Major challenges in data integration for water quality nowcasting and forecasting include data availability, interoperability, and reliability [7,8,9]. This is true for either fundamental trend analyses or more sophisticated model calibration and validation. Spatially and temporally sparse sampling, various methods, and protocols for any given parameter; missing ancillary data; incomplete or nonexistent metadata; and difficulty in accessing data portals all contribute to uncertainties [7] and lead to questions about the robustness of the data that do exist, such as: do we know the metadata provenance and trustworthiness? Do we know that proper quality controls were implemented, such as standard calibrations? Answers to these questions and others are critical for determining whether data are findable, accessible, interoperable, and repeatable [10], which are necessary to ensure, for example, that the science used in decision making is based on the best available and trusted data.

There is also a tendency for sampling programs to neglect exchanges and interactions across systems. For example, sampling only the receiving water body, such as an estuary or lake, but not the supplying rivers, streams, and groundwaters prevents consistent and comprehensive monitoring of water quality across the aquatic continuum and hinders the assessment and understanding of processes and pressures that operate across these interconnected systems. Evaluating such systems requires a better mechanistic understanding of the fluvial continuum from catchment to receiving water body. Evaluation must also include the preconditioned natural forces of change acting on systems to assess their resilience to absorb human-induced perturbations and extreme events for effective management, and to avoid tipping points [11]. This requires long term data streams that are consistent both spatially and temporally, which can be a challenge to find among government agencies. Data continuity and fidelity can also be hindered by short project funding cycles that have long term consequences.

Developing countries and remote locations often have minimal technical or institutional capacity to monitor water quality, manage the data, and report results [12]. This compounds the challenges of moving toward integrated water quality measures. Although international standards for water quality assessment (e.g., ISO—International Organization for Standardization) and data sharing, including metadata (e.g., FGDC—Federal Geographic Data Committee, ODC—Open Data Cube), are in place, no single standard is widely adopted and followed, thereby contributing to this challenge. All this can lead to large observational gaps of robust water quality data, needed to underpin water management decisions. One solution is to rely on innovative monitoring approaches, such as via the integration of satellite and drone remote sensing, citizen science, and models. A demonstration project in Bangladesh recently highlighted the impact of the use of satellite data by decision makers, where coordinated or centralized capabilities ensured users were trained for the use of satellite information and understood the limitations of different satellite products [13]. Global provision of water quality data often requires the integration of different types of data from different sources for use by different communities. Scattered initiatives provide databases and platforms but lack concrete sustainable support to maintain and widely use those infrastructures. A global platform will enable the community to access information globally, on which analysis can be carried out, decisions can be made, and policy questions can be answered.

Measurements from different observing platforms, sensors, and data processing systems have different capabilities and characteristics, including differences in accuracy, number and type of parameters provided, temporal frequency, spatial coverage and resolution, and cost efficiency (Figure 3). The physical, chemical, or biogeochemical parameters of interest are manifold, but preference is increasingly given to those advanced as Essential Biodiversity, Climate or Water Variables [14,15,16]. In situ measurements often have high vertical (water column) resolution but low horizontal resolution, here defined as the capability to distinguish horizontal gradients, whereas, ocean color satellite data provide the converse. For example, in monitoring a river outflow or wastewater discharge, field sampling and sensors can provide data on many parameters at high frequency, potentially at high spatial resolution (in centimeters) and in both horizontal and vertical dimension, but low spatial coverage, whereas a satellite sensor (10 to 30 m) or drone (10 to 100 cm) may provide relevant information at sufficient spatial resolution and high spatial coverage, but miss variability in the vertical dimension. Another example may be a lake with a public drinking water intake. Here, an ocean color satellite may be useful for monitoring the entire lake, in situ sensor may benefit for high frequency sampling at the depth of the intake, discrete sampling enables event-driven responses, and models may be useful to understanding how upstream anthropogenic stressors may impact this resource and test scenarios to protect and restore water quality. If these sources are properly integrated, as proposed, many of the trade-offs and strengths are mitigated, enabling a more comprehensive toolbox for monitoring and interpreting water quality.