Themes: Geological Sciences, Earth, Universe
location: Research Laboratory on the Future of the Radioactive Sites Pollution (LELI) - Fontenay-aux-Roses (92)
Master or engineer degree in environmental sciences, advanced notions in geostatistics are recommended.
Age limit: 26 years old unless otherwise stated.
In situ gamma-ray spectrometry is widely used for the characterization of sites contaminated by radionuclides, regardless of the extent of the contamination (from the building to the territory) or its origin (natural or artificial). This kind of measurements guides the sampling strategy of soil collected for laboratory analysis and supports the interpretation of the results, in particular the mapping of probability of exceeding critical thresholds. Indeed, in situ measurements can be set with a very high frequency to collect a high data density and on different carriers (e.g., backpack, car, helicopter) to explore various spatial scales. Spatial data analysis tools, in particular from geostatistics, are commonly used to analyze in situ measurement: estimation methods can be used to reconstruct the measurements between the acquisition lines and to quantify the corresponding uncertainty (e.g., by kriging); and then to interpolate the activities deduced from analyses on soil samples (e.g., by co-kriging). However, the application of these geostatistical techniques requires to make assumptions on the degree of stationarity of the spatial properties of the measured variable on the domain of interest. In the case of punctual source of pollution, the contamination is most often organized in space in the form of plumes and strong contrasts can be observed in measurements at relatively short distances. The assumption of strict stationarity is thus generally not verified. In addition, in situ measurements are related to a detection field with a variable geometry, depending on the acquisition conditions or the physical properties of the investigated area. The link between in situ measurements and analyses on soil samples, related to a well-constrained support, can therefore vary locally. The integration of these two types of data is thus a non-stationary problem. The thesis will focus on the non-stationarity of radiological contamination, more precisely on its characterization and on the ways to improve both the acquisition strategy and the geostatistical processing of in situ measurements. The work will be based on a main data set, acquired following the Fukushima nuclear accident, with a density and quality that allow to address several key questions. The resulting methods and tools will be applied to other sites related to specific applied issues.