What is SAVI? (Soil-Adjusted Vegetation Index)
What is SAVI?
SAVI, the Soil-Adjusted Vegetation Index, is a version of NDVI that corrects for the influence of bare soil. It is the index you reach for when vegetation is sparse — young crops, recently planted fields, arid pastures — where the soil showing through the canopy drags NDVI down and makes a healthy young crop look worse than it is. If NDVI is the everyday choice, SAVI is the early-season and dryland specialist.
How it’s calculated
NDVI assumes the only thing affecting red and near-infrared reflectance is the vegetation. When the canopy is open, that assumption breaks: the soil underneath contributes to both bands, biasing the index downward. SAVI inserts a soil-brightness correction factor L into the denominator:
SAVI = ((NIR − Red) / (NIR + Red + L)) × (1 + L)On Sentinel-2 these are Band 8 (NIR) and Band 4 (Red). The L factor is typically 0.5 — a compromise for intermediate vegetation density. L = 0 reduces SAVI exactly to NDVI (no soil adjustment); L = 1 applies maximum correction for very sparse vegetation on bright soils.
A critical detail: Sentinel-2 surface-reflectance values are stored on a 0–10,000 scale. Because SAVI adds L to the bands, the scale does not cancel — L must be scaled to match, implemented as 5000 (0.5 × 10000) in DN space. Getting this wrong is the most common silent bug in SAVI implementations.
Typical value ranges
SAVI occupies the same −1 to +1 range as NDVI, but for sparse canopies it reads higher because it removes the soil-brightness penalty.
| SAVI range | Meaning | Typical for |
|---|---|---|
| 0.6 – 0.9 | Dense, healthy vegetation | Peak-season closed canopy (SAVI ≈ NDVI here) |
| 0.4 – 0.6 | Moderate vegetation | Developing crops, healthy pasture |
| 0.2 – 0.4 | Sparse but healthy | Young crops, arid rangeland — where SAVI shines vs NDVI |
| 0.0 – 0.2 | Very sparse / bare | Newly emerged crop, bare soil |
| Below 0 | Water, snow, cloud | Open water, snow cover |
The difference between SAVI and NDVI is largest when vegetation cover is low (roughly 10–40%). As the canopy closes, the two indices converge.
When to use it
A practical rule: if you can see soil in the image, prefer SAVI. Once the canopy closes, switch back to NDVI. In early growth, SAVI sits above NDVI by a margin that shrinks as the canopy closes — that gap is the soil-brightness error NDVI was making, now corrected. SAVI is especially valuable in arid rangelands, shrublands, and any field with variable soil brightness that would add noise to NDVI comparisons.
The L factor is fixed, not adaptive — you pick L = 0.5 and accept it as a compromise. If your field has very high or very low vegetation density, a fixed L is suboptimal. MSAVI2 solves this with a self-adjusting formulation.
Comparison with other indices
SAVI is NDVI with a fixed soil-brightness correction — it shines in sparse and early-growth conditions where soil drags NDVI down. MSAVI2 improves on SAVI by computing the soil correction automatically per pixel (no fixed L), making it more accurate for heterogeneous fields. NDVI remains the better choice once the canopy closes and soil is invisible. For high-biomass canopies where saturation is the problem (not soil), use EVI instead.
Try it free: Get SAVI values for your field — free, no signup. Works on any land worldwide. Check my field’s establishment →
Frequently asked questions
What is the difference between SAVI and NDVI?
SAVI is NDVI with a soil-brightness correction. NDVI assumes only vegetation affects the reflectance; SAVI inserts an adjustment factor L so that bare soil showing through a sparse canopy does not drag the index down. SAVI and NDVI give nearly identical values once the canopy closes; SAVI matters most when vegetation cover is low (10–40%).
What is the L value in SAVI?
L is the soil-brightness adjustment factor, typically set to 0.5 as a compromise for intermediate vegetation density. L = 0 turns SAVI into NDVI; L = 1 applies maximum soil correction for very sparse vegetation on bright soils. In Sentinel-2’s DN space (0–10,000), L = 0.5 is implemented as 5000.
When should I use SAVI instead of NDVI?
Use SAVI when soil is visible in the pixel: young or widely spaced crops, arid rangelands, shrublands, and any field with variable soil brightness. Once the canopy closes (mid-to-late season for most annual crops), switch back to NDVI — the soil influence is gone and SAVI offers no further benefit.
What is the difference between SAVI and MSAVI2?
SAVI uses a fixed soil-adjustment factor L = 0.5. MSAVI2 computes the adjustment automatically from the data itself with an iterative formula, so the correction adapts to the actual vegetation density in each pixel. For heterogeneous fields or very sparse vegetation, MSAVI2 is the more accurate choice.