Skip to main content
— Presentation at SPIE REMOTE SENSING 2016 —
October 31, 2016
Fig. 1 Crop yield prediction method
SPIE remote sensing 2016 international conference was held in Edinburgh, UK, from Sep. 25 to 29, 2016. As an important symposium for remote sensing technology, SPIE remote sensing has been held for engineers, researchers, and scientists all over the world. There were 11 main sessions included in the conference.
Hitachi has been involved in the collaboration research with Campinas University of Brazil for estimation of sugarcane sucrose content (Fig. 1). As a report of research, I made an oral presentation titled "Dynamics modeling for sugarcane sucrose estimation using time series satellite imagery" in the session of "Remote Sensing for Agriculture, Ecosystems, and Hydrology".
Ethanol production has been paid attention as one of the most important agricultural derivatives, and there is increasing demand for quantity and quality incensement. Sugarcane, as one of the most mainstay crop in Brazil, plays an essential role in ethanol production. To monitor sugarcane crop growth, remote sensing technology plays an essential role in particularly for large-scale farming. In remote sensing analysis, high-resolution satellite image has high cost and low data availability while low-resolution image has high availability and low analysis precision. In this research, we focus on the integration of high and low resolution images to monitor and estimate sugarcane sucrose content.
Fig. 2 Farming information of production support cycle
In conventional method of estimation method, spectral features are extracted from satellite imagery while vegetation indices are calculated. Utilizing this data, statistical regression is applied to estimate the sucrose. However, sugarcane sucrose estimation using satellite images is considered difficult as the little correlation with sucrose content and main vegetation index such as Normalized Difference Vegetation Index (NDVI). We focused on the issues of sugarcane sucrose content estimation using time-series satellite image. For each growth stages respectively, sugarcane sucrose growth model is developed based on the understanding of biometry analysis. We calculated the spectral features and vegetation indices to make them be correspondence to the sucrose accumulation biological growth model. Nevertheless, we improved the statistical regression mix model considering more other factors.
The evaluation was performed through the sugarcane sucrose estimation using our proposed vegetation features in the sucrose accumulation cycle. We got precision of 90% which is about 9.6% higher than the conventional method. The validation results showed that our method could be applied in practice of large-scale sugarcane monitoring. As future works, it will be an important subject to make the prediction more robust and accurate. Moreover, with more data we are planning to contribute on the whole agriculture solutions (Fig. 2).
(By ZHAO yu)