Algorithmic identification of limnological features in vertical profiles from the Great Lakes
Abstract
High volume collection of environmental data in digital format presents a range of challenges for the researcher, from quality control and data management to efficient interpretation of the signal and the development of requisite information technology skills. These challenges have been termed the "data deluge". To aid in efficient data interpretation, we describe several algorithmic approaches for feature identification in signal streams, including gradient estimation, spectral analysis, and the hidden Markov model. These approaches are calibrated and evaluated over vertical temperature profiles from the Great Lakes obtained through the U.S. Environmental Protection Agency. To demonstrate the value of this data science approach, we describe how the algorithms can be integrated with the historical sampling record to yield an expert system that assists field technicians with adaptive sampling.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2013
- Bibcode:
- 2013AGUFMIN43A1647W
- Keywords:
-
- 1906 INFORMATICS Computational models;
- algorithms;
- 1916 INFORMATICS Data and information discovery;
- 1942 INFORMATICS Machine learning;
- 1976 INFORMATICS Software tools and services