Chris Chang (Graduate student in ATS)
Chris is a master student working on the linear stochastic dynamics and its application to the MJO initiation.
Chaos and Predictability
Our Lab focuses on the predictability and future projections of extreme weather from a dynamical system perspective. Using hierarchical modeling framework, data-driven method and statistical mechanics, we study the geophysical fluid dynamics across different temporal and spatial scales, such as tropical intraseasonal variability, teleconnections, baroclinic waves, and idealized 2D turbulence. Detailed information can be found in “Publications”.
PI, Kai-Chih Tseng got his PhD in June 2019 from Colorado State University, where the advisors were Dr. Elizabeth Barnes and Dr. Eric Maloney.
After Ph.D., Kai-Chih Tseng worked as a postdoctoral research associate at Princeton University and NOAA Geophysical Fluid Dynamics Laboratory with Dr. Nathaniel Johnson.
News
2023/09/05 Yu-Chuan Kan wins 2nd place in student poster competition of NTU AS summer program
2023/06/15 Dr. Kai-Chih Tseng is awarded 2030 Cross Generation Young Scholar Program by NSTC
2022/08/01 Dr. Kai-Chih Tseng is awarded Yushang Young Scholar by MOE.
Group Members
Yun-Hsuan Ho (Research Assistant in ATS)
Yun-Hsuan is a research assistant working on the prediciton and predictability of subtropical high. His research especially focuses on the anomalously strong case in 2020 which caused the record drought in Taiwan.
Jin-Yu Chang (Graduate student in ATS)
Jin-Yu Chang is an incoming M.S. student working on moist baroclinic wave and annular mode dynamics (using GFDL DyCore) and the related predictability.
Hsuan-Cheng Lin (Graduate student in ATS)
Hsuan-Cheng Lin is an incoming M.S. student working on monsoon depression/onset with model hierarchy and machine learning.
Ray Kuo (Undergrad student in ATS and Math)
Ray Kuo is an undergraduate student working on the application of ML-based statistical mechanics in idealized 2D turbulence
Yi-Ann Feng (Undergrad student in ATS and Math)
Yi-Ann Feng is an undergraduate student studying the numerical solution of 2D and 3D turbulence predictability
Yu-Chuan Kan (Undergrad student in ATS)
Yu-Chuan Kan is an undergraduate student studying the interaction between cumulus/cloud radiative feedback and tropical convectively coupled waves
Recent Work
Machine Learning, Statistical Mechanics and Weather Forecast
Weather is intrinsiclly unpredictable due to its chaotic nature. Thus, using a probability density function (PDF) to describe the time evolution of forecast error is central to the prediction problem. However, generating enough ensemble members to approximate the entire PDF can be computationally expensive and unaffordable due to limited window of time for decision making. In this study, we propose a proof-of-concept of generating infinity ensemble members by combing machine learning and statistical mechanics.
Tropical-extratropical Interaction
The dynamics of tropical-extratropical teleconnection has been well-studied since 1980 through the lens of linear Rossby wave theory. However, it's connection with predictability of extratropical weather and ensemble spread remain uncertain. In our research, we demonstrate the spread of a dynamical system (such as hydrological extreme or extratropical geopotential height) can be explained through linear wave dynamics, which bridges the gap of deterministic forecast and probabilistic forecast link
S2S prediction with Fluctuation Dissipation Theorem (FDT)
The use of fluctuation dissipation theorem can be traced back to Annus mirabilis in 1905. The year that Albert Einstein published 5 legendary paper which changed how we observe the world. Fluctuation Dissipation Theorem is one of them which transfers Newton's Law from a deterministic perspective to a probabilistic perspective. In this study, a linear inverse model (LIM, or FDT) is applied to study how the tropical forcing (i.e., Madden-Julian oscillation) and the stochastic extratropical internal modes trigger the large-scale extratropical teleconnections. link
Atmospheric Rivers in a warmer climate
Atmospheric river is characterized by strong moisture flux, which brings both water resource and disastrous flooding when it makes landfall. As climate warms, it is anticipated that the air can hold more moisture and lead to more extreme precipitation. It is therefore important to understand how the ARs will change in a warmer climate. In this research, we use large ensemble simulations generated by GFDL next generation climate model and identify when will humanity notice its influence on atmospheric rivers. In addtion, we also provide a more computationally efficient approach, where we derive a analytical solution to quantify the uncertainty in the foreced reponse caused by the internal climate variability(manuscript submitted)
Machine Learning and Climate Science
Machine Learning is a powerful tool which enables scientists to deal with non-Gaussian data. In climate research, non-Gaussian data exists almost everywhere including cumulus process or in any kind of nonlinear feedback. However, it has been dubbed as black box since its development. In our study, we demonstrate the potential of leveraging machine learning to the extended prediction and make it "physically more interpretable". link
Recent Publications (*=mentored students)
2023
Tseng K.-C., : Data-driven Statistical Mechanics for an Infinite-Member of Ensemble Weather Forecast (in preparation)
Tseng K.-C., and co-authors: Skillful forecasts of springtime CONUS tornado activity up to a year in advance (in preparation)
Schmitt, J.,(*) Tseng, K.-C, Hughes M., and Johnson, N., : Illuminating snow droughts: The future of Western United States snowpack in the SPEAR large ensemble. (submitted to JGR Atmosphere)
Tseng K.-C., and Y.-H., Ho(*): The Subseasoanl Predictability of North Pacific Subtropical High and the 2020 Record-breaking Event (submitted to npj, Climate and Atmospheric Sciences)
Bower, C., and coauthors: Atmospheric River Sequences as Indicators of Hydrologic Hazard in Present and Future Climates (submitted to Earth Future)
Jong, B.-T., and coauthors: Investigating Observed and Projected Increases in Extreme Precipitation over the Northeast United States using High-resolution Climate Model Simulations npj Clim Atmos Sci 6(18)
Zhang, W., Xiang, B., Tseng, K.-C, Johnson, N., Harris L., Delworth T.: Subseasonal prediction of wintertime atmospheric rivers in the GFDL SPEAR model. (submitted to Journal of Climate.)
2022
Tseng K.-C., and co-authors: When will humanity notice its influences on atmospheric rivers?127 e2021JD036044. link
Jia L.,and co-authors: Skillful seasonal prediction of North American summertime hot extremesJ. Clim., 35(13), 4331--4345 link (Nature Research highligh)
Bushuk B., and co-authors: Mechanisms of Regional Artic Sea Ice Predictability in Dynamical Seasonal Forecast Systems 35(13), 4207--4231 link
2021
Chen Y.-L.,and co-authors: Effect of the MJO on East Asian winter rainfall as revealed by a SVD analysis J. Clim., 34(25), 9729--9746 link
Zhang L.,and co-authors: Using large ensembles to elucidate the possible roles of Southern Ocean meridional overturning circulation in the Southern Ocean 36-yr SST trend (J. Clim, in press)
Tseng K.-C., and co-authors: Are multiseasonal forecasts of atmospheric rivers possible? Geophys. Res. Lett 48 e2021GL094000. link (NOAA-GFDL and Princeton CIMES Research highligh)
Bushuk B., and co-authors: Seasonal prediction and predictability of regional Antarctic sea ice J. Clim., 34(15), 6207--6233 link
Zhang G., and co-authors: Seasonal Predictability of Baroclinic Wave Activity npj Clim Atmos Sci 4(50) link
Tseng K.-C., N. C. Johnson., E. D. Maloney, E. A. Barnes, and S. B. Kapnick: Mapping Large-scale Climate Variability to Hydrological Extremes: An Application of the Linear Inverse Model to Subseasonal-to-Seasonal prediction J. Clim., 34(11), 4207--4225 link
Tseng K.-C., E. A. Barnes, and E. D. Maloney: The important role of the MJO for extratropical variability in observations and the CMIP5 climate models (submitted to JGR-Atmosphere)