Characterisation of fluorescence background in dye tracing

Abstract

Tracing with fluorescent dyes is a core technique in karst hydrogeology. The primary criterion for evaluation of a dye trace is whether a tracer can be demonstrated to significantly exceed background concentrations. The critical level of significance is generally based on the judgement of the operator, rather than any objective protocol. Such an objective criterion might seem desirable, but the dependence of background on the tracer, the environment, place and time of operation makes it impossible to implement. True background fluorescence arises from overlapping spectra of arbitrary fluorophores, but contamination by the actual tracer material generally poses a more serious threat. Tracer may be derived from elsewhere, or be present because of mishandling by the operator. The purpose of this paper is to develop a statistically valid sample of background variation at a single site over a sustained period of time. A total of 168 hand samples were collected approximately daily from a gauging station on Medway Creek, Ontario (catchment area 167 km2, mean discharge 2.47 m3 s–1), a surface drainage system draining an agricultural/suburban catchment on glacial till. Samples were analyzed by synchronous scanning on a PTI QM-1 spectrofluorophotometer at of 20 and 90 nm, and slit settings of 2 nm. There was considerable variability in synchronous spectra at both Δλ, and little serial correlation in time. A statistical analysis suggested that select areas of the spectrum were prone to greater increases in background: ~380nm at Δλ=90 and ~300,~325 and ~510 nm at Δλ=20 nm. The shorter wavelengths do not interfere significantly with conventional fluorescent tracers. However, 510nm at Δλ=20 corresponds exactly with the fluorescence peak of uranine (sodium fluorescein, Acid Yellow 73), probably the most popular tracer dye. The distribution of fluorescence intensity by wavelength shows a number of regions of transient fluorescence. This can be best portrayed in terms of the skewness of the distribution. Projection of fluorescence spectra in time allows an assessment of temporal variations in fluorescence. Based on daily samples, there is a strong, but inconsistent relationship with stream discharge. Fluorescence arises from release, flushing, supply exhaustion, and dilution of natural and artificial compounds. The results reported here can only be associated with the particular environment, and are based on a relatively small sample. However, they are probably indicative of background behavior in any moderately intensive agricultural/suburban environment. Broad organic background appears to vary considerably in composition over time, making routine extrapolative correction risky. Most tracer dyes, especially uranine, can appear as acute contaminants. Background problems are most simply solved by using overwhelming quantities of dye, which may not be acceptable. More sophisticated injection encoding may be necessary.