Time series classification using features extraction to identification of land use and land cover: A case study in the municipality of Itaqui, South Region of Brazil
One of the main applications in remote sensing is the analysis and classification of land cover and land use. Sensors, such as MODIS, are have been largely used for monitoring land cover change due to its high temporal resolution. Although several studies perform time series classification by features extracting or similarities measures to verify annual usage patterns and land cover, a little has been explored about the extraction of information by focal neighborhood operation and different sub-intervals of a time series whole. In order to explore the use of different features extracted of annual time series, for sub-intervals and focal neighborhood to identify patterns of land use and land cover, this work takes the use of statistical measures already extracted in the context of annual time series and presents an approach to information extraction for sub-intervals of year and focal neighborhood to characterise temporal patterns. To demonstrate the applicability of this study an experiments were conducted to classify of time series of land use and land cover using EVI MODIS sensor data, using Random Forest algorithm, where resulted in the creation of temporal maps identifying temporal patterns.