Optimizing flight-line distance for soil decontamination, application to contaminated Fukushima territories
Résumé
Geostatistical interpolation methods (block and point ordinary kriging) were used to produce soil contamination maps based on airborne gamma-ray measurements of cesium-137 deposit acquired in November 2013 with a flight-line distance of <1 km. The methods were applied to six different square tiles of ~20 km length located within 100 km from the Fukushima nuclear power plant to quantify the influence of an increase in the flight-line distance on the quality of the interpolated maps; indeed, smaller the flight-line distance, higher the quality and cost of produced maps. Therefore, kriging estimators were applied over different selections of flight-lines of decreasing density. The quality of the produced maps was quantified by the means of a correctness rate, calculated by comparing the predicted contamination surfaces, above some threshold, to that best estimated from all the flight-lines. The results showed that the calculated correctness was highly correlated with the flight-line distance (r2=0.83): increasing the flight-line distance for every 1 km decreased the correctness of the contamination map by ~1%. In fact, the correctness also depends on variogram modeling which is a subjective step of the kriging methods. The influence of variography was also analyzed: the contamination maps were reproduced, this time using a fixed variogram model (that of all the flight-lines) to find out how much flight-line distance influences the correctness, while the variogram model is unchanged. It was found that correctness is influenced from the variogram model by less than 6%, meanwhile at least 94% being attributed to the flight-line distance. Hence the relation of correctness versus flight-line distance is reliable.
Domaines
Sciences de l'environnement
Origine : Fichiers produits par l'(les) auteur(s)