Sentinel-2 data and vegetation indices

Sentinel-2 Data and Vegetation Indices


sat1

Learning outcomes

  • Learn the principle of the spectral index calculation
  • Familiarize with different vegetation spectral indices
  • Get to know the Sentinel-2 spectral bands
  • Learn about the index database and tools available to calculate indices
  • Learn how the spectral indices are used in different applications: agriculture, forestry, vineyards, natural hazards: fires, windstorms

BoK Concepts

  • [IP3-1-2] Spectral indices
  • [[PP1-3-4] Spectral Signature of Vegetation, Water, Soil
  • [PP1-1] EM radiation

Module 1


Introduction to spectral indices

Spectral Indices Background

Spectral indices are used to enhance particular land surface features or properties, e.g. vegetation, soil, water.

There are developed based on the spectral properties of the object of interest.


sat1

Spectral Indices Background

Spectral signatures of clean/turbid water (in reflectance)


Source: Ma, M., et al. (2007) Change in area of Ebinur Lake during the 1998–2005 period. International Journal of Remote Sensing, 28(24), 5523-5533.

Spectral Indices Background

Spectral signatures of snow and clouds (in reflectance)


Source: Dong, C. (2018) Remote sensing, hydrological modelling and in situ observations in snow cover research: A review. Journal of Hydrology, 561, 573-583.

Spectral Vegetation Principles

Knowledge of the leaf cell, plant structure, status, condition and its spectral properties is essential to perform the vegetation analysis using remote sensed data.


sat1

The structure of a chloroplast, and its location within a plant cell and leaf.

Source: Levetin and McMahon (2008)

Spectral Vegetation Signatures


sat1

Dominant factor controlling leaf reflectance.

Vegetation spectra correspond to bundles of leaves and steams of Spartina alternifora, a wetland perennial grass from Kokaly et al. (2017), Soil spectrum from Clark (1999). Figure adapted by Denis, A. (2018) from Kokaly et al. (1998), Bowker et al. (1985), Curran (1989) and Thenkabail et al. (2013)

Spectral Vegetation Signatures

    In green healthy plant
  • chlorophyll absorbs large proportion of red and blue spectrum for photosynthesis and strongly reflects in green
  • strong reflectance in near infrared (NIR) due to leaf structure and condition
  • lower reflectance in shortwave infrared (SWIR) influenced by water content, which absorbs infrared energy.

sat1

Spectral Vegetation Signatures

Monitoring vegetation disease


sat1
Source: Ashraf, M. A., Maah, M. J., & Yusoff, I., (2011) Introduction to remote sensing of biomass. In Biomass and remote sensing of biomass, 129-170. IntechOpen.

Vegetation indices

Spectral indices dedicated to vegetation analysis are developed based on the principle that the healthy vegetation reflects strongly in the near-infrared (NIR) spectrum while absorbing strongly in the visible red.

sat1
Source: Wu Ch-D., et al. (2014)

Rational for Vegetation Indices

To explore and highlight the unique spectral signatures of vegetation, which allows to delineate it from the other earth objects.

To delineate the subtle changes in the spectral signatures of vegetation caused by changes in plant vigour and health that cannot be distinguished by human eye.

Spectral vegetation indices have been found to be related to various biophysical parameters, i.e. Leaf Area Index (LAI), percent vegetation cover, fraction of absorbed photosynthetically active radiation (fAPAR), photosynthetic capacity, and carbon dioxide fluxes.

How to calculate spectral indices?

There are calculated using a mathematical equation applied to two or more wavelengths across the optical spectrum.

Spectral indices varied from a simple spectral rationing to more complicated combination of multispectral bands.

They are used to combine the multiple spectral bands into a single image, which highlights particular land surface features or properties.

Sentinel-2 Spectral Bands


sat1
Sentinel-2 spectral characteristics, © ESA

Simple band ratio

A common practice in remote sensing is the use of band ratio to eliminate various effects such as:

  • irradiance (topography)
  • transmittance (atmospheric effects)
  • Example of simple ratio index:


    RVI - Ratio Vegetation Index

    RVI = (NIR / Red)
    For Sentinel-2 the formula is: B8 / B4, where: B8 = 842 nm, B4 = 665 nm

    Simple band ratio

    Examples:

    PSSRa - Pigment Specific Simple Ratio (chlorophyll index) algorithm (Blackburn, 1998) investigates the potential of a range of spectral approaches for quantifying pigments at the scale of the whole plant canopy.


    PSSRa = (NIR) / (Red)
    For Sentinel-2 the formula is: B7 / B4, where: B7 = 783 nm, B4 = 665 nm

    Combination of multispectral bands

    NDVI - Normalized Difference Vegetation Index

    Effective for quantifying green vegetation. Positively correlated with vegetation greenness.


    NDVI = (NIR – Red) / (NIR + R)
    For Sentinel-2 the formula is: (B8 - B4) / (B8 + B4)
    where: B8 = 842 nm, B4 = 665 nm

    NDVI range value is -1 to 1

    sat1
    Source: Wu Ch-D., et al. (2014)

    Examples of the NDVI related indices

    TNDVI - Transformed Normalized Difference Vegetation Index, indicates a relation between the amount of green biomass that is found in a pixel. It has always positive values and the variances of the ratio are proportional to mean values (Senseman et al. 1996).

    TNDVI = (sqrt(NDVI + 0.5))


    NDI45M - Normalized Difference Index (Delegido et al. 2011), is more linear, with less saturation at higher values than the NDVI

    NDI45 = (NIR – R) / (NIR + R)

    For Sentinel-2 the formula is: (B5 - B4) / (B5 + B4), where: B5 = 705 nm, B4 = 665 nm

    Examples of the NDVI related indices


    sat1
    NDVI and TNDVI calculated based on Sentinel-2 data © IGIK

    Indices with green wavelength

    Green spectral band was found to be more efficient than the red spectral band to discriminate nitrogen.

    GNDVI - Green Normalized Difference Vegetation Index (Gitelson et al. 1996) is more sensitive than NDVI to different concentration rates of chlorophyll, which is highly correlated at nitrogen.

    GNDVI = NIR - Green / NIR + Green

    Indices with green wavelength


    MCARI - Modified Chlorophyll Absorption Ratio Index - is responsive to both leaf chlorophyll concentrations and ground reflectance.

    MCARI = ((Red2 - Red1) - 0.2 * (Red2 - Green)) * (Red2 / Red1)


    For Sentinel-2 the formula is:
    [(B5 - B4) - 0.2 * (B5 - B3)] * (B5 - B4),
    where: B5 = 705 nm, B4 = 665 nm, B3 = 560 nm

    Indices with red-edge wavelengths

    The 'red edge' is the name given to the abrupt reflectance change in the 680 ± 740 nm region of vegetation spectra that is caused by the combined effects of strong chlorophyll absorption and leaf internal scattering.

    Red edge, as the inflection point of the strong red absorption to near infrared reflectance, includes the information of chlorophyll content, nitrogen and growth status.

    Indices with red-edge wavelengths

    S2REP - Sentinel-2 Red-Edge Position Index is sensitive to both crop (chlorophyll content) N and growth status. Generally, the higher the S2REP value, the higher is the chlorophyll content (Guyot and Baret, 1988). The Sentinel-2 Band 6 (740 nm) measures the reflectance situated at the top of the linear part of the 'red edge' slope.

    S2REP = 705 + 35 * ((Red1 + NIR)/2) - Red2 / (Red3 - Red2))


    For Sentinel-2 the formula is:
    705 + 35 * ((B4 + B7)/2) - B5 / (B6 - B5)),
    where: B7 = 783 nm, B6 = 740 nm, B5 = 705 nm, B4 = 665 nm

    Indices with red-edge wavelengths

    IRECI - Inverted Red-Edge Chlorophyll Index algorithm incorporates the reflectance in four bands to estimate canopy chlorophyll content (Guyot and Baret, 1988; Clevers et al., 2000).

    IRECI = (NIR - Red1) / (Red2 / Red3)


    For Sentinel-2 the formula is:
    (B7 - B4) / (B5 / B6),
    where: B7 = 783 nm, B6 = 740 nm, B5 = 705 nm,
    B4 = 665 nm

    Vegetation Indices Issues

  • Atmospheric interference – best practice to perform the atmospheric correction.

  • Empirically derived NDVI products have been shown to be unstable, varying with soil color, soil moisture, and saturation effects from high density vegetation.

  • Soil brightness variations complicate the vegetation indices response especially when the vegetation cover is low – it is necessary to remove the effect of soil brightness and isolate reflectance changes from vegetation.
  • Indices dedicated to reduction of soil noise

  • Reduce soil noise at the cost of decreasing the dynamic range of the index.

  • These are slightly less sensitive to changes in vegetation cover than NDVI at low levels of vegetation cover.

  • These are more sensitive to atmospheric variations than NDVI.
  • Indices dedicated to reduction of soil noise

    SAVI - Soil Adjusted Vegetation Index - use a transformation technique that minimizes soil brightness influences from spectral vegetation indices involving red and near-infrared (NIR) wavelengths.

    SAVI = (1+L) * (NIR - Red) / (NIR + Red + L)


    For Sentinel-2 the formula is:
    (B08 - B04) / (B08 + B04 + L) * (1.0 + L); L = 0.428
    where: L is a soil brightness correction factor ranging from 0 to 1
    L = 1 low vegetation cover, L = 0 high vegetation cover, L = 0.5 intermediate vegetation cover.

    Indices dedicated to reduction of soil noise

    Sentinel-2 - SAVI

    https://custom-scripts.sentinel-hub.com/custom-scripts/sentinel-2/savi/
    sat1

    Vegetation Moisture Sensitive Indices

    The moisture sensitive indices are calculated using SWIR and NIR.

    The SWIR bands are sensitive to vegetation water content and spongy mesophyll structure in the vegetation canopy. The NIR reflectance is affected by leaf internal structure and leaf dry matter content but not by water content.

    There are used to assess the vegetation moisture decline that is particularly useful for drought monitoring.

    There are widely applied in agricultural and ecological applications including surface water body characteristics, vegetation water stress, soil water content assessment and wetlands monitoring.

    Vegetation Moisture Sensitive Indices

    NDMI - Normalized Difference Moisture Index (Gao, 1995).

    sat1
    .

    Index Database


    ! 249 spectral indices available !
    https://www.indexdatabase.de/db/is.php?sensor_id=96.

    sat1

    Sentinel-2 Indices

    Large collection of Sentinel-2 spectral indices and Java scripts available on

    https://custom-scripts.sentinel-hub.com/custom-scripts/sentinel-2/indexdb/


    sat1

    Sentinel-2 Indices


    sat1

    Module 2


    Application of vegetation spectral indicess


    Agriculture

    Analysing the crop health and vigour

    Variation of maize condition depending on drought conditions. The condition of maize, expressed by the vegetation index - NDVI, calculated using Sentinel-2 data compared with the maps of agricultural droughts for the same decades of the year, calculated on the basis of NOAA AVHRR data.


    sat1
    © IGIK

    Agriculture

    Using multi-temporal vegetation indices can help management of individual fields during the growing season. They can also be used for within-field monitoring of growth and application of fertilizers.

    sat1
    © Geografiska informationsbyrån

    Forestry

    Analysing the crop health and vigour

    Forest health and condition

    Combination of different spectral Spectral-2 indices (NDVI, NDMI and PSSRa) allows to assess and monitor forest health condition over time.


    sat1
    © IGIK


    Forestry

    Multi-temporal Sentinel-2 vegetation indices are applied to monitor bark beetle outbreak in Białowieża National Park - Poland

    DSWI - Disease Water Stress Index sensitive to stress due to water shortage and plant damage

    DSWI = (NIR-GREEN) / (SWIR1 + RED)


    sat1
    Source: Hoscilo et al. (2016)


    Vineyards

    Monitoring condition of vineyards using NDVI

    (April-September 2019)

    Vineyards Srebrna Góra, Poland


    Sentinel-2
    sat1

    sat1 © IGIK & VINUM 4.0 Sp z o.o

    Vineyards

    Monitoring condition of vineyards using Sentinel-2 moisture index (NDMI)

    (April-September 2019)

    Vineyards Srebrna Góra, Poland

    NDMI
    sat1


    sat1
    © IGIK & VINUM 4.0 Sp z o.o.

    Natural hazards - Fire

    Mapping burnt areas with Normalized Burnt Ratio index (NBR). Example from forest fire 2014 in Sala, Sweden.


    Sentinel-2 image
    brakuje mi tu obrazka

    sat1
    © Geografiska informationsbyrån

    Natural hazards - Windstorm

    Assessment of forest damage caused by the windstorm

    11-12 August 2017 in Poland, using Sentinel-2

    NDMI - Normalized Difference Moisture Index
    NDMI = (NIR−SWIR) / (NIR+SWIR)
    NIR (band 8) = 842 nm
    SWIR (band 11) = 1610 nm


    Source: Hoscilo and Lewandowska (2018)
    Thank you for your attention