Lissage par ligne brisee pour l'interpolation de series de donnees incorporant une variable explicative avec des observations plus denses: application a l'humidite du sol et aux precipitations

Broken line smoothing is a simple technique for smoothing a broken line fit to observational data and provides a flexible means of interpolation. Here an extension of this technique is proposed, which can be utilized to perform various interpolation tasks, by incorporating, in an objective manner, an explanatory variable available at a considerably denser dataset than the initial main variable. The technique incorporates smoothing terms with adjustable weights, defined by means of the angles formed by the consecutive segments of two broken lines. The mathematical framework and details of the method as well as practical aspects of its application are presented and discussed. Also, examples using both synthesized and real-world (soil water dynamics and hydrological) data are presented to explore and illustrate the methodology. [GRAPHICS] Editor Z.W. Kundzewicz; Associate editor A. Carsteanu