Relatorio DA Fase 1 Do Projeto Estima Desenvolvimento de Um Modelo Para Estimativa DA Produtividade Agricola DA Cana-de-Acucar Baseado Em Dados Landsat E Agrometeorologicos (Report on Phase 1 of the Project Estimate Development of

摘要

The objective was the development of a model for sugar cane (Saccarum officinarum) yield estimation in a test site corresponding to the sugar cane crop production area of an industrial plant, based on LANDSAT and Agromet data. Information on agricultural production system of the Barra Grande Plant located at the municipal area of Lencois Paulista, Sao Paulo State, for the crop year of 1983/84, 1984/85, and 1985/86 was obtained. Initially, the spectral response of the main varieties at different stages (number of harvesting cuts) was analyzed based on LANDSAT MSS data. The correlation between LANDSAT MSS and actual yield, estimated at the plant was investigated for the main planted varieties and development stages. Afterwards, a representative sample of the production system taking into account the different varieties and growth stages was selected for each crop year. Digital counts of the LANDSAT imagery analyzed were extracted using an image processing laboratory implemented at INPE. Vegetation indices based on the digital counts of the sugar cane selected fields were computed and correlated to the actual observed at the plant. The yield variation explained by the vegetation index was 41 percent. Estimated yield based on an Agromet model was correlated to the actual yield as well, and 57 percent of yield variation was explained by this model. When the vegetation index and the Agromet data were jointly correlated to the actual yield, 72 percent of the yield variation explanation resulted. This result is as accurate as the yield estimation done at the field level for the plant operation management purposes.