Wheat Yield Monitoring in Southern Spain Using the GRAMI Model and a Series of Satellite Images.

Recent studies show that the global demand for food will increase for at least another 40 years. It is estimated that the global population will reach 9 billion people by the middle of this century. Nowadays, more than one in seven people do not have access to recommended levels of protein and energy. This increase in demand and the limited resources available to produce food make it necessary to develop tools that allow estimations of crop production, thereby helping to manage the way food is produced, stored and distributed. The aim of this study was to develop a methodology to estimate wheat yield at the field scale using remote sensing data. Information from meteorological stations and TM and ETM+ sensors onboard LANDSAT 5 and 7, respectively, were used during the 2008 and 2009 growing seasons. The GRAMI model was adapted to be applied over a semiarid area in Southern Spain (Genil-Cabra Irrigation Scheme). The model parameters, light-use efficiency, crop phenological stage and yield partitioning factor, were calibrated using information collected from 30 durum and bread wheat experimental plots. According to the phenology data, the varieties were divided into two groups with similar development. Spectral radiometry measurements were taken over every plot throughout the growing season to obtain experimental relationships between the normalized difference vegetation index (NDVI) and leaf area index (LAI). Forty-nine wheat fields, each large enough to be resolved by the satellite imaging sensor while avoiding edge effects, were chosen each year in order to evaluate the model. Yield data for each field were supplied by farmers. The light-use efficiency (ε) used was 2.5 g/MJ PAR along with a 0.8 yield partitioning factor.Yield estimation errors of 12.76 and 12.80 % were obtained in each experimental group with a root mean squared difference (RMSD) of 531.96 and 514.89 kg/ha, respectively. When the model was applied over the irrigation district at the field scale with the seasonal development of LAI estimated from the series of satellite images, yield estimation errors of 25.15 and 22.93 % were obtained, with RMSD = 1051 and 1077 kg/ha for the 2008 and 2009 seasons, respectively. The significantly better results achieved under experimental plot conditions can be explained by the similar management practices applied to all these fields. By contrast, changes in crop management (planting date, irrigation, fertilization, etc) at the field scale reduced the model’s estimation ability. However, even at this scale, the results of the model were satisfactory for crop yield estimation over large areas, without the need for annual calibration of the model.