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 synthetic aperture radar for land monitoring on highland coffee farms. The immediate decision is to select coffee plots for field review. Useful reference evidence includes shade cover, coffee growth stage and terrain-aware field notes. The case is an illustrative planning context, with assumptions to check against the actual sensor, landscape and project requirements.
Radar measures a different interaction with the landscape
Synthetic aperture radar is an active observing technique: the instrument transmits microwave energy and measures the returned signal. Its image therefore represents a different interaction with the landscape from reflected visible or near-infrared sunlight. Radar can provide observations in conditions that limit optical imagery, but its response still depends on the instrument and scene. Backscatter should not be read as a universal moisture or vegetation score. Surface structure, water, roughness, geometry and the arrangement of scattering elements can contribute to the measured return.
Wavelength and polarisation influence which interactions are emphasised. The same agricultural field can change its radar response after rainfall, tillage, crop development or harvest, and several of those processes may occur together. The viewing direction and incidence angle also matter. A time series assembled without controlling these factors can contain differences that look like an environmental trend. Retain orbit, geometry, polarisation and processing metadata, and choose comparisons that are appropriate for the physical interpretation the application is intended to support.
Compare like with like before drawing conclusions
Radar images commonly contain speckle, a granular appearance associated with coherent measurement. Filtering or aggregation can make patterns easier to interpret, but it also changes spatial detail and statistical behaviour. Select processing around the reporting scale and evaluate its effects near boundaries. Terrain can create additional geometric challenges, including compressed or hidden areas in the radar view. A map should identify where the observation is difficult to interpret rather than filling every location with an equally confident result.
Complementary evidence helps distinguish plausible explanations for a radar change. Combine suitable optical observations, rainfall or management records and field measurements when they address the actual question. Validate a proposed model on independent locations and conditions, including cases where the expected relationship fails. A dark water surface in one setting does not establish a universal flood rule, because rough water, vegetation and built structures can produce different responses. Radar becomes valuable when its particular measurement physics is incorporated into a carefully defined monitoring workflow, with geometry and uncertainty remaining visible to the analyst and user.
The setting: Highland coffee farms
Coffee farms in highland terrain can combine coffee plants, shade trees, paths, soil and other crops within the same image footprint. The observed canopy may therefore describe the wider agroforestry system as much as the coffee itself. Before choosing an analytical method, decide whether the intended output concerns farm-level vegetation, shade management or a coffee-specific variable. These are distinct tasks and require different reference observations and levels of spatial detail.
Slope, aspect and changing illumination complicate comparisons between locations and acquisition dates. A dark patch on a shaded slope should not immediately be classified as stressed vegetation. Inspect terrain and shadow conditions alongside the image quality information, and compare areas with similar observation geometry where possible. Farm boundaries should also be checked carefully because small geolocation differences can move a sampling footprint from coffee into a neighbouring tree canopy or access track.
Field teams should document coffee stage, shade structure, recent pruning and the location of the plants or plots they measure. Samples collected beside a convenient path may not represent the full farm. A practical service could help organise visits or describe changes in farm cover, but claims about coffee condition or production need their own validation. Preserve these distinctions in the report so that a useful landscape observation is not presented as a plant-level diagnosis.
Represent gaps without inventing observations
A monitoring sequence often contains periods without a usable observation. The reason may be cloud, shadow, missing coverage, a failed transfer, or a quality condition that excludes the data. Begin by representing these states explicitly. A missing observation is different from a low measurement, an unchanged surface, or a confirmed absence of the feature being studied. The distinction should survive both the analysis and the presentation of the result.
Define when a temporal gap is acceptable for the intended decision. A workflow may continue using the latest qualified observation, request another source, or suspend an alert until sufficient evidence is available. Whichever response is selected, show the age of the supporting observation. If interpolation is used for a suitable analytical purpose, preserve the distinction between an estimated value and a directly observed measurement, together with the assumptions needed to interpret it.
Review how gaps affect the coverage of the final evaluation. Locations with frequent missing data may differ systematically from those with regular observations. A performance estimate based only on clear, complete records can therefore describe a narrower population than the service is expected to support. Report usable coverage alongside analytical results and use recurring gaps to guide acquisition planning, reference collection, and the selection of complementary observation methods.
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.
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.



