Large-area maize yield forecasting using leaf area index based yield model

Large-area yield prediction early in the growing season is important in agricultural decision-making. This study derived maize (Zea mays L.) leaf area index (LAI) estimates from spectral data and used these estimates with a simple LAI-based yield model to forecast yield under irrigated conditions in large areas in Sinaloa, Mexico. Leaf area index was derived from satellite data with the use of an equation developed with LAI measurements from farmers' fields during the 2001-2002 autumn-winter growing season. These measurements were correlated with the normalized difference vegetation index values from 2002 Landsat ETM+ (enhanced thematic mapper) data. The equation was then tested with 2003 Landsat imagery data. A yield model was validated with maximum LAI and yield data measured in farmers' fields in northern and central Sinaloa during three consecutive autumn-winter growing seasons (1999-2000, 2000-2001, and 2001-2002). The yield model was further validated with 2002-2003 autumn-winter ground LAI (gLAI) and satellite-derived LAI (sLAI) data from 71 farmers' fields in northern and central Sinaloa. Grain yield was predicted with a mean error of -9.2% with maximum gLAI and -11.2% with sLAL Results indicate that the yield model using LAI can forecast yield in large areas in Sinaloa in the middle of the growing season with a mean absolute error of -1.2 Mg ha(-1). The use of sLAI in place of ground measurements increased the mean absolute error by 0.3 Mg ha(-1). Nevertheless, the use of sLAI would eliminate laborious LAI measurements for large-area yield prediction in Sinaloa.