NDVI and NDMI: What They Show—and What They Do Not

NDVI compares reflectance in the near-infrared and red parts of the spectrum. Healthy green canopy reflects strongly in the near-infrared and absorbs red light, so denser and more vigorous vegetation produces higher values.

That relationship is robust but indirect. NDVI describes how a surface reflects light at a moment in time. It does not distinguish a thin stand from a stressed stand, and it saturates once canopy closure is complete, which limits its sensitivity at the upper end of the growing season.

NDMI compares near-infrared reflectance with a shortwave-infrared band that is sensitive to water in leaf tissue. It therefore responds to canopy moisture content rather than greenness alone.

Read alongside NDVI, it can help separate patterns. A canopy that remains green while NDMI declines may be entering moisture stress before visible change; a canopy where both decline together is consistent with a more advanced condition. Neither reading establishes a cause on its own.

A single NDVI value has little meaning without a reference. Crops, soils, planting dates and regional norms all shift the range that should be considered typical.

Comparison is what makes the number interpretable: the same field against a same-season reference median, or a sequence of observations across the analysis period. A trajectory that falls steadily across four usable scenes carries more information than one low reading that may reflect timing, haze or a partially obscured scene.

Optical satellite observation is interrupted by cloud, and cloud is not random. Extended unsettled weather removes scenes precisely during the periods a reviewer most wants to examine, and thin cirrus or cloud shadow can depress values without triggering an obvious flag.

For this reason the count of cloud-valid observations belongs beside the indicator, not in a technical appendix. A trend built from two usable scenes should not be presented with the same visual confidence as one built from six.

Programs turn on definitions: insured interest, covered peril, period of cover, measurement basis and documented loss. An index value addresses none of these directly.

Treating an indicator as a decision also removes the ability to explain the outcome. A reviewer asked to justify a position needs to point at conditions, dates and evidence quality — not at a composite number whose derivation is not visible.

Indicator values appear with the seasonal reference, the observation dates behind them, the number of cloud-valid scenes and the resulting data-quality state. Charts carry an accessible text summary and a table fallback so the same values can be read without relying on the visualisation.

Screening conditions are shown as text with the observed value beside each one, so a reader can reproduce the outcome by inspection rather than trusting an opaque calculation.