Análise da relação entre os perfis de NDVI obtidos dos sensores AVHRR/NOAA e MODIS nas áreas produtoras de cana-de-açúcar em São Paulo

The production of cane sugar in Brazil, in general, has been growing in recent years and has increasingly strategic role in the economy. Thus, it is important to propose innovative and technologically feasible solutions to help the generation of more efficient, objectives, precise, early and adequate models to monitoring and forecasting of the national crop. This study has the main objective of compare NDVI data from AVHRR/NOAA and MODIS/TERRA in productive sugarcane areas in São Paulo, from April 2001 to March 2010. Initially, were made the selection of sugarcane planting points in São Paulo and the extraction of NDVI profiles from AVHRR/NOAA and MODIS/TERRA. After that, clustering analysis using K-Medoids and DTW were made in 5 clusters with NDVI profile and, finally, the statistical analysis (Pearson correlation) done with clusters of the planted area, harvested area and crop yield. The clusters identified pixels related to the sugarcane planting areas and spectral mixture. The MODIS/TERRA showed values of R greater than the AVHRR/NOAA, because the sensor has a spatial resolution of 250m, with more details of sugarcane areas. However, the AVHRR/NOAA also obtained satisfactory results. This shows that it is possible to use data from low spatial resolution sensors to monitor and follow harvests of crops planted over large areas. The advantage of these sensor systems is their high temporal resolution, low cost and global coverage.Pages: 640-64