A method that produces a plausible result has not automatically demonstrated that it works for its intended users. Validation requires a defined question, suitable reference evidence, and an evaluation design that reflects how the method will be used. This perspective examines which observations should be compared, how they should be sampled, and which errors matter in practice. It also considers the difference between evidence used to develop a method and evidence reserved for evaluating it. The objective is a test that can reveal weaknesses as well as confirm useful performance.
This working guide considers spectral band alignment and common grids in open-field tomato production. The immediate decision is to select tomato rows for crop review. Useful reference evidence includes transplanting dates, canopy observations and field management notes. The case is an illustrative planning context, with assumptions to check against the actual sensor, landscape and project requirements.
The setting: Open-field tomato production
Open-field tomato production can involve short management cycles, staggered planting and repeated operations within neighbouring parcels. The field calendar is therefore central to interpreting a satellite observation. Newly transplanted plants, established rows and areas nearing the end of production can show different responses for ordinary reasons. Record the planted area and the relevant management unit carefully, particularly where a commercial holding contains several crops or planting stages at the same time.
The image footprint may mix plants with soil, mulch, access routes and irrigation infrastructure. Determine whether the spatial resolution supports the intended row-level interpretation or whether a broader field summary is more appropriate. Examine the quality and age of observations before proposing a management use that depends on rapid response. A monitoring workflow should communicate when cloud gaps or the acquisition schedule leave insufficient current evidence for that decision.
Validation should document crop stage, canopy continuity, recent operations and the area represented by each sample. Field staff may use a mapped anomaly to choose an inspection location, but the subsequent diagnosis requires suitable observations and expertise. If production or quality is the target, collect reference information for those outcomes rather than assuming that greener vegetation implies more marketable fruit. The evaluation should measure whether the product improves a defined scouting or management process.
A multiband value should refer to the same ground
A multispectral analysis combines information from several channels, so their geographic relationship matters. If the channels are displaced, a calculation near a field boundary may combine vegetation in one band with a path, ditch or neighbouring surface in another. The resulting pattern can look like a meaningful spectral difference even though it is partly an alignment artefact. A workflow should therefore consider band alignment before producing indices, classifications or crop parameter estimates, particularly when the target landscape contains small fields and sharp transitions.
Alignment is not the same as assigning every band the same pixel dimensions. Bands may have different native sampling and observation characteristics. Resampling them onto one grid creates a shared representation, but it does not make all channels contain the same original spatial detail. The selected interpolation method can change values around edges and influence derived statistics. Record the target grid, coordinate reference system, pixel alignment and resampling choices so that another analyst can reproduce the combination and evaluate its consequences.
Assess the effects of resampling
Check alignment using features that can be recognised reliably in the relevant channels and are stable at the time of acquisition. Inspect several parts of the area rather than relying on one visually convenient location. A displacement may vary across the image or become more obvious in particular terrain. Quantitative checks should use suitable independent references and report the scale of the remaining error. A correction that appears convincing on a broad overview still needs examination at the spatial scale used by the application.
When evaluating an agricultural model, test whether alignment sensitivity changes its result. For example, compare field statistics with and without a documented interior boundary exclusion, or examine the effects of a small plausible displacement. Such checks help reveal whether a conclusion depends mainly on edge pixels. Avoid using repeated interpolation steps where one controlled transformation would suffice. A clear geometric processing chain makes spectral combinations more defensible and helps ensure that a reported crop difference reflects observations of comparable ground rather than a hidden mismatch between image channels.
Keep an auditable review record
A review record should explain how a result moved from an analytical output to an accepted interpretation or operational decision. Preserve the product identifier, the evidence considered, the reviewer role, and the outcome of the review. A short, consistent record is often more useful than an extensive free-text note that cannot be compared between cases. The objective is to make important decisions traceable without turning routine review into unnecessary administrative work.
Separate observations from conclusions within the record. Describe the evidence visible to the reviewer, the alternative explanations considered where relevant, and the reason for accepting, qualifying, or rejecting the result. If a correction is required, identify what should change and how the revised product relates to the earlier version. Keep unresolved questions visible so that they can be addressed by another observation or by the person responsible for the next action.
Use the accumulated record to improve the workflow. Recurring review reasons can reveal unclear requirements, missing contextual information, unstable processing steps, or an evaluation gap. Examine whether similar cases receive similar decisions across reviewers and over time. Preserve the relationship between feedback and subsequent revisions so that improvements can be assessed against the evidence that motivated them. An auditable record supports learning, accountability, and consistent delivery while allowing the technical process to evolve.
Plan the evidence needed for calibration
Calibration connects an instrument's recorded response with a measurement that can be interpreted consistently. Begin by identifying which aspects of the response matter to the application: variation across channels, changes with operating conditions, geometric alignment, or differences between repeated observations. A calibration plan should state the quantities being assessed and the evidence needed to determine whether the response is sufficiently characterised for the intended use.
Design tests around controlled and documented conditions. Record the reference source, instrument configuration, exposure settings, temperature conditions, and processing applied to the measurements. Repeat suitable observations to investigate variability and distinguish a consistent response from a single favourable result. The reference itself has limitations, so preserve its characterisation and avoid assuming that every difference originates in the instrument under evaluation.
Connect calibration activities with the expected operating environment and processing chain. A correction that works under one test condition may require further evidence before being applied elsewhere. Define how calibration parameters are versioned, checked, and associated with the observations they support. Include a plan for recognising drift or other changes that could affect interpretation. The outcome should be an instrument response supported by traceable evidence, together with acceptance criteria that remain relevant to the application.
Bring the evidence back to the decision
The validation record should explain the population represented by the test, the reference observations used, and the limitations of the comparison. Report meaningful errors alongside useful successes, and retain examples that challenge the method. Treat wider deployment as a new opportunity to evaluate transfer across relevant conditions. A well-designed validation programme gives users a defensible basis for deciding where the method is useful, where additional evidence is needed, and where another approach may be appropriate.
Sources & further reading
Explore the bhumisat spacecraft concept and our vision towards 2030, or return to the journal.



