Full text
| |
| Author(s): |
Alexandre Esteves Almeida
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Instituto de Computação |
| Defense date: | 2017-08-04 |
| Examining board members: |
Ricardo da Silva Torres;
Bernardo Monteiro Flores;
Fábio Luiz Usberti
|
| Advisor: | Ricardo da Silva Torres |
| Abstract | |
Detecting and characterizing temporal changes are crucial indicators in the process of understanding how complex mechanisms work and evolve. The use of remote sensing images and techniques has been broadly employed over the past decades in order to detect and investigate temporal changes on the Earth surface. Such change detection in time series data may be even further refined by isolating the additive long-term (trend) and cyclical (seasonal) components from the underlying noise. This work investigates the particular Breaks For Additive Season and Trend (BFAST) method for the analysis, decomposition, and breakpoint detection of time series associated with remote sensing data. The derived outputs from that method are, then, used in three distinct ¿ but highly interconnected ¿ research venues: in a better comprehension of climatic phenomena; in the correlation to human-induced disturbances data; and in data classification problems using time series dissimilarity functions discovered by a Genetic-Programming-(GP)-based evolutionary framework. Performed experiments show that decomposition and breakpoints produced insightful and effective results when applied to the ecological data studies, but could not further improve the classification results when compared to its raw time series counterpart. The achievements in those three contexts also led to the creation of two open-source web-based time series analysis tools. One of those tools was so well received by the target community, that it is currently integrated into a private cloud computing platform (AU) | |
| FAPESP's process: | 15/02105-0 - Identifying temporal changes in tropical South American vegetation: a breakpoint detection approach |
| Grantee: | Alexandre Esteves Almeida |
| Support Opportunities: | Scholarships in Brazil - Master |