May 19, 2009 An automatic analysis method for real-time monitoring of sugar-cane harvesting has been developed in Montpellier in the framework of a thesis co-directed by Cemagref and Cirad. The goal is to design a true decision-aid tool based on the expert knowledge available in the sugar-cane industry, that can also be adapted to other fields such as wine growing and forestry.
On the island of Réunion, sugar cane, a tall grass, covers over 25 400 hectares. The harvest period can span six months of the year. However, once the cane has been cut, it must be processed in a factory within 48 hours, otherwise decomposition hinders the industrial crystallisation process, among others. Because the factories operate with zero stock and at constant output levels, the companies increasingly use data supplied by satellite-image experts to estimate the progress of harvests on the island.
Question of logic
The goal of the SUCRETTE programme managed by CIRAD is to develop remote-sensing methods and products to meet the needs of the sugar-cane industry. With that in mind, Mahmoud El Hajj developed an automatic analysis method for satellite-image time series implementing the expert knowledge. He used the FisPro software, developed in 2000 by Cemagref and INRA, that can be used to build computer systems based on fuzzy logic. Whereas binary logic uses only 0 and 1, fuzzy logic is capable of grasping the nuances between the two and thus achieve great flexibility. In addition, by assigning linguistic labels to the different numerical values, fuzzy logic can produce rules similar to those used in human thought. As a result, it is possible to create true decision-aid tools whose design is based on human knowledge.
Models to fill in missing links
Another original aspect of the work carried out by the young researcher is a crop-growth model integrating biophysical, climate and agronomic parameters. It is thus possible to estimate the growth of the sugar cane in spite of any missing data in the satellite images. Large cumulonimbus clouds that mask large stretches of terrain and reduce the value of satellite images are no longer a problem. Because sugar-cane growth is rapid, the model is also designed to determine the probability of harvesting having taken place between two images taken at an interval of two or three months. This is useful because, seen from space, a fully grown field ready to be harvested is indistinguishable from a field with new growth, two months after the harvest.
Toward an operational tool
By combining data from satellite-image time series, crop-growth models and expert knowledge, this new method will, in time, make it possible to design a true decision-aid tool. Whereas standard image analysis requires several days of work by an expert, the new tool needs just a few hours to process all sugar-cane fields throughout Réunion. The only element missing today is a graphical interface to facilitate use by an engineering firm.
In the future, it will be possible to expand the method for real-time monitoring of dynamic processes in all types of agricultural and forestry land cover.
Focus on FisPro
The FisPro software serves to build fuzzy-inference systems and then use them to process data, in particular for simulation of physical or biological systems.
The software was developed by Cemagref and INRA for a technology-transfer project funded by the Languedoc-Roussillon region in France, with an industrial partner, the "La Malepère" wine company in Arzens (Aude department, France).
FisPro is available at no cost and may be freely downloaded from http://www.inra.fr/Internet/Departements/MIA/M/fispro/indexfr.html
Other social bookmarking and sharing tools:
Note: Materials may be edited for content and length. For further information, please contact the source cited above.
Note: If no author is given, the source is cited instead.