A convincing image can still contain measurements that are unsuitable for a particular task. Quality control asks whether the inputs, transformations, and outputs meet the requirements of the intended application. This perspective follows evidence through the workflow, from acquisition metadata and reference observations to the final interpretation. It encourages teams to define acceptance conditions before seeing the result, rather than adjusting them to favour a preferred conclusion. The purpose is a repeatable way to distinguish usable observations, observations requiring qualification, and observations that should be excluded from a decision.
This working guide considers estimating coastal water quality from ocean colour across peatland forest landscapes. The immediate decision is to select locations for hydrological field review. Useful reference evidence includes water-level records, land-cover observations and dated site visits. The case is an illustrative planning context, with assumptions to check against the actual sensor, landscape and project requirements.
The setting: Peatland forest landscapes
A peatland forest landscape connects vegetation, water conditions, drainage features and land-management history. A satellite image can reveal patterns relevant to these relationships, but a canopy response does not directly measure the depth or condition of the peat below it. Define the monitoring objective carefully. Mapping cover change, identifying areas for hydrological inspection and evaluating a carbon-related parameter require different observations and should not be presented as a single interchangeable result.
A landscape inventory should distinguish forest, open peat, managed land, canals and other surface types that contribute to an observation. Seasonal water conditions and the timing of management events may affect the image record. Optical observations can be limited by cloud and haze, while complementary measurements require their own interpretation. Track usable acquisitions and preserve uncertainty about the timing of a change when it falls within a gap between observations.
Field validation should retain water-level measurements and canopy or land-cover observations as separate evidence streams. Record the date, location and spatial setting of each sample, including its relationship to drainage infrastructure where relevant. A vegetation pattern may help focus a visit without establishing the cause of a hydrological change. More specific claims about emissions, peat condition or restoration performance need suitable measurements and an evaluation method that addresses those variables directly rather than inferring them from greenness alone.
Identify the quantity behind the colour
Coastal water colour carries information about materials that absorb and scatter light, including phytoplankton pigments, suspended particles, and dissolved organic matter. A satellite measures radiation reaching the sensor; estimating water quality requires separating the atmospheric contribution and then interpreting the remaining water signal. Specify the target variable and units before choosing an algorithm. Chlorophyll concentration, turbidity, suspended sediment concentration, and water clarity are related in some settings, but they are not interchangeable measurements.
Coasts are especially difficult because several optically active constituents can vary independently. A band ratio calibrated for clear offshore water may fail in a sediment rich estuary. Reflected light from nearby land can contaminate coastal pixels, sun glint can dominate the surface signal, and shallow bottoms can influence apparent colour. Cloud, haze, whitecaps, and mixed land water pixels further reduce usable coverage. Selecting an atmospheric correction method and applying its quality flags are scientific choices, not merely file preparation. Examine spectra and spatial patterns before accepting a plausible looking concentration map.
Match observations to a changing coast
Design validation around the speed of coastal change. Field measurements taken several hours away from the satellite overpass may describe another water mass after tides, currents, or river discharge have shifted conditions. Record sampling time, position, depth, instrument calibration, laboratory methods, and the spatial variability around each station. Satellite pixels integrate an area, while a bottle sample represents a much smaller volume. Use suitable matchups and inspect surrounding pixels rather than pairing every station with the nearest value automatically.
Develop or select retrieval relationships using samples that span the expected range of conditions. Reserve independent data for evaluation and report bias, error, valid sample count, and cases rejected by quality screening. Compare algorithms on the same matchups to avoid misleading differences caused by sampling. Provide acquisition dates, water masks, uncertainty information, and clearly labeled units with every map. A surface optical product does not establish conditions throughout the water column, and a pigment estimate does not by itself identify a harmful species or toxin. Used with field observations and local hydrology, the imagery can reveal spatial gradients and help prioritize where additional sampling is needed.
Design an evaluation that reflects deployment
A model evaluation should resemble the conditions in which the model will be used. Start by identifying which relationships the development data share: nearby locations, repeated observations of the same unit, common collection teams, or measurements from a single period. These relationships can make records less independent than they appear. An evaluation that ignores them may test the ability to recognise familiar conditions rather than the ability to operate in a new setting.
Choose a split strategy that matches the intended transfer. Holding out locations, seasons, operational units, or acquisition conditions can answer different questions. Document which question each split is designed to address, and keep the held-out evidence separate while selecting features and tuning the method. Repeatedly inspecting the same evaluation set to improve a model gradually makes that set part of development, even if it retains its original label.
Report results at a scale that is meaningful to the user. A strong average can hide poor performance for important subgroups or unusual conditions. Examine error patterns and retain examples of both successful and unsuccessful predictions. Where the evidence covers only a limited range of conditions, describe that range explicitly. The evaluation then becomes a practical guide to suitable use, further testing, and the collection of additional reference observations.
Translate the decision into observation requirements
Start with the decision that the information is intended to support. Describe the geographic unit, the time available for acting, and the consequence of an incorrect interpretation. A requirement such as detecting an unusual pattern across management blocks is different from identifying its cause. Keeping that distinction explicit helps a team select measurements that are relevant to the application and prevents later evaluation from drifting towards whatever result happens to be easiest to demonstrate.
Build a compact requirements matrix linking each decision to the evidence needed, the proposed observation source, and an acceptance condition. Include geographic accuracy, temporal relevance, quality screening, and the context required for interpretation. The matrix should also identify the evidence used to justify each requirement. An engineering preference or a convenient data source can inform a study, but neither should silently replace a documented information need.
Review the matrix with the people who will use the result. Ask how their response would change when information is delayed, incomplete, or uncertain. Record unresolved requirements as questions for a pilot rather than converting them into confident specifications. This produces a practical framework for comparing technical options, designing tests, and explaining why an observation is useful to the intended workflow.
Bring the evidence back to the decision
Quality control is most useful when its results travel with the information being delivered. Preserve the reasons for accepting or rejecting an observation, and expose limitations that could change the user's interpretation. Review recurring failures as clues about the design of the workflow. A clear quality record supports consistent comparisons between operators and processing versions, while helping the team decide whether a problem originates in acquisition, preparation, interpretation, or the underlying observation requirements.
Sources & further reading
Explore the bhumisat spacecraft concept and our vision towards 2030, or return to the journal.



