Walter Scott, Jr. College of Engineering

Graduate Exam Abstract

Savini Samarasinghe
Ph.D. Preliminary
Dec 04, 2018, 3:00 pm - 5:00 pm
LSC 328
Causal inference using observational data - case studies in climate science
Abstract: We are in an era where atmospheric
science is data rich in both
observations (e.g., satellite/sensor
data) and model output. Our goal with
causal discovery is to apply suitable
machine learning methods to climate
data to identify potential cause-effect
relationships between climate
variables without using any
experimental controls. These methods
produce probabilistic graphical models
that encapsulate the potential
dependency structure of the variables
of interest. Extracting and studying
causal signals in climate can offer
deeper insights of the processes
governing the Earth’s climate as well
as of the effects of climate change.

Many different frameworks can be
used to infer causality. Graphical-
Granger methods, constraint based
structure-learning methods, Gaussian
graphical methods, and information
theoretic approaches are only a few
examples. With a plethora of
alternative procedures and frameworks
to infer causal relationships, it is very
likely that an investigator interested in
inferring causality in a spatiotemporal
setting comes across many questions
and challenges. What framework is
most suitable for the research
question? How do you set up the
inference problem in a physically
meaningful way? How do you select
the model variables? How should you
preprocess variables to extract the
strongest causal signals in the
presence of noise? What could you do
if the variables are non-stationary? –
are few such questions. An additional
limitation of most inference
frameworks is that they assume that
there are no hidden common causes
acting upon the variables included in
the model. This assumption is often
violated in practical applications
resulting in the standard methods
producing spurious/incorrect results.
Furthermore, it can be challenging to
identify suitable inference methods
that are scalable to high dimensional
problems, especially when the data is
non-Gaussian and the relationships
between the variables are non-linear.

With our research, we will attempt to
ease this burden imposed on a
researcher by investigating how
several of these issues can be
addressed. Specifically, we propose to
analyze a few causal inference
questions in climate science, case-by-
case, and document the scientific
thought process of setting up the
problem, the challenges faced, how
the challenges are dealt with, and - in
the process - also generate new
scientific findings of interest to the
climate science community. The main
objective of this proposed research is
to make causal inference methods
more accessible for a
researcher/climate scientist who is at
entry level to spatiotemporal causality.
As the case studies, we propose to
study (1) the causal relationships
between the Arctic temperature and
mid-latitude circulations, (2)
relationships between the Madden
Julian Oscillation (MJO) and the
European weather and (3) the causal
interactions of atmospheric
disturbances at different spatial scales
(e.g., Planetary vs. Synoptic). In
addition to these case studies, we also
propose to study the suitability of
existing inference methods to identify
causal relationships in the presence of
hidden common causes. We also
intend to create tutorials and sample
codes as a gateway point to causal
inference in spatiotemporal settings
with the aim of bridging the gap
between the causal inference domain
and the climate science domain.
Adviser: Dr. Imme Ebert-Uphoff
Co-Adviser: N/A
Non-ECE Member: Dr. Michael Kirby, MATH
Member 3: Dr. Edwin Chong, ECE
Addional Members: Dr. Chuck Anderson, CS
Peer reviewed journals:
1. Samarasinghe SM, McGraw MC, Barnes EA, Ebert‐Uphoff I. A study of links between the Arctic and the midlatitude jet stream using Granger and Pearl causality. Environmetrics. 2018; e2540.
2. Baker, A. H., Hammerling, D. M., Mickelson, S. A., Xu, H., Stolpe, M. B., Naveau, P., Sanderson, B., Ebert-Uphoff, I., Samarasinghe, S., De Simone, F., Carbone, F., Gencarelli, C. N., Dennis, J. M., Kay, J. E., and Lindstrom, P.: Evaluating lossy data compression on climate simulation data within a large ensemble, Geosci. Model Dev., 9, 4381-4403,, 2016.
3. S.M Samarasinghe, Y. Deng, and I. Ebert-Uphoff, A Causality-Based View of the Interaction between Synoptic- and Planetary-Scale Atmospheric Disturbances, J. Atmospheric Sci., under review.

Peer reviewed workshop (short paper) publications:
4. S. Samarasinghe, E. Barnes and I. Ebert-Uphoff, Causal discovery in the presence of latent variables for climate science, Proceedings of the Eighth International Workshop on Climate Informatics (CI 2018), 4 pages, Sept 2018 (accepted, in press).
5. J. Ramsey, K. Zhang, M. Glymour, R. Sanchez Romero, B. Huang, I. Ebert-Uphoff, S. Samarasinghe, E. Barnes and C. Glymour, A Toolbox for Causal Discovery, Proceedings of the Eighth International Workshop on Climate Informatics (CI 2018), 4 pages, Sept 2018 (accepted, in press).
6. S. Samarasinghe, M. McGraw, E.A. Barnes, I. Ebert-Uphoff, A Study of Causal Links Between the Arctic and the Midlatitude Jet-Streams, Proceedings of the Seventh International Workshop on Climate Informatics (CI 2017), NCAR Technical Note NCAR/TN-536+PROC, Sept 2017.
7. S. Samarasinghe, Y. Deng and I. Ebert-Uphoff, Structure Learning in Spectral Space with Applications in Climate Science, Workshop on Mining Big Data in Climate and Environment (MBDCE 2017), 17th SIAM International Conference on Data Mining (SDM 2017), April 27 - 29, 2017, Houston, Texas, USA, 5 pages, 2017.
Program of Study:
ECE 512
ECE 513
ECE 514
ECE 516
ECE 520
ECE 656
MATH 532