versión impresa ISSN 0034-737X
JUNGES, Amanda Heemann y FONTANA, Denise Cybis. Agrometeorological-spectral model to estimate wheat yield in the State of Rio Grande do Sul, Brazil. Rev. Ceres (Impr.) [online]. 2011, vol.58, n.1, pp. 09-16. ISSN 0034-737X. http://dx.doi.org/10.1590/S0034-737X2011000100002.
This study aimed to estimate the wheat yield within the Cotrijal Cooperative's operational area (northern Rio Grande do Sul), using spectral and meteorological variables. Yield data (1991-2006), monthly agrometeorological data (1991-2006) and spectral data (NDVI/MODIS, 2000-2006) were used in the analyses. The existence of a significant increase in grain yield due to the incorporation of new technologies was analyzed (technology trend).The choice of the independent variables was based on the analysis of the correlation between yield and spectral and meteorological data. For the multiple linear regression of grain yield estimation the following independent variables were used: rainfall (October), frost damage index (September), degree-days (accumulated from May to October) and NDVI (integrated from June to October). The multiple linear regressions showed satisfactory results with estimation errors below 10%, in most examined years. Accuracy, easy implementation and low cost of regressions pointed to the possibility of joint use of spectral and agrometeorological data in estimating wheat yield in the Cotrijal region. However, further studies are recommended to verify the results of the generated models when incorporating a longer series of spectral data.
Palabras llave : Normalized difference vegetation index; frost index; pluvial precipitation; degree-day.