Prediction sugarcane yields using remotely sensed images

Organised by Laboratory of Geo-information Science and Remote Sensing

Tue 29 August 2017 09:30 to 10:00

Venue Gaia, gebouwnummer 101
Room 1

By Jos Goris (The Netherlands)

Sugarcane is one of the most important crops in Rwanda. Demand on the national market is claimed to rise, while domestic production only covers 30% of the demand. Precision agriculture can help increase sugarcane yields. Monitoring of crop status could lead to a more realistic prediction of the yield than current yield predictions. In this research, remotely sensed imagery aimed to provide information about crop status. This information was used in a yield prediction model. Multiple models were evaluated and the most suitable model was chosen based on the requirements and available data. Green chlorophyll index data were calculated from the green and near-infrared light reflectance band and used as an indicator of crop status. The bulk harvest of sugarcane stalks was deemed linearly related to the green chlorophyll index times the incident photosynthetically active radiation. Historic data about yield per hectare per zone were used to calibrate the regression model. As a result of poorly calibrated remotely sensed imagery and lack of data covering the full growth cycle, the predicted yields are not accurate enough for Kabuye Sugar Work to be useful. In future endeavours, crop status over the whole growth cycle and correctly calibrated remotely sensed imagery are necessary to predict sugarcane yield more accurately.

Keyword: Yield prediction; Rwanda; sugarcane; remote sensing; vegetation index; green chlorophyll index; Kabuye sugar works