Utilização de perfis multi-temporais do NDVI/AVHRR no acompanhamento da safra de soja no oeste do Parana / Use of NDVI/AVHRR time-series profiles for soybean crop monitoring in the west of Parana, Brazil

AUTOR(ES)
DATA DE PUBLICAÇÃO

2007

RESUMO

Crop yield forecasting systems are important sources of information for helping Governrnent and private institutions in agricultural trade matters. The traditional methods of crop yield forecasting norma1ly do not provide precise estimations, due to its subjectivity. The use of Geothecnology has been studied for the development of more efficient crop yield forecasting systems. In this study, NDVI time-series, obtained ITom A VHRR/NOAA imagery, were used for the soybean (Glycine max (L.) Merril) crop monitoring in the western region ofthe Paraná State, Brazil, in the 2002/2003, 2003/2004 and 2004/2005 cropping seasons. Landsat images were processed to identify the soybean crop lands. Temporal profiles, describing the biomass status throughout the phenological cyc1e, were generated for 18 municipal districts. Automatic systems for image processing and spectral information extracting were developed to speed up the study. Quantitative parameters were measured ITom the temporal profiles, considering the whole cycle and parts of it. Linear regressions between the quantitative parameters and the municipal mean yields have shown that, in the 2002/2003 and 2003/2004 cropping seasons, the most significant correlations occurred when the whole cyc1e was considered. When smaller periods of the cyc1e were considered, prior to harvest, the correlations lowered. The NDVI monitoring during these two cropping years, which presented different c1imatic conditions, allowed the explanation of a greater part of the soybean yield variability at the municipal leveI. Inverse correlations were obtained in the 2004/2005 season, with more significant coefficients two months prior to harvest. The results have shown the potential of the multitemporal analysis using NDVI profiles as a source of data to be added to agrometeorological-spectral models for the soybean yield forecasting. The inftastructure presented in this study allowed a higher leveI of automation in the multitemporal analysis, mainly in the precise NOAA image navigation, which did not demand human intervention

ASSUNTO(S)

geoprocessing sistema de informação geografica sensoriamento remoto soja remote sensing geometric correction agricultura - estimativa de rendimento crop yield forescating processamento de imagens image processing multitemporal analysis

ACESSO AO ARTIGO

Documentos Relacionados