Abstract: Weather radar systems are vital to ensuring the safety of society by providing timely, accurate
data products used to forecast the behavior and development of weather phenomena. In order
to do this, high-quality weather radar data, in the form of high spatiotemporal resolution data,
is necessary. Current weather radar scan strategies require a slow scan rate in order to collect
high-resolution polarimetric observation data. This thesis proposes the use of a deep learning,
super-resolution model in order to augment the current weather radar operational paradigm and
collect high-resolution weather radar data with faster scan rates. Specifically, this thesis focuses
on evaluating the performance of the super-resolution generative adversarial network in generating
physically-realistic, pseudo-high-resolution radar scans – referred to as super-resolution scans – of
the same quality of high-resolution radar scans from input low-resolution weather radar scans.
Super-resolution generative adversarial networks, since their inception, have proven to be useful and effective techniques, specifically for processing and enhancing the resolution quality of
video and image datasets. In doing so, both the quality and quantity of information procured from
that data are also enhanced. This paper aims to utilize the image processing and generating capabilities of super-resolution generative adversarial networks within the scope of generating superresolution weather radar scans from low-resolution weather radar scans. In order to accomplish this
task, multiple experiments are setup to test the model’s capabilities in conducting super-resolution
for different radar scan types, at different resolution scaling factors and for different downsampling
methods, one of which simulates the characteristics of actual LR weather radar scans. The results
of the SRGAN’s performance on generating super-resolved weather radar scans will be quantified,
comparing between the super-resolution scans and the baseline methods.

Adviser: Dr. Chandra
Co-Adviser:
Non-ECE Member: Ketul Popat, Mechanical Engineering
Member 3: Margaret Cheney, ECE
Addional Members:

Publications:

Program of Study:
BIOM 570
BMS 500
ECE 512
ECE 514
ECE 521
ECE 537
ECE 580B4
ECE 581B2