Author(s): ALBINO, T. C.; VICTORIA, D. de C.; SILVA, G. B. S. da; BATISTELLA, M.

Summary: The increase in demand for food and agricultural products has led to an expansion of agricultural areas in Brazilian lands over the last decade, notably the region between the states of Maranhão, Piaui, Tocantins and Bahia. In order to evaluate and map such expansion, cropland mask from 2001 to 2007 were created based on remote sensing images. The production cycle of grain crops, with its rapid development and harvest, results in a strong seasonal signal that can be identified in time series of vegetation index images. Annual time series of the Enhanced Vegetation Index from MODIS sensor were analyzed and Fourier transforms were applied to each multitemporal image in order to isolate the annual and 6-month variation in the vegetation cover, typical of short cycle agricultural crops. Results were compared to official planted area statistics for the microregions encompassed in the study area. Both cropland masks and official statistics showed a strong increase in croplands. However, the cropland mask generated underestimated total planted area in 20%. These differences could be due to the account of double season crops in the planted area statistics or to a confusion caused by the Caatinga vegetation, which shows a strong seasonal vegetation cycle similar to the pattern observed in grain crops.

Publication year: 2013

Types of publication: Paper in annals and proceedings

Unit: Embrapa Territorial

Keywords: Fourier, MODIS, Séries temporais, Tim e series