Sugar cane harvested area monitoring using MODIS images / Monitoramento da área colhida de cana-de-açucar por meio de imagens do sensor MODIS

AUTOR(ES)
DATA DE PUBLICAÇÃO

2007

RESUMO

Brazil is among the world leaders in producing and exporting several agricultural products and is the largest producer and exporter of sugar and alcohol. The State of São Paulo is responsible for approximately 60% of the entire national production of cane, alcohol and sugar and for 70% of the exportations. In the current globalized market there is a great demand for reliable and objective information on the amount of raw material available to sugar and alcohol production, since this affects directly the quotation of theses products in stock exchange. In addition, information about the type of harvest (with burning or without burning raw cane) is relevant to the environmental. Remote sensing techniques associated to GIS technology have a great potential to monitor the sugarcane harvest activity and provide reliable and objective information not only on the amount of harvested area but also on the type of harvest. Considering that the sugarcane crop has several favorable characteristics to be identified, mapped and monitored through remote sensing satellite images it seems that free of charge MODIS images are a promising alternative to monitor the sugarcane harvest activity. This hypothesis is based on the high temporal resolution of the MODIS images which maximizes the chance of obtaining cloud free images. Therefore, the objective of this work was to develop an operational procedure using MODIS images in São Paulo State in order to provide objective information about the sugarcane harvest activity. To achieve this objective different image processing techniques were tested, i.e., linear spectral mixture model, image algebra and the use of vegetation index (NDVI) with the intention to point out advantages and disadvantages of each technique to estimate sugarcane harvested area. These techniques were applied over multitemporal compositions of NDVI (MOD13Q1), compositions developed to detect the type of harvest (MODCSH) and daily reflectance images. To validate the estimates medium spatial resolution images were used (Landsat-5 and CBERS-2) and field data provided by a sugar and alcohol Plant. Best results of area estimates were obtained with the subtraction between the MODCSH compositions which estimated 95.2% of the reference area and achieved best temporal precision (R=0.95; for the regression between dates of reference and harvested estimates). It was possible to estimate the type of sugarcane harvest applying linear spectral mixture model over the daily reflectance MODIS images. Finally, the presented methodology, with the use of MODIS images, allowed to monitoring the harvest activity of the sugarcane crop.

ASSUNTO(S)

modis Índice de vegetação da diferença normalizada remote sensing sugar cane modelo linear de mistura espectral modis são paulo (estado) cana-de-açucar são paulo (state) normalized diffe sensoriamento remoto

ACESSO AO ARTIGO

Documentos Relacionados