Lab Home | Phone | Search
Center for Nonlinear Studies  Center for Nonlinear Studies
 Home 
 People 
 Current 
 Executive Committee 
 Postdocs 
 Visitors 
 Students 
 Research 
 Publications 
 Conferences 
 Workshops 
 Sponsorship 
 Talks 
 Seminars 
 Postdoc Seminars Archive 
 Quantum Lunch 
 Quantum Lunch Archive 
 P/T Colloquia 
 Archive 
 Ulam Scholar 
 
 Postdoc Nominations 
 Student Requests 
 Student Program 
 Visitor Requests 
 Description 
 Past Visitors 
 Services 
 General 
 
 History of CNLS 
 
 Maps, Directions 
 CNLS Office 
 T-Division 
 LANL 
 
Monday, August 20, 2018
11:00 AM - 12:00 PM
CNLS Conference Room (TA-3, Bldg 1690)

Seminar

Machine learning, soft computing, and complex systems analysis: emerging approaches for discovering and predicting nonlinear phenomena in water resources and climate

Dr. Sean W. Fleming
White Rabbit R&D LLC, Oregon State University, University of British Columbia, Los Alamos National Laboratory

Water and climate have always been existential drivers of human civilization and today are more relevant than ever as both enablers of progress and wealth and threats to safety and well-being. What can modern data analytics teach us about these things? Geophysical processes are, in general, nonlinear phenomena generated by complex open systems, and some threads in academic research have long recognized the power of advanced mathematical concepts and computational data analytics to illuminate their presence and inner workings. However, mainstream research and practical applications often employ streamlined and, typically, linearized representations and methods. There can be powerful logistical motivations and justifications for doing so, but over the long run, it may also hamstring progress. The increasing accessibility of a rapidly growing and diversifying portfolio of powerful data science techniques can facilitate bridging this gap between theory and practice – provided that data-driven discovery and prediction methods are used in a physics-aware manner, and with an eye to providing demonstrably superior practical solutions. This introductory tour, though very far from being in any sense comprehensive, offers a few examples to illustrate these new developments: using neural networks to detect and understand high-frequency nonlinear dynamics in river flow and to perform operational flood forecasting; using information-theoretic methods to discover and characterize nonlinear climate teleconnections to northern-hemisphere water supply; discovering nonlinear phase transition-like behavior and filtering effects in local-scale hydroclimatology using exploratory nonparametric statistical data analysis; applying complex network theory to understand regional-scale hydrometeorological patterns and optimize environmental monitoring systems; building fuzzy expert systems to integrate nonlinear experiential knowledge and holistically summarize and track environmental conditions; and applying a suite of methods to instrumental and paleoclimatic records to work toward coming to terms with fractal, chaotic, and nonstationary dynamics in water and climate.

Host: Katrina E. Bennett