Written and maintained by CASRAI Editorial Board
Last updated
NDVI (normalized difference vegetation index) is a ratio of near-infrared and red reflectance, calculated as (NIR − Red) / (NIR + Red), with values from −1 to 1 once scaled. The formula is simple; the research risk is in the data. Landsat, Sentinel-2 and MODIS each store NDVI differently, so the band numbers, scale factors, fill values and compositing method must be handled correctly and cited.
Definition and scope
What is NDVI, the normalized difference vegetation index?
NDVI is a band-ratio index derived from satellite or airborne imagery. It contrasts reflectance in the near-infrared (NIR) band with reflectance in the red band, and is computed per pixel as (NIR − Red) / (NIR + Red). The USGS gives this formula for Landsat, applied to that mission’s NIR and red bands. Because the result is a normalized ratio, it ranges from −1 to 1 when stored values are scaled correctly.
NDVI is one member of a family of vegetation indices. For research use, treat it as a derived data product: its value depends on the sensor, the processing level and the compositing rules applied before the number reaches your spreadsheet. This page covers the data-handling side. For the wider field, see What Is Remote Sensing?
NDVI formula by sensor
How do you calculate NDVI (and other vegetation indices)?
Use the same ratio, but select the correct bands for the sensor. The table lists only what the cited agencies state.
| Product | Red band | NIR band | Formula as published | Resolution and storage |
|---|---|---|---|---|
| Landsat 4–7 | B3 | B4 | (B4 − B3) / (B4 + B3) | Signed 16-bit integer, −10,000 to 10,000, scale factor 0.0001 |
| Landsat 8–9 | B4 | B5 | (B5 − B4) / (B5 + B4) | Same storage; fill value −19,999 (Collection 2), saturation 20,000 |
| Sentinel-2 | B4 (10 m, about 665 nm) | B8 (10 m, about 833 nm) | (B8 − B4) / (B8 + B4) | Computed by the user from L2A (bottom-of-atmosphere) reflectance |
| MODIS MOD13Q1 v061 | Supplied as ready-made NDVI and EVI layers | Not computed by the user | 250 m, 16-day composite; int16, fill −3000, scale factor 0.0001, valid range −2000 to 10000 | |
Sources: USGS Landsat NDVI page; NASA Earthdata MOD13Q1 v061 record; Copernicus SentiWiki Sentinel-2 mission page (all read 2026-11-07). Sentinel-2 central wavelengths vary slightly between S2A, S2B and S2C (664.6 to 666.5 nm for red, 832.8 to 834.6 nm for NIR).
NDVI data: products and archives
- Landsat NDVI. The USGS derives it from Level-2 surface reflectance and delivers it on demand through ESPA, in files named with the suffix
_sr_ndvi.tif. Landsat 9 data exist only in Collection 2. - MODIS MOD13Q1. NASA LP DAAC distributes the Terra 16-day, 250 m vegetation index product under DOI 10.5067/MODIS/MOD13Q1.061.
- Sentinel-2. Level-1C is top-of-atmosphere; Sen2Cor converts it to bottom-of-atmosphere Level-2A reflectance, available systematically worldwide since 13 December 2018. Check the product documentation for the processing baseline before comparing dates.
Dataset-handling checklist
Before publishing an NDVI result, record each of these. Every item follows from the product documentation cited above.
- Processing level. Surface reflectance (Landsat Level-2, Sentinel-2 L2A) and top-of-atmosphere (Sentinel-2 L1C) are different inputs. State which you used.
- Scale factor. Landsat and MODIS NDVI are stored as integers. Multiply by 0.0001 to obtain −1 to 1.
- Fill and saturation values. Mask −19,999 and 20,000 in Landsat Collection 2 (−9,999 in Collection 1) and −3000 in MODIS before computing statistics.
- Composite method. MOD13Q1 selects the best available pixel in each 16-day window (low cloud, low view angle, highest NDVI or EVI), so a value is not a single-date observation.
- Version and collection. Record Landsat Collection 1 or 2 and the MODIS version (v061).
- Citation. NASA LP DAAC states that citation is critically important for dataset documentation and discovery. The record gives: Didan, Kamel (2021). MODIS/Terra Vegetation Indices 16-Day L3 Global 250m SIN Grid V061 [Data set]. NASA LP DAAC. https://doi.org/10.5067/MODIS/MOD13Q1.061. See data citation principle, DOI and data repository for how identifiers and archives support reuse.
Related indices: NDVI, GNDVI, EVI and SAVI
Which spectral region is most commonly used in vegetation indices such as NDVI, EVI and SAVI?
The red and near-infrared regions. NDVI uses exactly those two bands, as the formulas above show. EVI is supplied alongside NDVI in the MODIS product.
What is the difference between NDVI and GNDVI?
GNDVI follows the same normalized-difference structure but uses the green band in place of the red band, so it responds to a different part of the spectrum. Use the definition and band assignment given in the documentation of the product or paper you are reproducing, and state it in your methods.
Common misunderstandings
- Stored integers are not NDVI values. A raw Landsat or MODIS pixel of 6000 means 0.6 only after the 0.0001 scale factor is applied.
- A MODIS composite is not a daily value. It is the selected best pixel from a 16-day window.
- Sensors are not interchangeable. Bands, resolution and processing differ between Landsat, Sentinel-2 and MODIS, so index values from different sensors should not be merged without checking how they were produced.
- Class charts are not in the product documentation. The USGS page gives no table assigning value ranges to land-cover classes; any thresholds you use need their own citation.
FAQ
What is NDVI?
A normalized difference of near-infrared and red reflectance, (NIR − Red) / (NIR + Red), calculated per pixel from satellite imagery.
Where do researchers get NDVI data?
From archives such as USGS (Landsat, on demand through ESPA) and NASA LP DAAC (MODIS MOD13Q1), or by computing it from Sentinel-2 Level-2A bands.
How do I cite NDVI data?
Cite the dataset record, including version and DOI, as in the MOD13Q1 example above, and report the processing level and scale factor in your methods.
Sources
- USGS, Landsat Normalized Difference Vegetation Index: usgs.gov (read 2026-11-07)
- NASA Earthdata, MOD13Q1 v061: earthdata.nasa.gov (read 2026-11-07)
- Copernicus SentiWiki, Sentinel-2 mission: sentiwiki.copernicus.eu (read 2026-11-07)
