Source Water Quality in a Changing Climate: Rethinking What, When and Where We Monitor
No monitoring program can see everything, but it needs to see enough to manage the risks. Climate change keeps moving the goalposts.
After a long drought, catchments build up sediment, nutrients and microbial contamination, and the first big storm washes it into rivers and storages. Bushfire makes this worse: burnt, water-repellent soils shed rain quickly, carrying ash, organic carbon and metals with it.
No monitoring program can see everything; it needs to see enough to manage the risks, through normal conditions and events. What has changed is the risks themselves. Drought–flood cycles, bushfire and longer, hotter summers are shifting baseline quality and making events more frequent and different in character. In some systems, what is sampled, when and where no longer matches where the risk now sits.
The bar has also risen. Utilities are now expected to manage trihalomethanes, not just protect disinfection, and that means understanding the organics coming into the plant: what kind, not just how much. In many catchments, storms and bushfire are changing exactly that.
The goalposts are moving. The question is whether monitoring programs are moving with them.
Stratification: longer, stronger and harder on treatment
Warmer conditions are making reservoir stratification stronger and longer-lasting. With little vertical mixing, the bottom layer runs out of oxygen. Manganese, iron and phosphorus are released from the sediments, and organic matter breaks down anaerobically, changing the character of the organics in that water. At the surface, cyanobacteria take advantage of the light and nutrients.
What reaches the plant depends on where the offtake sits. Storages with a variable offtake can usually select better water during stratification, but are exposed when turnover mixes bottom water through the column. Storages with a fixed or low-level offtake may be drawing that water for much of the season.
Either way, surface grab samples and fixed-depth sondes can miss it. Depth profiling of temperature, dissolved oxygen and metals through summer and autumn shows what is coming. When low-oxygen water is being drawn, chlorine demand and THM formation potential are worth tracking alongside manganese and iron, because coagulation, taste and odour, and disinfection by-products can all shift quickly when the source water changes character.
Post-bushfire signatures and the multi-year tail
Burnt catchments behave differently for years after a fire. Rain on burnt slopes delivers ash, sediment, nutrients, metals and organic carbon, and the organics arriving differ in character from those of an unburnt catchment. For treatment, that shows up as changes in coagulant demand, chlorine demand and disinfection by-product formation that bulk DOC alone won't reveal. Nutrient and organic loads can also drive algal growth and deepen bottom-water oxygen depletion in receiving storages, compounding the stratification problem, for two to three years after the fire.
Most post-fire monitoring captures the first wet season and then reverts to the standard program. The event response isn't the problem; its scope and duration are. Programs in fire-prone catchments need pre-season baseline data, rainfall-triggered sampling, and a parameter set covering organic character (UV254 or THM formation potential, for example), trace metals and nutrients. That expanded program should run until the data show the catchment has settled, not for a fixed period, recognising that the settled state may not match the pre-fire baseline.
Storm events: triggers built on yesterday's storms
More intense rainfall produces faster, higher turbidity peaks, and the first flush from a dry or burnt catchment carries a disproportionate share of the load. Most utilities already respond with continuous intake turbidity monitoring and event-triggered sampling. The question is whether the triggers still fit. Rain-gauge thresholds, alarm levels and assumed lead times set from historical events may now fire too late, or not at all, for the storms the catchment is getting.
Turbidity also tells only part of the story. Organic carbon doesn't necessarily peak with turbidity, so a plant tracking turbidity alone can be caught out on coagulation and chlorine demand after the visible event has passed. Storms also carry the highest pathogen loads. If event sampling misses the peak, the data underpinning a supply's source water category, and the log reduction target that follows from it, may understate the real risk.
Keeping the program in step
The test for a monitoring program isn't whether it sees everything. It's whether it still tells you enough to manage the risks you now have. That belongs in the periodic review the ADWG Framework already expects, not a one-off redesign.
A good starting point is the events of recent years that caught the plant out. From there, the review comes down to three questions:
What. Is the program measuring what the plant actually responds to, including the character of incoming organics, not just bulk DOC? After a fire, was the parameter set expanded, and for long enough?
Where. Does sampling cover the depths, offtakes and tributaries that now carry the risk?
When. Do event triggers, alarm levels and lead times reflect the storms the catchment is now producing, and is event sampling capturing the pathogen peaks that inform source water categorisation?
A season of more intensive depth profiling and event sampling is a practical way to answer these questions and to set a baseline for the next review. Targeted continuous monitoring at one or two critical points will often add more than extra routine grab samples, although grab samples still do verification work instruments can't.
Because the goalposts are moving, the review isn't a project with an end date. It should recur on a set interval and be brought forward by a major fire, flood or shift in raw water behaviour.
Climate change hasn't made source water monitoring wrong, but it is changing the risks monitoring was set up to manage, and it will keep doing so. Three checks are worth making now. Find out when event triggers and alarm levels were last set, and against what. Ask whether the parameter set reflects what the plant now responds to: metals and cyanobacteria during stratification, organic character for THM control, pathogen peaks during storms. And pull the data from the last few events that challenged the plant, and ask whether monitoring saw them coming.
Comments