How to Make Outstanding Maps with Sentinel-2 and ArcGIS Pro – Part 2: Spectral Indices

Introduction

This is the second part of the blog post series about making outstanding maps using Sentinel-2 imagery in ArcGIS Pro. In Part One I briefly described what is Sentinel-2 data, where and how to get it and how to process it, to produce various band combinations, in order to understand specific phenomena.

The different band combinations are key to point-out such phenomena and in many cases they can serve as excellent basemaps for thematic mapping. But they are suitable only for qualitative interpretation. In order to extract measurable information from multiband imagery, we often conduct mathematic calculations among various bands to produce new, quantified, single-band raster files, the so-called Spectral Indices.

Spectral Indices are extremely useful to monitor specific phenomena, such as floods, wildfires, deforestation etc. It is common practice for scientists to use spectral indices to make measurements or detect changes on the dynamic surface of our Planet. However, I often use (or misuse!) spectral indices to produce more realistic maps.

Have a look at Picture 1, which shows the Natural Color band composite of my area of study. I can easily distinguish three principal geomorphological zones: the coastline, which is quite complex in this example, the lowland occupied by cultivations and the mountainous area. One way to produce a map is to just use the Natural Color as a basemap and superimpose vector data.

Another way (my way) is to calculate certain spectral indices, to extract data related to the three aforementioned zones and then process and style this data according to my needs and taste. In ArcGIS Pro there are two ways to calculate such indices:

• Through the Indices Gallery at the Tools group on the Imagery contextual tab on the ribbon and
• Through the Band Arithmetic function from the Raster Functions pane.

At the following paragraphs I will demonstrate how I calculate three specific indices in ArcGIS Pro, using both ways. More precisely, I will calculate the:

• Normalized Difference Vegetation Index (NDVI),
• Soil-Adjusted Vegetation Index (SAVI),
• Modified Normalized Difference Water Index (MNDWI).

All calculations will happen in the multiband image I produced in Part One. In order to reference its bands correctly, I will always have to consult Table 2 in Part One, which shows the correspondence between original Sentinel-2 bands and the composite’s bands in ArcGIS Pro.

Normalized Difference Vegetation Index (NDVI)

The Normalized Difference Vegetation Index (NDVI) is a standardized index commonly used to generate an image displaying greenness (relative biomass). NDVI takes advantage of the contrast of the characteristics of two bands from a multispectral raster dataset: the chlorophyll pigment absorptions in the red band and the high reflectivity of plant materials in the NIR band.

I always use NDVI to extract data related to vegetation, which I further style appropriately and include in my layers stack, especially when I produce topographic or hiking maps. For example, at the hiking maps on the trailhead signs for Kythera Trails the vegetation has been produced with NDVI on Sentinel-2 imagery!

So, let’s do this right now in Pro!

I select the Band Composite multiband image at the Contents pane and I go to the Imagery contextual tab at the ribbon. I go to the Tools group and I click on Indices to open the Indices Gallery. From the dropdown list I select the first index which is the NDVI (Picture 2).

The NDVI calculator opens as a pop-up on my screen (see Picture 3). According to Table 2 from Part One, the NIR band of the Band Composite is Band_4 and the Red Band is Band_1. So I enter those values and I click OK (Picture 3).

The NDVI is being calculated and added on the Contents pane as a new layer. This layer is a single-band raster file, with pixel values ranging from minus one (-1) to plus one (+1) and by default is symbolized in grayscale (Picture 4).

Every pixel below zero represents areas with no vegetation, while pixels above zero represent areas with vegetation. The larger the pixel number the most dense, vigorous and healthy the vegetation is. To illustrate this, I change the Primary symbology from Stretch to Classify and I create ten classes above zero with an interval of 0.1 (see Picture 5). I then assign a color to each class, according to the following Color Scheme:

The NDVI layer look like the one in Picture 5. Everything below zero, including surface water and the sea, appear in red. Areas with values between +0.1 and +0.4 have weakly irrigated, sparse vegetation, areas with values between +0.4 and +0.8 are adequately watered, healthy crops and areas with values between +0.8 and +1.0 are the most healthy, robust crops and other areas with physical vegetation.

In the final part of these blog post series, I use the NDVI layer to illustrate only the crop fields at the lowland zone. For this, I clip it inside the boundaries of the lowland zone.

Modified Normalized Difference Water Index (MNDWI)

The Modified Normalized Difference Water Index (MNDWI) uses the Green and SWIR bands for the enhancement of open water features. It also diminishes built-up area features that are often correlated with open water in other indices.

I normally use the MNDWI to extract inland water bodies, such as lakes, reservoirs, rivers, as well as the sea, hence the coastline. MNDWI does a great job in distinguishing water from other physical features.

For demonstration purposes, I will calculate this index using the equivalent Method from the Raster Functions pane. So, in ArcGIS Pro, I open the Band Arithmetic function, for Raster I select the Band Composite, for Method I select MNDWI from the exhaustive dropdown list, and at the Bands field I write the numbers 2 and 6. The number 2 is the Band_2 of the Composite which represents the Green band of Sentinel-2, while the number 6 is the Band_6 of the Composite which represents the SWIR-1 band of Sentinel-2 (see Table 2 in Part One). My screen looks like the one in Picture 6.

I then click on the Create new layer button and the newly created single-band MNDWI layer is added on the Contents pane in grayscale. At the Symbology properties, I change the Primary symbology from Stretch to Classify and I create six classes, above zero, with an interval of 0.2 (see Picture 7). I then assign a color to each class, according to the following Color Scheme:

The pixel values of the MNDWI range from minus one (-1) to plus one (+1). Every pixel below zero represents areas with no water, while pixels above zero represent areas with water. Areas with pixel values between 0.0 and 0.6 are mostly irrigated crops, while areas between 0.6 and 1.0 include water bodies and of course the sea (Picture 7).

Comparing the NDVI with the MNDWI is also a very good way to understand what every class on both indices represent in the real World.

In the final part of these blog post series, I use the MNDWI layer to illustrate only the coastline and the sea, as well as the inland water bodies. For this, I mask the layer to only show pixels with values greater than or equal to 0.8.

Soil-Adjusted Vegetation Index (SAVI)

The Soil-Adjusted Vegetation Index (SAVI) method is a vegetation index that attempts to minimize soil brightness influences using a soil-brightness correction factor. This is often used in arid regions or desert areas, where vegetative coverage is low.

It is actually an alternative to the classic NDVI, which I I normally use to extract vegetation in areas sparsely covered by crops or forests, such as the bald mountains of Greece.

For demonstration purposes, I will calculate this index using the equivalent Method from the Raster Functions pane. So, in ArcGIS Pro, I open the Band Arithmetic function, for Raster I select the Band Composite, for Method I select SAVI from the exhaustive dropdown list, and at the Bands field I write the numbers 4, 1 and 0.5 (see Picture 8).

The number 4 is the Band_4 of the Composite which represents the NIR-1 band of Sentinel-2, while the number 1 is the Band_1 of the Composite which represents the Red band of Sentinel-2 (see Table 2 in Part One).

The number 0.5 is the soil brightness correction factor, represented with the letter L. The soil brightness correction factor varies depending on the amount of green vegetative cover. In areas with no green vegetation cover, L=1; in areas of moderate green vegetative cover, L=0.5; and in areas with very high vegetation cover, L=0, which is equivalent to the NDVI method.

My screen looks like the one in Picture 8.

I then click on the Create new layer button and the newly created single-band SAVI layer is added on the Contents pane in grayscale. At the Symbology properties, I change the Primary symbology from Stretch to Classify and I create six classes, above zero, with an interval of 0.5 (see Picture 9). I then assign a color to each class, according to the following Color Scheme:

The pixel values of the MNDWI range from minus one (-1.5) to plus one (+1.5). Every pixel below zero represents areas with no vegetation, while pixels above zero represent areas with vegetation . Areas with pixel values between 0.0 and 0.5 are mostly crops, while areas between 0.5 and 1.5 include areas with dense vegetation (Picture 9).

In the final part of these blog post series, I use the SAVI layer to illustrate vegetation only in the mountainous zone of my area of study. For this, I will use an elevation layer to clip SAVI within the boundaries of the mountainous zone.

Conclusion

Thank you for reaching the end of this post. In the third part, I will describe how the three indices I created here will be further masked, clipped and combined to produce a realistic basemap for the area of study.

In the meantime, I would be very happy to know that anyone got inspired by Part One and Part Two and actually produced data with Sentinel-2. I would also be happy to receive any feedback or suggestions for improvement.

Kindest regards from Crete, Greece

Spiros