Multi-temporal remote sensing for yield prediction in sugarcane crops

Abstract

Sugarcane yield prediction is critical for in season crop management and decision making processes such as harvest scheduling, storage and milling, and forward selling. This presentation reports on a recently published method of predicting sugarcane yield in the Bundaberg region (Qld) using time series Landsat data. From the freely available Landsat archive, 98 cloud free (<40%) Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper (ETM+) images, acquired between November 15th to July 31st (2001-2015), were sourced for this study. The images were masked using the field boundary layer vector files of each year and the green normalized difference vegetation index (GNDVI), an indicator of crop vigour was calculated. An analysis of average GNDVI values from all sugarcane crops grown within the Bundaberg region over the 15 year period using a quadratic model identified the beginning of April as the peak growth stage and, therefore, the decisive time of image capture for a single satellite image based yield forecasting. The model derived maximum GNDVI was regressed against historical sugarcane yield data obtained from the mill. The coefficient of determination showed a significant relation between the predicted and actual sugarcane yield (t/ha) with R2 = 0.69 and RMSE 4.2 t/ha. Results showed that the model derived maximum GNDVI from Landsat imagery would be a feasible technique to predict sugarcane yield in Bundaberg region. This research, however, warrants further investigation to improve and develop accurate operational sugarcane yield prediction model across other domestic and global growing regions, as the influence of environmental conditions and cropping practices will likely vary the relationship between GNDVI and yield (t/ha).