Pathogen Pathways – Riparian Management II
2. Delivery of faecal contamination to waterways by surface runoff – large scale rainfall simulator experiments
2.1 Introduction
To improve understanding of the delivery of microbes to waterways by surface runoff, five experiments were conducted under Objective 2 of the PTRRP using a large-scale rainfall simulator located upon hill-country pasture. The experiments aimed to quantify microbial delivery under heavy rainfall and examine the variation of delivery with stock history. Objective 5 has continued this series of experiments with a further six simulator runs being conducted. Together these 11 events (Table 1) encompassed two simulated rainfall rates and variable antecedent soil moisture conditions. Additionally, they have been conducted between zero (stock still present) and 75 days after grazing by sheep, in both summer and winter. The analysis reported here includes all 11 simulated events.
2.2 Methodology and site description
The large-scale rainfall simulator (Photograph 1) encompasses 1050 m2 of a topographically-convergent hillslope in the Pukemanga catchment, within the Whatawhata Research Station, west of Hamilton. The simulator consists of 13 rainfall stands at 9 m spacing. Each stand consists of a 5 m high pole with four closely spaced upward-spraying sprinklers mounted on a frame. The design rainfall rate is 35 mm/hr, providing a total application rate of 10.2 L/s, applied over approximately one hour during each experiment. This application rate and duration corresponds approximately to an 8-year recurrence interval. By removing two sprinklers from each stand it was possible to halve the design rainfall rate, and, during objective 5, two experiments were conducted at this lower intensity. On occasion the simulator was run one day prior to an event in order to wet up the soil.
Mean slope angle at the site is 18°, and the hillside is underlain by a yellow-brown earth soil (Kaawa hill soil). Vegetation is predominantly ryegrass-clover. No artificial borders were used around the area encompassed by the simulator since surface runoff converged and flowed naturally into a headwater stream. A 10 m wide wing-wall dug 50 mm into the ground was used to guide the flow to a flume (Photograph 2) at the outlet of the simulator area. Flow rate was measured using a 25 L tipping bucket. Samples of the outflow were collected manually throughout an experiment and taken immediately to the laboratory for microbial (E. coli) analysis.
Photograph 1: The rainfall simulator and experimental site
Photograph 2: The flume and catchment outlet
Table 1: A description of the “11 rainfall simulator events”.
| Event | Load | Conc. | Days | Flow | 3d-rain | Graze | Stock |
|---|---|---|---|---|---|---|---|
| 1 | 9 × 108 | 6 × 103 | 42 | 14.1 | 15.8 | 4 | 49 |
| 2 | 4 × 1011 | 3 × 106 | 0 | 13.7 | 70.2 | 7 | 130 |
| 3 | 2 × 109 | 1 × 104 | 55 | 12.9 | 34.3 | 3 | 130 |
| 4 | 6 × 1011 | 6 × 106 | 0 | 9.9 | 7.2 | 3 | 412 |
| 5 | 2 × 1010 | 1 × 105 | 14 | 20.5 | 53.6 | 3 | 412 |
| 6 | 4 × 108 | 5 × 103 | 75 | 6.8 | 6.5 | 3 | 340 |
| 7 | 5 × 109 | 2 × 105 | 0 | 2.4 | 0.0 | 3 | 225 |
| 8* | 1 × 1011 | 7 × 105 | 2 | 18.4 | 50.5 | 3 | 225 |
| 9 | 2 × 109 | 7 × 104 | 13 | 2.1 | 0.0 | 3 | 225 |
| 10* | 4 × 109 | 5 × 104 | 28 | 7.8 | 31.5 | 3 | 225 |
| 11 | 2 × 108 | 3 × 103 | 43 | 5.4 | 31.6 | 3 | 225 |
Notes:
Load - the total number (MPN) of E. coli in the outflow over the duration of an experiment
Conc - the event-mean concentration (MPN/100 mL)
Days - the number of days elapsed since the paddock was last grazed
Flow - the total flow passing over the outlet flume during an event (m3)
3d-rain - the 3-day antecedent rainfall total (mm), and includes simulated rainfall used to wet-up the soil prior to an event
Graze - the number of days that sheep grazed the paddock during the last grazing period
Stock - is the number of sheep per hectare during the last grazing period
*A reduced rainfall rate was applied over a period of two hours. In all other experiments, the design rainfall rate of 35 mm/hr was applied over one hour
2.3 Results
Total flow over each event varied markedly between events (Table 1), reflecting variation in antecedent soil moisture and the rate and duration of rainfall application. Time-series outflow and outflow E. coli concentrations for each event are shown in Figure 1. These often illustrate an initial flush of relatively high bacterial concentration, prior to the attainment of peak flow. This is followed by a gradual decline in concentration over the remainder of the event. The total number or load of bacteria washed across the outflow flume during an event varied between 2 × 108 and 6 × 1011 MPN. Since the outflow drains into a headwater stream, these loads represent a substantial delivery (2 × 105 to 6 × 108 MPN/m2) of E. coli direct to the stream network.
Both the loads and event mean concentrations of bacteria (Table 1) correlate with the recent stock history prior to each experiment (Figures 2 and 3). Events undertaken immediately following sheep removal (or whilst sheep were still present) gave an event-mean concentration of between 2 × 105 and 6 × 106 MPN/100 mL. Concentrations then declined exponentially with the time elapsed since the paddock was last grazed (Figure 2). A similar exponential decline in event load was observed with the time elapsed since the last period of grazing (Figure 3).
Multiple Regression was used to attempt to develop a statistical model with which to predict event load and event-mean concentration. In addition to the number of days elapsed since grazing, total flow, the 3-day antecedent rainfall, the number of animals and their duration of grazing were all examined as independent variables, in an interactive stepwise selection procedure. The strength of relationships was assessed using the coefficient of determination (r2), expressed as percentage and adjusted for degrees of freedom. In the prediction of event-mean concentration, ‘time elapsed since grazing’ alone was the strongest predictor (r2=0.78; Figure 2). That is, the addition of the other independent variables made no improvement to the model fit. For the prediction of event load, the combination of two independent variables – ‘time elapsed since grazing’ and ‘total flow’ explained 67% of the observed variance. The addition of other variables made no improvement to the model fit.
Figure 1: Time-series outflow (L/s, 2nd Y axis) and outflow E. coli concentration (MPN/100 mL, Y axis) for the 11 rainfall simulator events. Units of time are minutes after rainfall began
Figure 2: The decline in event-mean concentration with the time elapsed since the last grazing period
Figure 3: The decline in event load with the time elapsed since the last grazing period
Contact for Enquiries
Phil Journeaux
Manager
North Island Regions
Sector Performance Policy
MAF Policy
Private Bag 3123
Hamilton
NEW ZEALAND
Phone: +64 7 957 8313
Fax: +64 7 957 8315
Contact this person
