Recently, the environment, sustainable development and natural risk management have become a major challenge. New Caledonia probably represents a first rank environmental observatory.  Thus, the bulk of the great barrier reef of New Caledonia has been declared a World Heritage by the UNESCO. Protecting and managing such a fragile environment owing to major mining projects and an increasing population are major challenges. Human activities enhanced erosion of naturally fragile areas (thick tropical regoliths), resulting in dangerous processes for persons, goods and resources. On another hand, an increasing landslide frequency in the southern part of French Alps has been correlated with climate change and geodynamic forcing (earthquakes). Additionally, wrong land use has reinforced the process. These examples underline the need for a global approach of environmental managing, especially as far erosion is concerned.`

 

These last years, the increasing amount of geosciences data have lead to new promising applications. For example, the use of very high resolution satellite images now enables the observation of small objects. However, actual data analysis techniques suffer from the huge amount of complex data to process. Indeed, this environmental data is often heterogeneous, multi-scale, incomplete, and composed of complex objects.

 

In this context, this computer science project aims at providing to geologists a semi-automatic and complete process for monitoring soil erosion. This process will be based on multi-temporal very high resolutions satellite images coupled with digital elevation model (DEM), sensor data and/or expert knowledge. The project will focus more precisely on two important aspects of this process: segmentation of satellite images based on collaborative methods, and construction of descriptive (patterns, clustering,  …) and predictive (decision trees, …) spatio-temporal models. New methods, algorithms and softwares on environmental data will be proposed to assist experts in their knowledge discovery.

 

 

This project is composed of academic laboratories and one company. This multidisciplinary consortium is composed of

 

  1. computer scientists of the LIRIS, LSIIT, LISTIC and PPME laboratories, specialists in data mining and image processing,
  2. geologists of the PPME and IPGS laboratories, with a strong knowledge in soil erosion mechanisms and
  3. the Bluecham Company, specialist in decision support systems.

 

Skills of each partner cover most of the KDD (Knowledge Discovery in Databases) process, from selection and preprocessing to geologic knowledge discovery. Moreover, our industrial partner will transfer our works to the society.