Materials and Methods

Site descriptions

The 2 DBs investigated during this current study, along with concurrent studies investigating the effect of DBs on hydrology and nutrients during the same storm events from 1 December 2017 to 30 November 2018, were located on pastoral dairy farms in the north-western portion of the Lake Rotorua catchment. Table 1 describes site characteristics reported by the concurrent studies. The Oropi series soils at the the Hauraki site, and Waiteti series soils at the Awahou site are both free draining, with >72 mm/h permeability in slowest horizon (Rijkse & Guinto, 2010).

Event types

Event types reported on in this, and concurrent studies, were differentiated according to the mode(s) ponded water was discharged from the DB. ‘Overflow Events’ occurred during larger runoff events when inflow continued to be delivered to the pond after the pond height exceeded the height of the upstand riser (Fig. 1 ). After 3 days of ponding, any residual ponded water was evacuated when the outlet valve was opened, creating ‘release discharge’. Therefore, ‘Overflow Events’ had both overflow and release discharge components. In contrast, ‘Non-overflow Events’ were smaller storms that did not contribute enough runoff to overtop the riser. Non-overflow Events included events when at the end of the 3-day treatment period, either had a portion of ponded runoff to discharge by opening the release valve, or all ponded runoff leaked and infiltrated the soil so there was no water left to discharge.

Equipment

Isco® (California, USA) 6712 portable auto-samplers, capable of filling 24 x 1 L bottles collected inflow and discharge samples at each site when triggered by a telemetered UNIDATA® Neon® 2013 F 3G External Memory Metering Module data loggers linked to UNIDATA®6527 Starflow® QSD flowmeters. The auto-samplers were triggered to collect 1 L samples when flows exceeded 7 L/s (Harmel, King, Wolfe, & Torbert, 2002). Calibration and maintenance of the monitoring equipment followed standard quality controls (NIWA, 2004).
Inflow auto-samplers collected a 1 L sample every 20 min for the first 10 samples, then one 1 L sample/h thereafter (Harmel, King, & Slade, 2003; Stanley, 1996). The mouth of a rain guarded 750-mL self-sealing bottle using a ping-pong ball inside the bottle, was installed at ground level near the pond outlet valve to capture a sample of the initial flush of surface runoff generated before the inflow auto-sampler was triggered. The ping-pong ball bottle sample was used as the concentration of the initial runoff and used in calculating event inflow yields
Discharge auto-samplers were programmed to collect a 1-L sample/h (Harmel et al., 2003; Stanley, 1996). Sampled discharge flows were generated if the pond height exceeded the upstand riser height during pond filling (i.e. ‘overflow discharge’) (Fig. 1 ), and when the valve at the base of the riser was opened to release the pond at the end of the event treatment, typically on the third day of ponding (i.e. ‘release discharge’).
Throughout all ponding at both sites an intractable leak at the connection point of the outlet valve pipe and the base of the upstand riser generated a continual flow of ~2-4 m3/h. Attempts at sealing this leak during the study period were unsuccessful. Under normal sampling conditions, the leak flow was too low to trigger the auto-samplers, although leak samples were collected during 4 events at the Hauraki site, and 1 event at the Awahou site, in order to characterise the SS concentrations of the leak discharge.
Water samples were collected from the field within 24 h of the end of the ponding event and kept refrigerated at 4 °C prior to subsampling (within ~24 h of collection). Two separate subsamples (~30 mL) were taken from the field sample after vigorously shaking the bottle, to analyse total and dissolved N and P. The remaining field sample was kept refrigerated until being analysed for SS concentrations used in this current study, following the standard procedure from the American Public Health Association (2005).

Calculations

Mean flow proportional concentrations

Event and annual mean flow-proportional (MFP) SS concentrations were calculated by dividing the inflow and discharge loads by their respective volume (Tanner & Sukias, 2011). The average difference between the event MFP inflow and leak samples collected during 5 events was +3%, with no consistent increase or decrease. Due to the negligible difference between the MFP inflow and leak concentrations, the MFP inflow concentration was applied to the entire leak volume for each respective event in which the leak discharge was not sampled. The applied leak concentration was used to calculate the event MFP discharge concentrations and event discharge loads. All inflow and discharge MFP concentrations will be referred to only as inflow and outflow concentrations.

Loads and yields calculation

Loads of SS in inflows and each discharge type were determined for all ponding events. Inflow loads of SS were calculated by multiplying the measured concentration of the runoff samples collected by the ping-pong ball sample bottle and auto-samplers, and using interpolated concentrations based on the linear rate of change between measured concentrations, by the interval flow volume measured every 5 minutes. Inflow loads were corrected on a pro rata basis (15% increase at the Hauraki site and 9% increase at the Awahou site) to account for the small catchment area between the inflow monitoring site and the DB (Table 1) .
Discharge loads were calculated for overflow discharge (combining upstand riser and spillway breaching), release discharge (which occurred during Overflow events and Non-Overflow events), and leak discharge. The load of each discharge type was calculated from flow measurements and sample concentrations taken from the DB outlet pipe, except for emergency spillway breaching. Emergency spillway loads were calculated by applying the MFP concentration of the overflow discharge generated by ponded water discharged by going over the upstand riser to the volume breaching the spillway calculated by the concurrent study reporting on hydrology. Yields refer to the load per unit of contributing catchment area and expressed as mm for runoff volumes, and kg ha-1 for SS loads.

Data analysis

Water sample data analysis

Events were grouped by site, then analysed for overall annual results, and results for each event type. Changes to concentrations were calculated as the percent difference between inflow and discharge concentrations. The percent difference between inflow and discharge yields were reported as ‘yield treatment efficiencies’. Inflow yield data for each site was also organised by austral seasons (i.e. summer from December to February) to compare differences between the sites and identify seasonal patterns for SS inflow yields.