River Turbidity: Baseline, Episodic Peaks and a Maintenance Outlier
A volunteer group samples River Alden weekly for 12 weeks at two sites: Upstream (U), in woodland, and Downstream (D), below a housing development and a storm drain. Grab samples are taken Tuesdays at 09:00. Turbidity medians are 4.2 NTU at U (IQR 3.1–5.8) and 11.6 NTU at D (IQR 8.9–15.3). Two heavy rain events occurred, 28 mm in week 5 and 41 mm in week 9, both on Sundays. At D, turbidity peaked at 46 NTU and 61 NTU one day after each event and returned to baseline within 3–4 days; at U, the peaks were 9 and 12 NTU. In week 10, a D sample measured 210 NTU after a culvert repair upstream of the site. The group has no flow data and no continuous logger. A draft report says downstream turbidity is 2.8 times upstream, proving the housing development causes chronic pollution.
1. Evaluate the draft report's claim that downstream turbidity is 2.8 times upstream and that this proves chronic pollution from the housing development.
Read the worked answer
The arithmetic is roughly right for the medians: 11.6 ÷ 4.2 ≈ 2.76, so '2.8 times' is a fair summary of the median ratio. The interpretation is not supported.
First, the medians mix baseline days with the tails of two storm events. The event peaks at D (46 and 61 NTU) are much larger than at U (9 and 12 NTU), so a median calculated over all 12 weeks can be raised by a few event-influenced samples. 'Chronic' implies a persistent baseline difference; the data as described show a clear episodic response. The group should separate dry-week samples from event samples and compare the dry-week medians.
Second, causation is not established. There is no before-and-after data, no measure of flow, and only one downstream site. Turbidity depends strongly on discharge, so if the downstream reach has a different channel slope, drain input or flow regime, the ratio could reflect hydrology rather than development. The week 10 value of 210 NTU is a known maintenance disturbance and should not be used as evidence of routine pollution.
A defensible statement is: 'Downstream turbidity is higher than upstream, particularly after rain, consistent with a downstream sediment source; flow and event-focused sampling are needed to test whether development is responsible.'
2. Explain how the week 10 value of 210 NTU should be handled in the analysis and what sampling change would reduce this kind of problem in future.
Read the worked answer
The value should not be silently deleted. It is a real measurement taken after a known culvert repair, so it is not representative of ordinary conditions but it is relevant to disturbance impacts. The group should flag it in the dataset with the repair date, report results with and without it, and state clearly that it was excluded from the baseline comparison only because the cause is known. If the report is about pollution from the development, including it would distort the conclusion; if the report is about sediment risks, it is an important event.
To reduce the problem, the group should pair grab sampling with a continuous turbidity logger at each site, or at least trigger extra samples when rainfall exceeds a threshold and when maintenance is scheduled. A logger would capture the rising and falling limb of events that weekly Tuesday sampling misses. Field blanks, calibration checks and a note of any upstream works would also help distinguish instrument drift from real sediment.
3. Which single additional measurement would most strengthen the inference that the housing development increases turbidity, and why?
Read the worked answer
Stream discharge (flow) at both sites is the most useful addition. Turbidity is strongly flow-dependent: a higher flow can mobilise bed sediment and increase turbidity without any change in land use. If D has higher turbidity at the same flow as U, that is much stronger evidence of an additional downstream sediment source. If D is higher only because it carries more water after rain, the development explanation is weakened.
Flow data also allow the group to compare turbidity at matched discharges rather than at matched calendar dates, which is important because the two sites may respond to rain at different times. A continuous turbidity logger would be valuable, but without flow it still leaves the same confound. A second downstream site or before-and-after construction data would also help, but flow is the measurement that most directly separates 'more sediment' from 'more water'.