Monitoring the soil moisture level of crop fields is one of the most important thing to do for having an optimal crop yield. The finer the spatial and temporal resolution, the better it is to take prompt actions when necessary.
The Copernicus Sentinel-2 programme makes it possible to have a full coverage of any place in Europe at least every 5 days in 13 different multispectral bands. Based on this, we studied the Normalized Difference Water Index (NDWI) over the period from May to September 2016 in the agricultural region in the Northern part of Liege, Belgium. The NDWI is defined by the following formula:
NDWI = ( NIR – SWIR ) / ( NIR + SWIR )
Where NIR is for the Near infrared band and SWIR for Short-Wave Infrared band. For Sentinel-2 data, these bands are respectively 8A and 11, which spatial resolution is of 20m. Values range from -1, very bad moiture level, to 1 very high moisture level.
The reflectance in the NIR band is influenced by the leaf internal structure and the dry matter of the crop. The SWIR, on the other hand, reflects the changes in both the vegetation water content and the spongy mesophyll structure. Other things being equal, a darker area in the SWIR area of the spectrum highlights a higher water content. The combination of both bands removes the variations induced by the leaf internal structure and the leaf dry matter content. The accuracy in the water content measurement is thus enhanced.
Our time series study gave us the following result for the 5-month period.
The NDWI is globally very good for all the periods analysed. July seems however dry compared to the other dates. From the weather reports (meteobelgique.be), we know that the month of June was very rainy. July started very dry climate because of an anticyclone over the Atlantic and Brittany, giving Belgium wonderful summer days. This could explain why the crops are drier in this scene of July.
By looking more closely to the crop fields, we can spot some areas that need more attention than other parts. In the following comparison, we can see how “moist” the crops seen from space in True Colour are.
Among the crops highlighted in the blue circle, we see that there is no visible correlation between the greenness of the crops in True Colors and the Moisture level. In other words, darker crops in True Colour do not necessarily correspond to moister crops in NDWI.
The following example shows another area where it is not easy to see in True Colour which are the moistest fields.
The comparison shows how the moisture level varies inside a same crop field.
The following image shows the comparison between the NDVI on the left side, and the NDWI on the right side. Surprisingly on the image dating of the 7th of June, a very wet month, we see that the most photosynthetically active crop field (High NDVI) is one of the least moist. One of the hypothesis could be that active crops assimilate water faster than less active crops. In the latter case, water stays longer on the surface letting the field appear moister to the Sentinel-2 MSI instruments.
Sparsely planted fields appear less moist, as shown in the following image comparison. For an almost bare land, water does not stay long: it infiltrates quickly into the ground or evaporates faster than on a dense vegetated field.
For the beginning of the month of July, we see the opposite of what we have seen for the month of June. Here, a moister soil represents crop field more photosynthetically active. It is understandable that the crops needs less water during the first months of their growth. The more active they are, the more water they need.
Of course, these interpretations have to be confirmed by the experts (farmers, agronomy engineers, etc.) and unavoidable in-situ observations have to be carried out too.
In our Time Series analyses, we have a total of 9 “views” of the crops for over the 5-month period (May to September 2016). We have several other data that can come into play to make the temporal resolution higher. Landsat 7 and 8 scenes can be used as well as Sentinel-1 SAR data that can operate by day and night and by all kinds of weather.
What we have seen in this article is the Normalized Difference Water Index (NDWI) that brings additional information about the crops, and more precisely their hydric state. Having all the band combinations and indexes all together (True Colour, False Colour, NDVI, NDWI), and analysing them, makes it possible to fine-grain the analyses of the crops.
Quadratic studies problems encountered in the agro-industrial sector and develops for its partners real-time monitoring tools and decision support tools, intended to help better 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.