MODIS sensor images associated with an agronomic model to estimate soybean grain yield

The objective of this work was to estimate soybean yield in Rio Grande do Sul State, Brazil, for crop years of 2000/2001 to 2002/2003 through an agronomic crop yield model implemented in a geographic information system (GIS). Two approaches were used: an agronomic model (AGRO), with leaf area index (LAI) obtained from literature, and an agronomic-spectral model (AGROESPEC), with LAI estimated from MODIS (moderate resolution imaging spectroradiometer) images. Results were compared with the official estimates provided by Instituto Brasileiro de Geografia e Estatistica (IBGE), using the t test for paired observations. For crop years 2000/2001 and 2001/2002, there were no significant differences. For 2002/2003, AGRO and AGROESPEC approaches underestimated the yield in 7.87 and 7.04%, respectively, in relation to the official data provided by lBGE. Both approaches allowed a practical and quantitative evaluation of weather effects on soybean yield. However, the AGROESPEC provided more detailed estimates, in terms of soybean yield spatial variation, due to the use of LAI values from MODIS images.