Deep Learning Based Mineral Dust Detection and Feature Selection

Thumbnail Image

Files

CT2019Team4.pdf (339.06 KB)

Links to Files

http://hpcf-files.umbc.edu/research/papers/CT2019Team4.pdf

Permanent Link

http://hdl.handle.net/11603/19264

Collections

UMBC Information Systems Department
UMBC Faculty Collection
UMBC Student Collection

Author/Creator

Author/Creator ORCID

Date

Type of Work

technical reports
Text

Department

Program

Citation of Original Publication

Ping Hou et al., Deep Learning Based Mineral Dust Detection and Feature Selection, http://hpcf-files.umbc.edu/research/papers/CT2019Team4.pdf

Rights

This item is likely protected under Title 17 of the U.S. Copyright Law. Unless on a Creative Commons license, for uses protected by Copyright Law, contact the copyright holder or the author.

Subjects

UMBC High Performance Computing Facility (HPCF)

Abstract

Dust storm affects human health and the environment. In this study, we develop deep learning models to identify dust from cloud and surface using MODIS observations and CALIPSO data. We also identified the best subset of channels for dust detection by a shuffling procedure and a genetic algorithm. Results show the important features determined by the two methods are very similar. And the genetic algorithm selected a subset of features that have a comparable performance with the model with all features, proving the effectiveness of the genetic algorithm. The chosen subset will reduce future data collection efforts for dust detection.