Abstract
In this paper, we propose a knowledge discovery process for hydrologic data. We adopt a sequential pattern algorithm applied on data from stations located along several rivers. Data are pre-treated according to different spatial proximities and the analysis of the number of patterns obtained highlights the influence of such defined relations. We propose an objective measure of validation, called the measure of temporal contradictionless, to help the expert to discover usual information. With this work, we move towards a new way of spatial indicator discovery to help interpretation of ecological monitoring of watercourses and pressure data.