Indices & Variables Reference#
Complete catalog of the variables supported by ndvi2gif, organized by the bands they need and therefore by the sensors that can compute them.
Note
Many of the reference DOIs listed for the spectral indices below were sourced from the Awesome Spectral Indices catalogue (Montero et al., A standardized catalogue of spectral indices to advance the use of remote sensing in Earth system research, Scientific Data, 2023). We gratefully acknowledge this open community resource.
Tip
This page is kept in sync with the code, but the authoritative list for a sensor is always
NdviSeasonality(sat=...).get_available_indices(). Asking for an index a sensor cannot
compute raises a ValueError listing the ones it can.
Overview#
Group |
Count |
S2 |
Landsat |
MODIS |
S3 |
S1 |
ERA5 |
CHIRPS |
|---|---|---|---|---|---|---|---|---|
13 |
✓ |
✓ |
✓ |
✓ |
||||
13 |
✓ |
✓ |
✓ |
|||||
2 |
✓ |
|
||||||
9 |
✓ |
|||||||
10 |
✓ |
|||||||
9 |
✓ |
6 |
6 |
|||||
7 |
✓ |
|||||||
47 |
✓ |
|||||||
1 |
✓ |
|||||||
Total |
111 |
44 |
34 |
33 |
23 |
7 |
47 |
1 |
Usage is the same for all of them:
from ndvi2gif import NdviSeasonality
ns = NdviSeasonality(roi=roi, sat='S2', index='evi', key='median',
periods=12, start_year=2023, end_year=2023)
Units: surface reflectance on every optical sensor#
Formulas below use the standardized band names (Blue, Green, Red, Nir, Swir1,
Swir2, Red_Edge1-3) and assume surface reflectance in 0–1, which is what every
optical collection holds: Landsat is rescaled by scale_OLI / scale_ETM, and Sentinel-2
and MODIS — which store reflectance as integers multiplied by 10000 — are rescaled when the
collection is built.
Warning
Before v1.6.0 Sentinel-2 and MODIS were left as integers × 10000. Pure band ratios
(NDVI, NDWI, MNDWI, NDMI…) were not affected, but every index with a constant, a sum or an
inverse was: savi, evi, lai, avi, wi2015, cri1, cri2, fai, mcari, ireci,
and the magnitude of awei/aweinsh. Results computed with those indices on Sentinel-2 or
MODIS with earlier versions should be recomputed. See the changelog.
Sentinel-3 is the exception: the OLCI collection available in Earth Engine holds top-of-atmosphere radiances, not reflectance, so its indices are relative indicators rather than calibrated quantities.
Visible and NIR indices#
13 indices · Sentinel-2, Landsat, MODIS, Sentinel-3
They only need visible and near-infrared bands, so they are available on every optical sensor.
|
Name |
Formula |
Reference |
|---|---|---|---|
|
Normalized Difference Vegetation Index |
|
|
|
Enhanced Vegetation Index |
|
|
|
Soil Adjusted Vegetation Index |
|
|
|
Green NDVI |
|
|
|
Wide Dynamic Range Vegetation Index |
|
|
|
Advanced Vegetation Index |
|
Bannari, Asalhi & Teillet (2002), IGARSS |
|
Leaf Area Index (empirical) |
|
|
|
Chlorophyll Index Green |
|
|
|
Carotenoid Reflectance Index 1 |
|
|
|
Carotenoid Reflectance Index 2 |
|
|
|
Photochemical Reflectance Index (broadband) |
|
|
|
Normalized Difference Water Index |
|
|
|
Vegetation Condition Index (simplified) |
|
Notes on how some of them are implemented:
saviuses L = 0.428 rather than the 0.5 Huete suggested as a general value.laiis an empirical regression on EVI calibrated for agricultural crops; treat it as a relative indicator elsewhere.cri2uses the red band where the original index uses 700 nm, so that it can be computed on sensors without a red edge band.priis a broadband proxy: the original index contrasts two narrow bands at 531 and 570 nm that no sensor here has.vcinormally scales NDVI between its long-term minimum and maximum for each pixel; here fixed values (0.1 and 0.8) are used, so it is a rescaled NDVI. For the real VCI, compute the per-pixel extremes from a longNdviSeasonalityseries.
SWIR indices#
13 indices · Sentinel-2, Landsat, MODIS (not Sentinel-3, whose OLCI sensor has no SWIR)
|
Name |
Formula |
Reference |
|---|---|---|---|
|
Modified NDWI |
|
|
|
Automated Water Extraction Index (shadow) |
|
|
|
Automated Water Extraction Index (no shadow) |
|
|
|
Water Index 2015 |
|
|
|
Normalized Difference Moisture Index |
|
Hardisky et al. (1983); Wilson & Sader (2002) |
|
Moisture Stress Index |
|
|
|
Normalized Multi-band Drought Index |
|
|
|
Normalized Burn Ratio |
|
Key & Benson (2006), FIREMON, USDA RMRS-GTR-164-CD |
|
Normalized Burn Ratio (alias of |
|
as |
|
Normalized Difference Built-up Index |
|
|
|
Normalized Difference Snow Index |
|
|
|
Normalized Difference Tillage Index |
|
Van Deventer et al. (1997), PE&RS 63(1) |
|
Floating Algae Index |
|
Water indices (
mndwi,ndwi,awei,aweinsh,wi2015) are positive over water and negative over land.HydroperiodAnalyzeruses them with a default threshold of 0.ndtiis the tillage index (crop residue on soil), not the turbidity index that shares its acronym. For water turbidity seeturbidityandspmon Sentinel-3.mndwiandndsihave the same formula; the name only says what it is used for.faiuses SWIR2 as the baseline band, with a wavelength factorfper sensor (0.1109 for Sentinel-2, 0.1359 for Landsat 8/9, 0.1438 for MODIS). Hu (2009) used the 1240 nm MODIS band, which is not among the standardized bands.
Thermal indices#
2 indices · Landsat (both), MODIS (lst only)
|
Name |
Implementation |
Sensors |
|---|---|---|---|
|
Land Surface Temperature (°C) |
Landsat Collection 2 |
Landsat, MODIS |
|
Urban Thermal Field Variance Index (simplified) |
|
Landsat |
utfvi is a simplified empirical version, not the published UTFVI (which normalizes LST by
its mean over the scene). Both were offered on Sentinel-2 and Sentinel-3 before v1.6.0, where
they returned empty images; they are now only accepted on sensors with thermal bands.
Sentinel-2 red edge indices#
9 indices · Sentinel-2 only (bands B5, B6, B7 → Red_Edge1, Red_Edge2, Red_Edge3)
|
Name |
Formula |
Reference |
|---|---|---|---|
|
Normalized Difference Red Edge |
|
|
|
Chlorophyll Index Red Edge |
|
|
|
Modified Chlorophyll Absorption Ratio Index |
|
|
|
Inverted Red Edge Chlorophyll Index |
|
|
|
MERIS Terrestrial Chlorophyll Index |
|
|
|
Plant Senescence Reflectance Index |
|
|
|
Red Edge Inflection Point (nm) |
|
Guyot & Baret (1988) |
|
Sentinel-2 Red Edge Position (nm) |
|
|
|
Normalized Difference Chlorophyll Index (water) |
|
ndci is meant for chlorophyll-a in turbid and productive waters; the rest are vegetation
chlorophyll and senescence indices.
Sentinel-3 OLCI water quality indices#
10 indices · Sentinel-3 only
OLCI band names are standardized as Blue (Oa02, 412.5 nm), Blue2 (Oa03, 442.5 nm),
Blue_Green (Oa04, 490 nm), Green (Oa05, 510 nm), Red (Oa07, 620 nm), Red2 (Oa08,
665 nm), Red3 (Oa09, 673.75 nm), Red_Edge1 (Oa10, 681.25 nm), Red_Edge2 (Oa11,
708.75 nm) and Nir (Oa12, 753.75 nm). The collection holds top-of-atmosphere radiances,
so these are relative indicators for comparing places and dates, not calibrated
concentrations.
|
Name |
Formula |
Reference |
|---|---|---|---|
|
OLCI chlorophyll index |
|
|
|
Fluorescence Line Height |
|
|
|
Red edge position (nm) |
|
|
|
Floating algae (OLCI) |
|
|
|
Turbidity |
|
|
|
Suspended Particulate Matter |
|
|
|
Trophic State (spectral proxy) |
|
|
|
Coloured Dissolved Organic Matter |
|
|
|
Diffuse attenuation at 490 nm (proxy) |
|
Mueller (2000), SeaWiFS Postlaunch Tech. Report |
|
Water-leaving reflectance (proxy) |
|
floating_algae and tsi failed on every Sentinel-3 image before v1.6.0 (they looked up a
band that does not exist); they work from v1.6.0 on.
Raw Reflectance Bands#
9 bands · Sentinel-2 (9), Landsat (6), MODIS (6)
Spectral indices are ratios, so they cancel out multiplicative changes in brightness: a pixel can keep exactly the same NDVI while its reflectance drifts. When the question is about the radiometry itself rather than about vegetation, the bands are available as regular index names.
index='blue' # or 'green', 'red', 'nir', 'swir1', 'swir2'
index='red_edge1' # or 'red_edge2', 'red_edge3' — Sentinel-2 only
Units: surface reflectance, 0 to 1, on every sensor, so the same numeric threshold means the same thing whichever sensor produced it.
Not available on Sentinel-3, whose bands are top-of-atmosphere radiances
with a different band set, and not on Sentinel-1, which has vv and vh.
Typical use — radiometric stability: combined with the dispersion reducers, the bands map how invariant each pixel is over a time series, which is how pseudo-invariant features (PIFs) are chosen for relative radiometric normalization:
# Within-year dispersion of SWIR1, one composite per year
stability = NdviSeasonality(
roi=roi, start_year=2018, end_year=2025,
periods=1, sat='S2', index='swir1', key='std'
).get_year_composite()
# How many observations each pixel is based on
n_obs = NdviSeasonality(
roi=roi, start_year=2018, end_year=2025,
periods=1, sat='S2', index='swir1', key='count'
).get_year_composite()
Name |
Band |
S2 |
Landsat |
MODIS |
|---|---|---|---|---|
|
Blue (~490 nm) |
✓ |
✓ |
✓ |
|
Green (~560 nm) |
✓ |
✓ |
✓ |
|
Red (~665 nm) |
✓ |
✓ |
✓ |
|
NIR (~840 nm) |
✓ |
✓ |
✓ |
|
SWIR 1 (~1610 nm) |
✓ |
✓ |
✓ |
|
SWIR 2 (~2190 nm) |
✓ |
✓ |
✓ |
|
Red Edge 1 (~705 nm) |
✓ |
✗ |
✗ |
|
Red Edge 2 (~740 nm) |
✓ |
✗ |
✗ |
|
Red Edge 3 (~783 nm) |
✓ |
✗ |
✗ |
SAR Indices#
7 variables · Sentinel-1 (VV + VH, IW mode)
vv and vh are returned in dB, the usual scale for backscatter. The polarimetric
indices are defined on linear power, so each of them converts VV and VH to linear
(10^(dB/10)) before applying its formula. The default preprocessing (use_sar_ard=True)
applies terrain flattening and speckle filtering on linear power too.
|
Name |
Formula |
Reference |
|---|---|---|---|
|
VV backscatter (dB) |
|
— |
|
VH backscatter (dB) |
|
— |
|
Dual-pol Radar Vegetation Index |
|
|
|
VV/VH ratio |
|
|
|
Radar Forest Degradation Index (dual-pol) |
|
|
|
Modified Dual-Pol SAR Vegetation Index (DPSVIm) |
|
|
|
Vegetation Scattering Diversity Index — experimental |
|
none found |
rviranges roughly from 0 (bare soil, water) to 1 (dense vegetation).vv_vh_ratioin linear power is equivalent to the differenceVV − VHin dB.rfdiwas defined by Mitchard et al. with HH/HV; Sentinel-1 carries VV/VH, so the common dual-pol adaptation is used.dpsviis the modified form by dos Santos et al. (2021). Periasamy’s original DPSVI (2018) multiplies by a term that depends on the maximum VV of the whole scene, so it is not a per-pixel index and changes with the extent.vsdihas no known published definition. It is kept for compatibility, but prefer the documented indices for anything that has to be reported.
Warning
Before v1.6.0 the Sentinel-1 collection (served in dB by Earth Engine) was processed and
indexed as if it were linear power: terrain correction and speckle filtering ran on dB values,
and rvi, vv_vh_ratio, rfdi and dpsvi were computed on dB. Their values change from
v1.6.0 (e.g. rvi 2.42 → 0.70 over the same scene); vv and vh change on sloping terrain,
where the old terrain correction distorted them.
The normalize=True option of the SAR index methods returns a z-score (mean 0, standard
deviation 1) instead of the raw value.
ERA5-Land Climate Variables#
47 variables · ERA5-Land daily aggregates (ECMWF reanalysis, 1950–present, ~11 km)
Use the reducer that matches the variable: key='mean' for states (temperature, soil
moisture, pressure), key='sum' for fluxes accumulated over the period (precipitation,
evaporation, runoff, radiation), key='min'/'max' for extremes.
Temperature (24)#
Four variables, each as daily mean, daily minimum and daily maximum, in Kelvin or converted
to °C (_celsius suffix):
Variable |
Kelvin |
°C |
|---|---|---|
Air temperature at 2 m |
|
|
Dewpoint temperature at 2 m |
|
|
Skin (surface) temperature |
|
|
Soil temperature, layer 1 (0–7 cm) |
|
|
Precipitation & water balance (10)#
Each in meters of water, or in L/m² (= mm) with the _lm2 suffix:
Variable |
meters |
L/m² |
|---|---|---|
Total precipitation |
|
|
Total evaporation (negative = upward flux) |
|
|
Potential evaporation |
|
|
Total runoff (surface + subsurface) |
|
|
Surface runoff |
|
|
Soil moisture (4)#
Volumetric water content (m³/m³) by depth: volumetric_soil_water_layer_1 (0–7 cm),
volumetric_soil_water_layer_2 (7–28 cm), volumetric_soil_water_layer_3 (28–100 cm) and
volumetric_soil_water_layer_4 (100–289 cm).
Radiation (3)#
In J/m²: surface_solar_radiation_downwards_sum, surface_net_solar_radiation_sum and
surface_latent_heat_flux_sum.
Wind & pressure (3)#
u_component_of_wind_10m (east–west) and v_component_of_wind_10m (north–south), in m/s —
wind speed is sqrt(u² + v²) — and surface_pressure, in Pa.
Snow (3)#
snow_depth_water_equivalent (m of water), and daily snowfall as snowfall_sum (m of water)
or snowfall_sum_lm2 (L/m²).
CHIRPS Precipitation#
1 variable · CHIRPS daily (1981–present, ~5.5 km, 50°S–50°N)
sat='CHIRPS', index='precipitation'
Satellite and station blended daily precipitation in mm/day. Use key='sum' for
monthly or seasonal totals. Finer grid than ERA5-Land and calibrated against stations, but
limited to 50°S–50°N.
Usage Examples#
Optical#
from ndvi2gif import NdviSeasonality
ndvi = NdviSeasonality(roi=roi, sat='S2', index='ndvi')
evi = NdviSeasonality(roi=roi, sat='MODIS', index='evi', key='mean')
chlorophyll = NdviSeasonality(roi=roi, sat='S2', index='ndre') # red edge, S2 only
water = NdviSeasonality(roi=roi, sat='Landsat', index='mndwi')
surface_temp = NdviSeasonality(roi=roi, sat='Landsat', index='lst', key='max')
SAR#
# All-weather vegetation monitoring; the median also dampens residual speckle
rvi = NdviSeasonality(roi=roi, sat='S1', index='rvi', key='median', periods=12)
Climate#
temperature = NdviSeasonality(roi=roi, sat='ERA5', index='temperature_2m_celsius',
start_year=1980, end_year=2023, key='mean')
rain_totals = NdviSeasonality(roi=roi, sat='CHIRPS', index='precipitation', key='sum')
root_zone = NdviSeasonality(roi=roi, sat='ERA5',
index='volumetric_soil_water_layer_3', key='mean')
Several sensors in one classification#
Indices from different sensors can be combined in a single feature stack with
LandCoverClassifier (v1.6.0):
from ndvi2gif import LandCoverClassifier
s2 = NdviSeasonality(roi=roi, sat='S2', periods=4, start_year=2024, end_year=2024)
s1 = NdviSeasonality(roi=roi, sat='S1', periods=12, start_year=2024, end_year=2024)
clf = LandCoverClassifier([s2, s1])
clf.create_feature_stack(indices={'S2': ['ndvi', 'ndmi', 'ndre'], 'S1': ['vv', 'vh', 'rvi']})
References#
Climate datasets:
Muñoz-Sabater, J. (2019). ERA5-Land hourly data from 1981 to present. Copernicus Climate Change Service (C3S) Climate Data Store. DOI:10.24381/cds.e2161bac
Funk, C. et al. (2015). The climate hazards infrared precipitation with stations — a new environmental record for monitoring extremes. Scientific Data, 2, 150066. DOI:10.1038/sdata.2015.66
The references of each index are linked in its table.
See Also#
Datasets Reference - Detailed platform documentation
API Reference - Complete class documentation
Tutorials - Step-by-step index usage