Résumé
Hydrogeological tracer tests are primarily conducted with fluorescent tracers. Field fluorometers make it possible to monitor tracers at very low concentrations (<1 ppb) and at high frequency. However, changes in natural fluorescence at a site resulting from variations of dissolved and suspended inorganic and organic material may compromise the measurement of useful signals, thereby limiting the chances of identifying or quantifying the real tracer recovery. An elevated natural signal can mask small concentrations of the tracer while its variability can give the impression of a false recovery. This article shows how the use of a combination of several continuous measurements at different wavelengths allows a better extraction of the natural signal. Field multispectral fluorometers were installed at two Mediterranean karst outlets; both drain carbonate systems but have different environmental conditions. The fluorometers functioned over several hydrologic cycles, in periods affected or not by artificial tracers, making it possible to observe natural signal variations at these sites. The optical properties of this type of field fluorometer were used to calculate the spectral response of the different optics of the measuring probe. These responses, superimposed on three-dimensional excitation/emission matrices produced from laboratory fluorescence measurements, allowed an understanding of what the fluorometer sees under natural flow conditions. The result is an innovative method for correcting artificial tracer results. This type of correction makes it possible to fine-tune the effect of natural background variation on tracer recovery curves for a clear identification of the tracer presence and a more precise quantification of its recovery.