ASEN 6519

Data Assimilation and Inverse Methods
for Earth and Geospace Observations


Fall 2018 | T/TH 2:00-3:15pm | ECCR TBD

Course Description

Data assimilation and inverse methods play a key role in integrating remote-sensing and in-situ Earth and Geospace observations into a model of the Earth system or subsystems, enabling weather prediction and climate projection of high societal relevance. This course covers selected topics in probability theory, spatial statistics, estimation theory, numeric optimization, and geophysical nonlinear dynamics that form the foundation of commonly used data assimilation and inverse methods in the Earth and Space Sciences. The course materials are organized into three sections: (1) Statistical Principles and Background, (2) Building Blocks for Spatial Problems, and (3) Building Blocks for Spatial-Temporal Problems. Hands-on computational homework (in Matlab) and projects provide opportunities to apply classroom curricula to realistic examples in the context of data assimilation.

Class Learning Goals

The goals of this course are to provide the fundamental statistical background and context of commonly used data assimilation and inverse methods in the Earth and Space Sciences, and to equip students with the knowledge and skills to construct a data assimilation system on their own. Students will: (1) develop a deeper understanding of the underlying statistical principles of data assimilation and inverse methods; (2) actively apply their own understanding of the fundamentals and tradeoffs of different approaches in critiquing current data assimilation research; and (3) develop the skills, confidence and creativity to design and develop a data assimilation system of their choice.

Prerequisites

Some basic understanding of random vectors and matrices, estimation theory, numerical optimization techniques (e.g., ASEN 5044 Statistical Estimation for Dynamical Systems) as well as programming experience with Matlab are desired.

For more information:

ASEN 6337

Remote Sensing Data Analysis


Fall 2017 | T/TH 1:45-2:45pm | ECCR 151 (Lab: ECCR 239)

Course Description

With an explosive increase in the availability of high-resolution terrestrial remote sensing data, analyzing it has become a big data problem. Increasingly, machine learning is being recognized as a powerful tool for addressing this challenge. This course covers some of the most commonly used machine learning techniques in remote sensing data analysis, specifically for clustering, classification, feature extraction and dimensionality reduction. The course also covers inverse methods used to retrieve geophysical information from atmospheric remote sensing data. The course materials are organized into five sections: (1) Introduction, (2) Feature Extraction and Selection, (3) Clustering, (4) Classification, and (5) Inverse Methods for Atmospheric Remote Sensing Data. Hands-on computational homework (in Matlab or/and Python) and group and individual projects provide opportunities to apply classroom curricula to real remote sensing data.

Class Learning Goals

The goal of this course is to introduce commonly used machine learning techniques and inverse methods in remote sensing data analysis, equipping students with the knowledge and skills to apply modern data analysis techniques to remotely sensed data on their own. Students will: (1) develop a deeper understanding of machine learning and inverse methods in the context of remote sensing data analysis; (2) actively apply their own understanding of the fundamentals and tradeoffs of different approaches in critiquing current remote sensing data analysis research; and (3) develop the skills, confidence and creativity to design and solve a remote sensing data analysis problem of their choice.

Prerequisites

Some basic understanding of estimation theory and statistical learning techniques (e.g., ASEN 5044 Statistical Estimation for Dynamical Systems, ASEN 5307 Engineering Data Analysis Methods), as well as programming experience with Matlab or/and Python and familiarity with software engineering tools are desired.

For more information: