No abstract

No abstract

No abstract

The French Naval Oceanographic and Hydrographic Service (SHOM) operates two niultibeam echo sounders for hydrographic surveys. These syslems provide a full coverage of the sea bed with a fine resolution. The data sets contain sparse erroneous soundings which must be detected and removed. Cleaning methods in use are mainly manual. They ensure high quality data sets but they represent a considerable human workload.To reduce the cost of the data cleaning, SHOM is studying automatic algorithms to detect dubious soundings. This article presents three algorithms developed by 1 he (( Centre de geostatistique -Ecole des Mines de Paris>> and based on geostatistical techniques. The first algorithm tests the median of the depth increments between the selected salunding and its neighbours; the second algorithm defines ,an acceptance interval from the quantiles of the local distribution of the depth; the third one is a cross validation in which the kriging error is compared with the kriging standard deviation.These algorithms have been integrated in the data cleaning system developed at SHOM . The raw data set is first processed by the detection algorithms then the operator controls the detected data. The first tests on board show a dramatic reduction in the processing time with an increase in quality.

Domaining is very often a complex and time-consuming process in mining assessment. Apart from the delineation of envelopes, a significant number of parameters (lithology, alteration, grades) are to be combined in order to characterize domains or subdomains within the envelopes. This rapidly leads to a huge combinatorial problem. Hopefully the number of domains should be limited, while ensuring their connectivity as well as the stationarity of the variables within each domain. In order to achieve this, different methods for the spatial clustering of multivariate data are explored and compared. A particular emphasis is placed on the ways to modify existing procedures of clustering in non spatial settings to enforce the spatial connectivity of the resulting clusters. K-means, hierarchical methods and model based algorithms are reviewed. The methods are illustrated on a simple example and on mining data.