Monitoring the Evolution of Crops Using the NDVI

In a previous article, we followed the evolution of crop parcels in the Northern Part of Liege. We gathered the RGB bands (bands 4, 3, 2) of Sentinel-2 imageries taken at different moments. The bands of the same dates were combined to created True Colour scenes and these temporal scenes were assembled to create a Time Series. Having this as a basis, we were able to follow the evolution of the crops from the period of 01/05/2016 to the 25/09/2016, in True Colour.

Using these three bands in the visible electromagnetic spectrum already gives us tons of information. It is possible to see when and where the fields are harvested, to estimate the crop types using the calendars of harvestings, and we have insights about the colour variations between the crops.

Remote Sensing satellites instruments come fortunately with more than the 3 bands in the visible EM spectrum – 13 for Sentinel-2 imageries – and we know for sure they contain a zillion of information just waiting to be extracted.

In another previous article, we developed the Normalized Differenced Vegetation Index (NDVI) index that makes use of the red and near infrared (nir) bands. Having values comprised between -1 and +1, the NDVI reflects the plants’ health, vitality, the greenness and the amount of vegetation.

The idea is to create NDVI for a set of crop parcels at different dates and to see how it evolves over time.

Crop evolution near Liege (Belgium) using the NDVI in 2016

Crop evolution near Liege (Belgium) using the NDVI in 2016

We see the NDVI constantly increasing. The greener the area, the higher the index and thus, the higher the photosynthetic activity.

The first scene (01/05/2016) is our starting point. From a weather point of view, the last 10 days of April were very cold for the season. The mean temperatures observed were the same as a normal month of March. Crops accuses thus a little evolution delay as they would be in normal times.

In this scene we already see the greenest fields in true colour are not the most active. Here’s an obvious example:

The greenest fields (true colours) are not the most photosynthetically active ones

The greenest fields (true colours) are not the most photosynthetically active ones

Another example at the next scene.

The greenest fields (true colours) are not the most photosynthetically active ones

The greenest fields (true colours) are not the most photosynthetically active ones

Can we assume that a similar signatures evolution in both the True Colours and NDVI are of the same crop type? It is far too early to conclude that and this will be covered in a subsequent article. Facts show that different crops can have the same signatures and, for example, a weaker crop of type 1 can have the same signature as a strong crop of type 2.

In early July, we see a crop field that looks different in the True Colour scene, but has exactly the same NDVI – Their photosynthetic activity seems to be similar.

Different Crop type but same NDVI

Different Crop type but same NDVI

Are they of different crop type having the same activity? Are they maybe of the same crop type but at different stages in their evolution? Further analyses are also needed here, in situ would be the best welcome.

The scene of the 16th of August shows a dark green crop field in true colour, with a poor NDVI. Is this crop field sick? Or is it at the very beginning of its growth?

Green field (True Colours) and poor ndvi: Sick crops?

Green field (True Colours) and poor ndvi: Sick crops?

The following collage shows the evolution of this field and its neighbor. The NDVI stays rather low but for the neighboring field it is steadily growing. Is our field 1 sick, or not optimized?

Evolution of NDVI throughout the season

Evolution of NDVI throughout the season

NDVI is a powerful indicator of the photosynthetic activity. It can, unfortunately, not be used alone as we’ve seen and simply comparing with a true colour document is not sufficient either. We first need to know of which type the crops are and knowing about the seeding and harvesting period as well. In situ measures are in all cases necessary to validate the results. An NDVI Time Series document is however very useful to follow the photosynthetic activity of the crops, whatever they are, and to spot anomalies as soon as they occur!


Quadratic is currently studying problems encountered in the agro-industrial sector and develops for its partners real-time monitoring tools and decision support tools, intended to help them manage their logistics. Beside searching ways to reduce costs, the company also helps to monitor crop fields using the products of the ESA and European Union’s Copernicus (Sentinel) programme. Quadratic is member of the Copernicus World Alliance, which aims the development of synergies between organizations active in the Copernicus programme.

Share the joy
  •  
  •  
  •  
  •  
  •