NDVI and multispectral cameras usage

  • by

Increase in biomass is a key parameter allowing to evaluate the condition of cultivation in agriculture. In precision farming the source of this data can be air or satellite images, using cameras with infrared sensors. Images taken in this technology allow to set vegetation indices. In the database of the University in Bonn [1] we can find over 250 indices that can be used in order to specify vegetation of crops and not only. NDVI (Normalized Difference Vegetation Index) is one of the most common vegetation index and the most popular among farmers.

Basing on this index we can make e.g. nitrate dosing map.

The entire magic is hidden in one simple formula:

NDVI=(NIR – RED)/(NIR + RED) , where

NIR is the image representing light reflection in near infrared,
RED is the reflection of red light.

RED image can be made by standard camera or even phone, whereas image of NIR requires camera or rather a sensor enabling imaging of near infrared – this is enlargement of an image by additional information spectrum. That is why when we want to determine biomass increase we must obtain images from cameras equipped with this additional sensor – called multispectral cameras.

The cameras can be placed on satellites, airplanes or drones, therefore we have the division into air and satellite images.

We can obtain satellite image from generally available sources, e.g. Open Access Hub of Copernicus programme – European programme for Earth observation implemented by European Committee in cooperation with European Space Association (ESA).

Satellite data is characterised by quite low spatial resolution (from 10m to 60m per pixel depending on the sensor) – which means that if on 10m there is a large difference of biomass then it will be averaged on the image. Additionally it often happens that the weather makes it impossible to take a photo as the sky is clouded which makes it impossible to make a good quality image.

The alternative may be images taken by multispectral cameras (with additional spectres), like e.g. Sequoia+ or RedEdge-MX, placed on drone or airplane. Images are taken below cloud level so there are no interferences as in the case of satellite imaging. Images taken are characterised by high spectra resolution and quite small number of available spectral channels which means that the possibilities of field analysis are very limited.

The most sophisticated piece of equipment, characterised with high spatial and spectra resolution allowing for precise field analysis regarding vegetation, are hyperspectral cameras as NIR imaging is even more precise than in multispectral technology.

[1] Henrich V. (et al.) A database for remote sensing indices, https://www.indexdatabase.de/db/i.php, access: 26.06.2019 cc