Interpreting catchment management actions at the water treatment plant

Annual average modelling but daily problems

Seqwater is the Bulk Water Supply Authority, which operates 39 water treatment plants to provide safe, secure and cost-effective drinking water to over 3 million people, in South East Queensland (SEQ).

Along with its delivery partners, Seqwater operates an extensive source water protection program, involving on-ground activities, to improve the quality of water delivered to its water treatment plants.

The potential reductions in modelled, long-term, annual average loads to water treatment plants from the source water protection activities are predicted using the Catchment Investor (formerly known as the Catchment Investment Decision Support System).

However, water treatment plant operation constraints are based on daily water quality conditions. Specifically, source water protection impacts the day-to-day water treatment operations required. For example:

• the need to dose with flocculant
• managing sludge handling
• constraining throughput due to high suspended sediment loads
• managing disinfection processes due to microbial loads and suspended sediment loads.

The challenge is how to interpret changes in modelled, long-term annual loads to potential changes in day-to-day treatment plant operation.

The problem is that treatment plant operation depends on daily inputs, but current source protection modelling does not occur at a daily timestep.

Model at a daily timestep or disaggregate?

There are two basic solutions options:

1. to model all source protection intervention activities using a daily timestep model (e.g. using eWater Source), or
2. to disaggregate the modelled, long-term, annual load changes to represent the change in the temporal pattern of pollution at the plant.

Parameterising a daily timestep model is attractive; however the challenge is parameterising such a model to represent the 60+ source protection intervention activities and the 20+ measures that they impact upon in a timely manner. Building and calibrating such a model is difficult, and then running tens to hundreds of scenarios is very time consuming.

We advocate, and have implemented, a temporal disaggregation of long-term average annual loads because the Catchment Investor already handles the 60+ source protection activities and is designed for rapidly creating, running and exploring scenarios at speed.

How did we disaggregate?

Water treatment plant operation is focused on removing microbial pathogens. The efficiency of pathogen removal is heavily dependent on the suspended sediment load. Suspended sediment needs to be removed through flocculation and sedimentation to effectively remove pathogens. Our focus on disaggregation was therefore to represent the daily suspended sediment loads.

The suspended sediment loads are expressed in terms of water turbidity for different treatment plant risk levels.

Step 1: Total suspended sediment vs turbidity

By analysing measured total suspended sediment (TSS) and turbidity data collected by Seqwater and the Queensland Government, from a range of historical and contemporary monitoring programs, we determined tight correlations between TSS and turbidity (regionalised by catchment size). This means that if we can predict daily TSS loads, then we can convert that to an equivalent turbidity at each water treatment plant.

Step 2: Predicting TSS

eWater Source modelling has been conducted for several catchments in SEQ. The model generates discharge at a daily time step. There is reasonably good correlation between TSS and volumetric flow rate (flow) at a site.

A common approach would be to regionalise these relationships (based on catchment size) to be able to apply them to catchments that are not modelled. However, we do not have measured flow at all sites, hence we could not produce a daily time series of TSS everywhere.

So, we have used these relationships to create distributions that can be applied in ungauged sites.

Step 3: Temporal distribution of turbidity

By applying the TSS-turbidity relationship from step 1 to daily modelled TSS at the available sites, a distribution of turbidity was established. The regionalised TSS-flow relationships (step 2) were used to apply these distributions to catchments outside the modelling range.

Step 4: Water treatment plant risk profiles

Using the water treatment plant specific risk profiles (different risk levels for different turbidity values), and the turbidity distributions from step 3, we quantify the number of days per year (on average) that each water treatment plant is at different risk levels under the current catchment condition.

Step 5: Predicting changes to risk

Catchment Investor predicts long-term annual average reductions in pollutant loads. The application of source protection intervention activities effectively changes the TSS-flow relationship from step 2.

This change in relationship is not linear. Source protection has very little impact for extreme events (occurring <1% of the time). We adjusted the TSS-flow relationship (excluding extreme events) by scaling TSS concentration and combining with flow to derive the long-term load.

When combined with step 1, step 3 and step 4, we get a distribution of change in time-spent at each water treatment plant risk level for different predictions of long-term annual load reduction.

Putting it together

Catchment Invetstor predicts the long-term annual average load reduction from source protection intervention activities.

The result of the disaggregation process produces a water treatment plant-specific function that converts the load reduction to a change in the number of days at different risk levels.

Exploration of results in Catchment Investor then presents scenarios of source water protection in terms of what is useful for water treatment plant operators. For example, if we protect this sub-catchment with tree planting, the number of days spent at ‘risk level 4’ will reduce from 40 days per year to 10 days per year.

The approach presented here is a pragmatic one, constrained by the data available. We look forward to revising the analysis as more data becomes available.

Learn more about Catchment Investor >>