNDERSTANDING OF SUBDUCTION DYNAMICS What is the rate - limiting step for discovery?

Outline Rate-Limiting Process in Subduction Studies? Some Outstanding Questions in Subduction Dynamics Example Studies: Integrating Observations, Numerical Modeling and Visualization Cyber-infrastructure for Subduction Modeling? Accelerating the rate-limiting process. Conclusions

Question 1: How do slabs drive plate motion? What are the available driving forces? How are forces transmitted to the plate? What are resisting forces? What are the dynamic feedbacks that regulate plate speeds?

Question 2: What are the main controls on non-steady-state evolution of slabs? What is the effect of large-scale rheological structure (layered, slab)? What are the effects of phase changes & related processes (grain-size reduction, latent heat, localize torques on slab)? How is slab motion reflected in plate boundary motion, structure & deformation of overriding plate?

Question 3: How do slabs drive flow in the surrounding mantle? What determines the ratio of toroidal-to-poloidal flow in subduction zones? What conditions are necessary to generate significant trench-parallel flow?

Example 1: Snapshot of Flow & Deformation How does flow at a slab edge depend on rheology? Observations required to build model: Slab shape seismicity, tomography, tectonic history Thermal structure subducting plate age, overriding plate heat flow, terrane data, volcanism, magnetic data, crustal thickness from seismic reflection/refraction data

Example 1: Snapshot of Flow & Deformation Observation used to test model: Seismic anisotropy Observed plate motion Regions of uplift/subsidence (dynamic topography) Volcanism Need non-newtonian rheology, sharp slab edge, moderate plate boundary coupling, moderate slab yield stress to match suite of observations.

Example 2: Ridge-Trench Interaction What controls the process of slab detachment and ridge abandonment? Observations to constrain: plate age at trench, ridge-trench geometry, volcanism & deformation of upper plate, location of fossil slab beneath Baja.

Example 2: Ridge-Trench Interaction Slab detachment controlled by yield strength of the lithosphere. Older plate detaches first. Younger plate viscously stretches. Upwelling of hot mantle through slab gap.

Example 3: Slab Flattening & Slab-Parallel Flow Can variations in overriding plate thermal structure affect slab dip & mantle flow? Observations: some regions with colder overriding plate also have shallow slab dips which came first?

Example 3: Slab Flattening & Slab-Parallel Flow 3D Model Set-up Proto-slab Generic subduction zone set-up Create proto-slab with kinematic boundary conditions, then let run fully dynamically. Vary geometry and temperature structure of cold portion of overriding plate.

Example 3: Slab Flattening & Slab-Parallel Flow Variation in overriding plate thermal structure Causes shallowing of slab beneath colder overriding plate. 20 degree shallower Small initial slab-parallel flow Slab-parallel flow as strengthens as slab shape changes Slab-parallel flow beneath slab & in mantle wedge Result of non-steadystate deformation & flow. Compare with location of seismic anisotropy.

Outline Rate-Limiting Process in Subduction Studies? Some Outstanding Questions in Subduction Dynamics Example Studies: Integrating Observations, Numerical Modeling and Visualization Cyber-infrastructure for Subduction Modeling? Accelerating the rate-limiting processes. Conclusions

What do these examples have in common? State-of-the-art numerical modeling software 3D models very large model output (gigabytes per time-step!!) Visualization of data & model results is key to model analysis communication of results Observations inform model design used to test/analyze model results

Cyber-infrastructure for Subduction Modeling? What do we need? Access To And Sharing Of: NUMERICAL MODELING SOFTWARE VISUALIZATION/ ANALYSIS TOOLS DATA

Cyber-infrastructure for Subduction Modeling? What do we need? Access To And Sharing Of: NUMERICAL MODELING SOFTWARE VISUALIZATION/ ANALYSIS TOOLS DATA A few years ago, I would have said that new/better software is the rate-limiting process. Computational Infrastructure for Geodynamics Share and support existing community-developed codes Develop new cutting-edge codes Train next generation of geodynamicists

Cyber-infrastructure for Subduction Modeling? What do we need? Access To And Sharing Of: NUMERICAL MODELING SOFTWARE VISUALIZATION/ ANALYSIS TOOLS DATA Generic Encourage science-driven development of visualization /analysis tools Specialized Maps/2D 3D

Cyber-infrastructure for Subduction Modeling? What do we need? Access To And Sharing Of: NUMERICAL MODELING SOFTWARE VISUALIZATION/ ANALYSIS TOOLS DATA Data Access is the rate-limiting step for furthering understanding of subduction dynamics. 3D Volume/Point/Tensor Data 2D Map/Line/Point Data

Removing the re from Data Research The beginning of each modeling project is a hunt for data... Scour the literature for most recent data. Is data available? Yes excellent - download! No - try to contact PI willing/ready to share data??... followed by a lot of data combining and processing. What format is this data in? How can I combine this with N other data sets? What is best way to visualize data to look for patterns related to different processes? How do I extract derived quantities (e.g., surfaces) to use as input into numerical model?

Infrastructure for Data Access and Sharing How is Data Used in Model Design & Analysis? How is Data Accessed/Extracted? Need Up-to-date Data and Data Options. Enable flexiblity, but anticipate how scientific questions require data.

Need to make contributing data EASY. Success depends on making EarthCUBE THE place to contribute data: Motivation for PIs contributing data : Need to know that data accessed here will lead to referencing of related published papers. How can you make data contributing easy when there are so many different types of data? Point sets Maps Volumes Vectors Tensors Create a framework for the community to develop and share this infrastructure.

Conclusions Modeling of subduction dynamics is motivated by a range of scientific questions: How do slab drive plate motions? What are the main controls on slab evolution? How slabs drive surrounding mantle flow? Modeling subduction dynamics requires Cutting-edge modeling software (CIG) Flexible 2D/3D visualization tools (support science-motivated development) Data, lots-of, lots-of data (EarthCUBE) Acquiring and manipulating data into a useful form can be the rate-limiting sub-process Facilitate grass-roots development of visualization and data infrastructure that is driven by how the science is actually being done.