Predicting the denitrification capacity of sandy aquifers from shorter-term incubation experiments and sediment properties

Abstract

Abstract. Knowledge about the spatial variability of denitrification rates and the lifetime of denitrification in nitrate-contaminated aquifers is crucial to predict the development of groundwater quality. Therefore, regression models were derived to estimate the measured cumulative denitrification of aquifer sediments after one year of incubation from initial denitrification rates and several sediment parameters, namely total sulphur, total organic carbon, extractable sulphate, extractable dissolved organic carbon, hot water soluble organic carbon and potassium permanganate labile organic carbon. For this purpose, we incubated aquifer material from two sandy Pleistocene aquifers in Northern Germany under anaerobic conditions in the laboratory using the 15N tracer technique. The measured amount of denitrification ranged from 0.19 to 56.2 mg N kg−1 yr−1. The laboratory incubations exhibited high differences between non-sulphidic and sulphidic aquifer material in both aquifers with respect to all investigated sediment parameters. Denitrification rates and the estimated lifetime of denitrification were higher in the sulphidic samples. For these samples, the cumulative denitrification measured during one year of incubation (Dcum(365)) exhibited distinct linear regressions with the stock of reduced compounds in the investigated aquifer samples. Dcum(365) was predictable from sediment variables within a range of uncertainty of 0.5 to 2 (calculated Dcum(365)/measured Dcum(365)) for aquifer material with a Dcum(365) > 20 mg N kg−1 yr−1. Predictions were poor for samples with lower Dcum(365), such as samples from the NO3− bearing groundwater zone, which includes the non-sulphidic samples, from the upper part of both aquifers where denitrification is not sufficient to protect groundwater from anthropogenic NO3− input. Calculation of Dcum(365) from initial denitrification rates was only successful for samples from the NO3−-bearing zone, whereas a lag-phase of denitrification in samples from deeper zones of NO3− free groundwater caused imprecise predictions. In our study, Dcum(365) of two sandy Pleistocene aquifers was predictable using a combination of short-term incubations and analysis of sediment parameters. Moreover, the protective lifetime of denitrification sufficient to remove NO3− from groundwater in the investigated aquifers is limited, which demonstrates the need to minimise anthropogenic NO3− input.