Figure
2. Distributed sensor networks can be developed to support water
resource management : (a) REACTIVE ; a river catchment with sensor
locations denoted by numbers (1-7) spanning river channels. At t1, a
stressor (e.g. organic pollution, sedimentation) appears (upstream of
location 7) leading to enhanced ecosystem respiration (ER). Real-time
analytics and visualisation allow pollutant tracking through t2-t4,
enabling water abstraction (denoted by x) to be deactivated at t3. (b)PROACTIVE ; a river catchment with a large headwater reservoir.
Hydrograph shows discharge (Q) scenarios f1-4. Low flow f1 elevates ER
in the mainstem. With a regulatory target of ER 1-2.5, water release in
f2 modifies only the segment below the reservoir. Excessive water
release in f3 leads to overshoot of targets, allowing an optimal
solution in f4 to trade-off ecosystem recovery and water supply.
For example, abstractors using river water for drinking water supply can
identify contamination issues, such as high dissolved organic carbon
(DOC) concentrations upstream, thus avoiding problems whereby
disinfection byproducts make water unsuitable for human consumption
(Valdivia-Garcia et al., 2019). Specific examples include water
utilities and hydropower companies that withdraw, store, and
redistribute water around river systems facing management challenges
related to altered water quality (Gillespie et al., 2015). Such
approaches are already being tested, by diverting episodic events with
elevated DOC in raw water sources away from water treatment works
(Yorkshire Water, 2023), but sensor networks can be costly to implement
and maintain. The integration of forecasting into ML architecture
promises to strengthen and advance scientific understanding further by
feeding back to field sensors and samplers to collect higher resolution
data. For example, enhanced data collection during contamination events
could be used to support regulator investigations, and during storms
where runoff peaks are often missed, for enhanced understanding of water
quality and carbon cycle dynamics.
Regulators and decision makers need access to high-quality data to
develop, monitor and enforce catchment management plans and legislation,
and identify areas where persistent problems highlight the need for
restoration, such as through payment for ecosystem service or
nature-based solution initiatives. Additionally, managers of
agricultural basins, which are recognized as a leading source of global
water contamination (Liu et al., 2022) need evidence to manage and
reduce the effects of sediment loads and adsorbed contaminants
originating from soil erosion, and the use of agrochemicals (nutrients,
herbicides, pesticides), all of which can lead to elevated GHG emissions
from rivers (Xiao et al., 2021). By pinpointing river sections or
sub-catchments suffering from stressors, prioritized and targeted
management practices can meet multiple objectives to reduce emissions as
part of the water-energy-food nexus in global resource systems.