Phosphate and ammonium distribution in a pilot-scale constructed wetland with horizontal subsurface flow using shale as a substrate

Phosphate: (P) and ammonium (N) distributions were investigated in a pilot-scale constructed wetland system (CWS) with horizontal flow. Shale was selected as a substrate, on the basis of its P adsorption capacity as well as its suitability for plant growth, investigated in an earlier study. The system was set up in a greenhouse, and comprised of four tanks with and four tanks without Phragmites australis (common reed). Sampling ports were installed along the length of each tank, and at three different depths, in order to facilitate three-dimensional monitoring of nutrient distribution over the period of study (il months). In the planted systems, H2PO4--P (ortho-P) and NH4+-N concentrations were low (0.5-1.0 g m(-3)) at all depths throughout their length. Generally, ortho-P and NH4+-N concentrations decreased exponentially along the transect from the tank inlet to the outlet. In the unplanted systems, higher values (0.5-16.5 g m(-3)) of ortho-P were observed. NH4+-N in the unplanted systems was relatively high (10-30 g m(-3)) throughout the period of investigation. In both planted and unplanted tanks, NO3--N concentrations were very low at the inlets (0.02-0.05 g m(-3)), and increased only slightly towards the outlets. Although the presence of P. australis had a significant effect (p < 0.05) on P and N concentrations at all depths and along the length of the tanks, the nutrient distribution followed the same trend as in unplanted tanks. The results were consistent with the theoretical removal model which predicts an exponential decrease in pollutant concentrations to a background value approaching zero along the transect from wetland inlet to the outlet. A modification of the model was suggested where a first order area-based reduction rate constant k (m d(-1)) would be replaced with (k(p) + k(s)), k(p) representing the rate constant for the removal by the plants, and k(s) the rate constant for the removal by the substrate, and associated microbial populations they support. Approximate values for k(p) + k(s) Of 0.084 m d(-1) (P) and 0.065 m d(-1) (N) and for k(s) of 0.069 m d(-1) (P) and 0.034 m d(-1) (N) were obtained. The hydraulic residence time, now characteristics and permeability of the shale was investigated by a bromide (Br-) tracer. The tracer breakthrough curves showed a similar pattern in all tanks, with about 66% of the now occurring through the bottom zone (0.35 m). The actual hydraulic residence time (6 days) was slightly higher than the nominal one. The tracer study also enabled the calculation of the volumetric rate constants, k(v) and approximate values of 0.075 d(-1) (P) and 0.060 d(-1)(N) were obtained for planted and 0.061 d(-1)(P) and 0.020 d(-1)(N) for unplanted tanks. The results obtained from this pilot scale study, provide a better understanding of nutrient distribution and flow patterns that take place within CWS, and should enable more rational design parameters to be developed that help to optimise future large-scale systems. (C) 2000 Elsevier Science Ltd. All rights reserved.