Boundary-free monitoring

NDVI — Normalized Difference Vegetation Index

Sentinel-2 · B08 (NIR, 842 nm) · B04 (Red, 665 nm)

Chlorophyll absorbs red light strongly at its photosynthetic peak (~680 nm), while the palisade and spongy mesophyll layers of a leaf reflect and scatter NIR radiation intensively through multiple internal cell-wall interfaces. The reflectance contrast between these two regions therefore scales with canopy greenness. NDVI normalises this contrast to the interval [−1, 1]. Dense healthy canopy yields values of 0.6–0.9; sparse or stressed vegetation 0.2–0.5; bare mineral soil approximately 0.1; open water and cloud produce negative values. NDVI saturates in high-biomass canopies because chlorophyll absorption in the red channel reaches a floor before NIR reflectance reaches its ceiling with increasing leaf area index.

NDRE — Normalized Difference Red Edge

Sentinel-2 · B08 (NIR, 842 nm) · B05 (Red Edge, 705 nm)

The red-edge region (680–750 nm) is the steep reflectance slope between the chlorophyll absorption maximum and the NIR plateau. The position and gradient of this slope shift as a function of chlorophyll concentration: a nitrogen-deficient or senescing leaf narrows its absorption well, and reflectance at B05 (705 nm) rises accordingly. By comparing NIR to the red-edge band rather than to the red band, NDRE measures chlorophyll content rather than simply canopy cover. It does not saturate in dense crops where NDVI has ceased to discriminate, making it the preferred index for nitrogen management in precision agriculture. Typical values for a healthy crop canopy fall in the range 0.2–0.5.

NDWI — Normalized Difference Water Index

Sentinel-2 · B03 (Green, 560 nm) · B08 (NIR, 842 nm)

Open water reflects green light moderately and absorbs NIR strongly, producing a positive Green − NIR contrast. Vegetation inverts this relationship: chlorophyll partially transmits green light while the mesophyll strongly reflects NIR. The index therefore reliably separates open water bodies (positive values) from vegetated and dry surfaces (negative values). Following McFeeters (1996), this formulation uses B03 and B08; a related index by Gao (1996) substitutes SWIR for green and is better understood as a canopy moisture proxy — that version corresponds to NDMI below. In a crop context, NDWI is most informative at field margins, irrigation channels, and drainage features, and can detect inundation or waterlogging at plot scale.

NDMI — Normalized Difference Moisture Index

Sentinel-2 · B08 (NIR, 842 nm) · B11 (SWIR, 1610 nm)

Liquid water in plant tissue absorbs strongly in the short-wave infrared through overtone vibrations of O–H molecular bonds. A well-hydrated crop canopy therefore reflects strongly in NIR (from leaf structure) and absorbs in SWIR (from leaf water), yielding a large positive NDMI. As canopy water content declines under drought stress, stomata close, cells lose turgor, and SWIR reflectance rises as water is replaced by air, reducing NDMI. This makes NDMI a more direct estimator of plant water stress than NDVI, which responds primarily to chlorophyll concentration rather than hydration state. Typical irrigated crop values fall in the range 0.1–0.4; values approaching zero or below indicate severe moisture deficit.

EVI — Enhanced Vegetation Index

Sentinel-2 · B08 (NIR, 842 nm) · B04 (Red, 665 nm) · B02 (Blue, 490 nm)

NDVI is susceptible to two systematic biases which EVI corrects simultaneously. First, atmospheric aerosols preferentially scatter blue light through Rayleigh scattering, contaminating surface reflectance retrievals; the −7.5 ρ_Blue term in the denominator reduces this effect by coupling the blue channel to the aerosol signal. Second, at high leaf area index the red channel reaches its absorption floor before NIR reflectance reaches its maximum, causing NDVI to saturate; the linear denominator with coefficients C₁ = 6 and C₂ = 7.5 preserves sensitivity across the full biomass range. EVI was operationally developed for MODIS and transfers well to Sentinel-2, extending useful dynamic range in dense canopies and aerosol-prone conditions. Typical cropland values during active growth are 0.4–0.7.

SAVI — Soil Adjusted Vegetation Index

Sentinel-2 · B08 (NIR, 842 nm) · B04 (Red, 665 nm)

When vegetation is sparse and does not completely cover the ground, bare soil contributes directly to the pixel-average reflectance recorded by the sensor. Dry mineral soil reflects broadly across the visible and NIR spectrum; its specific NIR-to-red ratio varies with soil brightness and moisture, introducing noise into NDVI estimates that is uncorrelated with vegetation status. Huete (1988) introduced the soil brightness correction factor L to shift the denominator away from the singularity at the origin, where NDVI is most sensitive to soil background variation. At L = 0.5 (moderate fractional cover), SAVI reduces soil-induced variance by roughly half relative to NDVI. It is the appropriate index for early-season monitoring, arid and semi-arid landscapes, or any regime where fractional vegetation cover remains below approximately 30 %.

CIre — Chlorophyll Index Red Edge

Sentinel-2 · B07 (NIR, 783 nm) · B05 (Red Edge, 705 nm)

Rather than normalising the NIR–red-edge contrast, CIre expresses it as a simple ratio minus one, which preserves near-linear sensitivity over a wide range of chlorophyll concentrations. B05 (705 nm) sits within the chlorophyll absorption well at the base of the red edge, while B07 (783 nm) lies in the NIR plateau beyond chlorophyll absorption. The ratio ρ_NIR / ρ_RE therefore scales directly with chlorophyll content per unit leaf area, and subtracting one sets the baseline at reflectance equality. Unlike normalised-difference forms, CIre is unbounded above zero, maintaining sensitivity at high chlorophyll concentrations where NDRE begins to compress. Agricultural crops in active growth typically return values of 0.5–2.5; values above 2 indicate dense, well-nourished canopies with high photosynthetic capacity.