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 harmonising satellite observation time series across coconut production landscapes. The immediate decision is to identify areas needing plantation review. Useful reference evidence includes palm spacing, understory condition and farm-level observations. The case is an illustrative planning context, with assumptions to check against the actual sensor, landscape and project requirements.
The setting: Coconut production landscapes
Coconut production landscapes can range from organised plantations to widely spaced palms embedded in farms, settlements and coastal vegetation. These settings create very different observation problems. A satellite pixel may combine palm crowns, soil, grass, roofs or water rather than represent coconut vegetation alone. Define whether the task is mapping a production landscape, documenting canopy change or studying a crop-specific variable before choosing an image product or interpretation method.
Field and management boundaries are important even where ownership units do not coincide with a continuous canopy. Record the extent of intercropping and other vegetation that may contribute to an observed change. In coastal settings, inspect shoreline adjacency, temporary flooding and atmospheric conditions alongside the imagery. A field-average vegetation response should be accompanied by the area and quality of the observations from which it was calculated.
Validation should describe palm arrangement, canopy continuity, understory and recent land-management events at known locations. Samples drawn only from dense or easily accessible stands may not represent the wider landscape. A model intended for crop condition or production needs reference measurements suited to that claim and evaluation beyond its development sites. For many users, a transparent inventory and a well-supported list of locations requiring review provide a practical foundation for later, more specialised analytical services.
More observations require a common measurement definition
Combining observations from different satellites can increase the information available through time, but the resulting series needs a common analytical definition. Instruments may differ in spectral response, spatial sampling, viewing geometry and processing history. Two bands with similar names are not necessarily equivalent measurements. Harmonisation addresses selected differences so that observations can be used together more consistently. It should be understood as a documented processing approach with remaining limitations, rather than a guarantee that every value from every sensor is interchangeable.
A suitable workflow begins with the variables that need to be compared. Identify the input products, their physical meaning and the transformations already applied. A common grid helps organise observations spatially, while atmospheric correction, geometry normalisation and spectral adjustments may address other differences. These operations have distinct purposes. Applying only one of them should not be described as complete harmonisation. Retain sensor identity and acquisition metadata even when the user sees one combined series, because those attributes remain valuable for diagnosing an apparent discontinuity.
Validate continuity before interpreting a longer record
Test continuity using overlapping or otherwise comparable observations and independent evidence where available. Inspect several surface types and canopy conditions rather than evaluating only one convenient landscape. A consistent average relationship can still hide a meaningful difference for a particular crop or season. Examine how the combined record affects the final product, including index thresholds, fitted timing metrics and change alerts. The relevant question is whether the remaining differences are acceptable for that application, not merely whether the images appear similar on a map.
A longer record also requires a stable policy for quality screening and updates. Changes in input collections, calibration or processing versions can affect historical outputs. Version the workflow and preserve enough information to reproduce earlier results when necessary. When a series crosses a known processing change, evaluate the transition explicitly. A harmonised dataset can provide a strong foundation for monitoring, but field context and validation still determine the usefulness of its interpretation. Maintaining traceability allows additional observations to improve coverage without concealing their origins or implying a precision that has not been demonstrated.
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.
Design field observations around the information need
Field observations are most useful when they answer a defined question within the monitoring workflow. Begin by identifying the variable, condition, or management event that needs to be recorded and the geographic unit to which it applies. Distinguish direct measurements from an observer's interpretation. A consistent record of crop stage, management timing, and visible conditions can provide valuable context even when the field team is not collecting a complex set of measurements.
Prepare a collection protocol that reflects the variability of the site. Define where observations are made within a unit, how many locations are represented, and how the position and time are recorded. Avoid selecting only the most dramatic or convenient examples. Include apparently normal conditions and records that might challenge the remote interpretation. Where a measurement method has known limitations, preserve the information needed to assess whether those limitations affect the intended comparison.
Link field records to the corresponding remote observations without losing their original detail. Geographic matching, timing, and the scale represented by each record should remain visible. Review disagreements with the collecting team before changing a label or excluding an observation. A disagreement may reveal a timing mismatch, a mixed surface, or an incomplete interpretation rather than a simple error. The field programme should improve the understanding of the monitoring process as well as provide evidence for evaluating its outputs.
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.



