Agricultural land productivity scoring for rural valuation
Multi-year NDVI and EVI time series from Sentinel-2 and Landsat reveal the structural productive capacity of agricultural parcels, separating genuine fertility from a single good season and flagging drainage, compaction or salinity before a transaction closes.
Sensors
- Sentinel-2 MSI: 10 m resolution in the red and near-infrared bands used for NDVI; 20 m in the red-edge bands (B5, B6, B7) sensitive to chlorophyll stress before visual symptoms appear. Five-day revisit at the equator (two-satellite constellation) gives roughly 20–30 cloud-free observations per growing season in temperate climates, fewer in persistently cloudy tropics. Archive from 2015.
- Landsat 8/9 OLI: 30 m multispectral resolution; 16-day revisit per satellite, eight days when both are operational. The combined Landsat archive extends to 1972, making it the only freely available source for decadal productivity baselines. OLI's coastal/aerosol band aids atmospheric correction over irrigated fields with high aerosol loads.
- Planet SuperDove: 3–5 m resolution with daily revisit over most land areas. Adds eight spectral bands including two red-edge channels. Useful for resolving sub-parcel productivity variation (headland compaction, wet corners) that 10 m imagery blurs, but archive depth is shorter and access requires a commercial licence.
- NASA SMAP: L-band passive radiometer measuring surface soil moisture at approximately 36 km resolution with two- to three-day global revisit. Coarse for parcel-level work, but persistent low-moisture or high-salinity signals in SMAP time series corroborate spectral stress indicators and help distinguish drought response from structural soil problems.
Why a single season's yield figure misleads a buyer
A vendor presenting three years of strong yield data is presenting three years of weather. A wet spring and a favourable summer can make chronically compacted or saline land look productive. The satellite record does not care about the weather narrative; it shows what photosynthetic activity actually occurred, field by field, week by week, across every season in the archive.
The key distinction is between transient vigour and structural vigour. A parcel with a high mean NDVI but high interannual variance is weather-dependent. One with a consistently lower mean but low variance is structurally constrained. Neither fact appears in a vendor's yield declaration, and neither requires a physical soil survey to detect at the screening stage.
What a floating-point index reveals about the ground beneath it
NDVI (Normalised Difference Vegetation Index) is the ratio of near-infrared reflectance minus red reflectance to their sum. Dense, healthy canopies absorb red light strongly and reflect near-infrared strongly, pushing NDVI toward 1.0. Stressed, sparse or absent vegetation sits below 0.3. EVI (Enhanced Vegetation Index) adds a blue-band correction that reduces soil background noise and atmospheric aerosol effects, making it more reliable over irrigated fields with bare-soil periods.
Persistent low-vigour zones within a parcel are the diagnostic signal. A wet corner that reads low NDVI every spring points to poor drainage. A strip along a headland that never recovers to field-average values points to compaction from machinery turning. A diffuse low-vigour patch that tracks across seasons without moving suggests salinity or subsoil hardpan. None of these features are visible in aggregate yield figures, and some are invisible on a single-date image. They emerge only in time-stack analysis.
The red-edge bands on Sentinel-2 (B5 at 705 nm, B6 at 740 nm, B7 at 783 nm) are particularly useful here. Chlorophyll stress reduces red-edge reflectance before the crop visibly yellows, giving an earlier warning of sub-optimal conditions than broadband NDVI alone.
Building the productivity score: archive depth and statistical method
A credible productivity score requires at minimum five growing seasons, and ten or more to separate structural soil effects from multi-year climate cycles such as the North Atlantic Oscillation or ENSO. Landsat's archive to 1984 (with consistent OLI data from 2013) is the practical backbone. Sentinel-2 adds resolution from 2015. Together they allow a 10-year Sentinel-era score to be cross-validated against a longer Landsat baseline.
The standard approach is to extract peak-season NDVI or EVI composites for each year, compute per-pixel mean and coefficient of variation across the time stack, then aggregate to parcel level using cadastral boundaries. A high mean with low coefficient of variation is the signature of reliably productive land. Parcels scoring in the bottom quartile of mean vigour for their agro-ecological zone, or showing high spatial heterogeneity within the boundary, warrant further investigation before a price is agreed.
Cloud cover is the honest constraint. In maritime climates, achieving ten usable cloud-free observations per growing season from Sentinel-2 alone may require compositing across a two- to three-week window, which smooths phenological timing. Combining Sentinel-2 and Landsat through methods such as the Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) partially addresses this, though the fusion product carries its own uncertainty budget.
From pixel statistics to a number a valuer can use
The output of spectral time-series analysis is not, by itself, a land value. It is an independent productivity indicator that sits alongside soil survey data, water rights documentation and comparable transaction records. Its practical role in a valuation is to flag anomalies that justify a price reduction, commission a targeted physical survey, or walk away from a transaction entirely.
A parcel scoring in the top decile for its soil class and climate zone across a ten-year archive commands a different conversation than one sitting in the bottom third with high interannual variance. The satellite score does not replace the agronomist, but it arrives before the agronomist is hired and costs a fraction of a full soil survey. For portfolio acquisitions spanning hundreds of parcels, it is the only practical way to prioritise where physical due diligence is worth spending.
Satellize runs this class of analysis operationally. The Tonga crop-estimation programme, which applies NDVI time-series methods to smallholder parcels across the archipelago, uses the same spectral foundations, adapted for tropical phenology and the specific cloud climatology of the South Pacific.
Honest limits: what the method cannot tell you
Spatial resolution sets a floor on sub-parcel detection. At 10 m, a two-pixel-wide drainage problem is detectable; a single poorly draining furrow is not. Planet SuperDove at 3–5 m resolves finer features, but at commercial data cost and with a shorter archive.
The method scores photosynthetic output, not yield in tonnes per hectare. A high-vigour score on a parcel growing a low-value crop does not imply high revenue potential. Crop type identification requires additional classification steps and is not always unambiguous from spectral data alone, particularly where rotations change annually.
SMAP soil moisture at 36 km resolution cannot localise a salinity patch to a specific field. It can flag a landscape-level soil moisture anomaly that warrants closer investigation. Parcel-level salinity mapping requires either higher-resolution radar (Sentinel-1 SAR can help in some conditions) or ground-truth electromagnetic induction surveys.
Finally, the archive reflects what happened under the management regime of the current owner. A poorly managed parcel with structurally good soils will score below its potential. The score is a record of realised productivity, not a guarantee of future performance under different management.
Typical figures
| Spatial resolution (primary) | 10 m (Sentinel-2 red/NIR); 30 m (Landsat OLI); 3–5 m (Planet SuperDove, commercial) |
| Revisit cadence | 5 days (Sentinel-2 dual satellite); 8 days (Landsat 8+9 combined); daily (Planet SuperDove) |
| Archive depth | Landsat to 1972 (consistent OLI from 2013); Sentinel-2 from 2015; Planet from ~2016 |
| Key spectral bands | Red (665 nm), NIR (842 nm), Red-edge (705/740/783 nm) on Sentinel-2; Red (655 nm), NIR (865 nm) on Landsat OLI |
| Soil moisture ancillary | NASA SMAP L-band passive, ~36 km resolution, 2–3 day global revisit, from 2015 |
| Minimum detectable low-vigour zone | ~100 m² at 10 m resolution (single pixel cluster); sub-field features below ~30 m require SuperDove |
| Recommended time-series depth | Minimum 5 growing seasons; 10+ preferred to separate climate variability from structural soil effects |
| Cloud cover constraint | 10–30 usable observations per season typical in temperate climates; fewer in humid tropics without multi-sensor compositing |
| Delivery formats | GeoTIFF raster layers, parcel-level CSV/GeoJSON score tables, PDF valuation-support report |
Analytics Satellize can run
| Multi-year NDVI/EVI productivity score per parcel | Per-pixel time-stack mean and coefficient of variation from Sentinel-2 and Landsat composites, aggregated to cadastral parcel boundaries | GeoJSON parcel layer with mean vigour score, interannual variance, and decile rank within agro-ecological zone |
| Persistent low-vigour zone delineation | Spatial clustering of pixels below zone-specific NDVI threshold across 80% or more of observed seasons | Shapefile of flagged sub-parcel zones with probable cause classification (drainage, compaction, salinity candidate) |
| Phenological consistency profile | Time-series harmonic analysis (Fourier decomposition) to characterise growing-season timing and peak amplitude per parcel per year | Annual phenology summary table; anomaly flags for seasons with suppressed peak or delayed green-up |
| Red-edge chlorophyll stress index | Sentinel-2 red-edge band ratios (CIre: B7/B5 minus 1) to detect early-stage stress not yet visible in broadband NDVI | Seasonal raster stack and parcel-level stress frequency score |
| Soil moisture anomaly overlay | SMAP time-series anomaly detection (deviation from multi-year seasonal mean) co-registered to parcel boundaries | Ancillary data layer flagging parcels within landscape-level persistent moisture deficit or excess zones |
| Portfolio screening priority ranking | Composite scoring combining mean vigour, variance, low-vigour zone area fraction and soil moisture anomaly into a weighted index | Ranked parcel list with recommended tiers: proceed, commission targeted survey, or seek price renegotiation |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.