Assess hail damage in cornfields with satellite imagery

The region of Taber and Barnwell, in Alberta, Canada, has a large production of corn and other crops. In early August 2019, the upcoming harvest was severely impacted when a major hailstorm ripped through the region. Hailstones driven by winds that reached more than 140 kilometers per hour (or 87 miles per hour) tore through the area. In this tutorial, as an imagery analyst for a local farmer organization, you'll perform a first damage assessment, using ArcGIS Pro. You'll explore multispectral imagery captured before and after the hailstorm. You'll then apply a vegetation index on both images, compute the difference, and extract the average loss of healthy vegetation in each field.

This tutorial was last tested on July 2, 2024, using ArcGIS Pro 3.3. If you're using a different version of ArcGIS Pro, you may encounter different functionality and results.

Requirements

• ArcGIS Pro (see options for software access)
• ArcGIS Image Analyst extension

Get started and explore the imagery

To assess the hail damage in cornfields of the area, you'll use satellite imagery. First, you'll get set up with the project and the data, and then, you'll start exploring the imagery.

Download and open the project

First, you'll download the project that contains the imagery needed for the tutorial, and open it in ArcGIS Pro.

1. Download the Corn field damage.zip file.
2. Locate the downloaded Corn field damage.zip file on your computer.

   Note:

   Depending on your web browser, you may have been prompted to choose the file's location before you began the download. Most browsers download to your computer's Downloads folder by default.

3. Right-click the Corn field damage.zip file and extract it to a location you can easily find, such as your Documents folder.

   Next, you'll open the project in ArcGIS Pro.

4. Start ArcGIS Pro. If prompted, sign in using your licensed ArcGIS organizational account.

   Note:

   If you don't have access to ArcGIS Pro or an ArcGIS organizational account, see options for software access.

5. In ArcGIS Pro, in the Recent Projects area, click Open another project.

   

6. In the Open Project window, browse to the Corn field damage folder you extracted. Click Corn field damage.aprx to select it, and click OK.

   The project opens.

   

The project contains two images captured on August 4 and August 8, 2019, before and after the hailstorm in the Taber and Barnwell area in Alberta: Before_storm.tif and After_storm.tif. It also contains a Field boundaries vector layer that you'll use later in the tutorial.

Note:

The two images are PlanetScope satellite imagery produced by the Earth-imaging company, Planet Labs. PlanetScope is a constellation of 120 satellites that enables the capture of new images every day for any given area on Earth.

Observe the images in natural color

Next, you'll observe the two pre- and post-storm images. They are multispectral, which means that they contains several separate spectral bands. Each image contains three bands listed in the Contents pane:

• Red (or Band_3)
• Green (or Band_2)
• Blue (or Band_1)

They also contains a fourth band that is not currently visible:

• Near Infrared (or Band_4)

Note:

Near Infrared light is not visible to the human eye, but it is often captured by satellite and aerial imagery sensors, since it is useful for many applications, as you'll see later in the tutorial.

Currently, the images appear in natural color, using the Red, Green, and Blue bands, which correspond to the spectrum of light visible to the human eye. Natural color approximates how colors would look to a person. You'll observe and compare visually the two images.

1. Observe the current view on the map, which shows the Before_Storm.tif image. You can zoom in and out with the mouse's wheel button to see more details.

   

   You can see many fields; many are circular in shape, and others rectangular. They tend to be very green, as in August, before the storm, many crops are getting close to maturity. You'll now compare with the post-storm image, using the Swipe tool.

2. In the Contents pane, turn on After_storm.tif by checking the check box.

   

   Nothing changes visibly on the map since After_Storm.tif is displaying below Before_Storm.tif.

3. Click the Before_storm.tif image to select it.

   

4. On the ribbon, on the Raster Layer tab, in the Compare group, click Swipe.

   

5. In the map viewer, drag from top to bottom to reveal the After_Storm.tif image and compare it to the Before_Storm.tif image.

   

   At first glance, you can see that the post-storm image has somewhat lighter tones in some areas than the pre-storm image. In particular, there seems to be some light-tone traces crossing the area in the northwest to southeast axis. However, it is difficult to gather more precise information about crop damage at this point.

6. When you are done exploring, on the ribbon, on the Map tab, in the Navigate group, click the Explore button to exit the swipe mode.

   

To better visualize the crop damage caused by the storm, you'll turn your attention to the Red and Near Infrared bands.

Explore the Red and Near Infrared band values

To better visualize changes in the crops, you can use the Red and Near Infrared (NIR) bands, which are helpful to assess vegetation health.

• The chlorophyll in healthy vegetation absorbs most of the light in the Red band for use in photosynthesis, therefore reflecting very little of it.
• The cell structure of healthy vegetation strongly reflects NIR light.

Since the satellite sensor captures the amount of reflected light in the different bands, the values of an image pixel showing healthy vegetation will typically be low for the Red band and high for the NIR band. This is illustrated in the spectral profile graph below. In contrast, stressed or dying vegetation will absorb less red light (therefore reflecting more of it) and will reflect less NIR light. The graph also shows that a pixel representing bare soil would reflect even more red light and less NIR light.

To get a better sense of the variability of the Red and Near Infrared band reflectance values in your Alberta images, you'll use the Image Information tool, which provides spectral profile information at the pixel level.

1. In the Contents pane, verify that the Before_storm.tif image is selected. On the ribbon, on the Imagery tab, in the Tools group, click Image Information.

   

   The Image Information pane appears.

2. In the map viewer, point to a dark green field, full of dense and healthy vegetation.

   In the Image Information pane, a spectral profile graph for the pixel at your current pointer location appears. As expected, the Red band reflectance value (symbolized in red) is very low and the NIR band reflectance value (symbolized in gray) is high.

   

   Note:

   The bands are displayed on the graph in the order provided by the image, which is Blue (band 1), Green (band 2), Red (band 3), and Near Infrared (band 4).

3. In the map viewer, point to a beige or light brown area corresponding to bare soil and the absence of vegetation.

   In the Image Information pane, the spectral profile graph updates. The Red band reflectance value is now comparatively higher and the NIR band reflectance value lower.

   

4. Point to more areas of both images and observe how the Red and NIR values vary.

Now that you better understand the relation between the two bands, you can see that computing the gap between Red and NIR values could be a good method to measure the amount of healthy vegetation present on the ground. You'll do that by applying the SAVI index to your images.

Perform change analysis with the SAVI index

Next, you'll learn what the SAVI index is and how it relies on the Red and NIR band values to provide a measure of vegetation health. You'll apply the SAVI index to both pre- and post-storm images, compute the difference between the two resulting rasters, and extract the average loss of healthy vegetation in each crop field.

Note:

This type of analysis is performed at the pixel or cell level.

Satellite TIFF images, like the ones you are using in this tutorial, are rasters. A raster is data that is represented as a grid of cells or pixels.

When doing raster-based analysis, you compute values for every cell in the raster. Learn more about raster data.

Apply the SAVI index

A spectral index combines different spectral bands through a mathematical formula, usually computing some type of ratio. The resulting output is a new raster image.

Note:

There are many different indices combining different spectral bands and using different mathematical formulas. Each index is meant to highlight a different phenomenon, such as healthy vegetation, water, urban development, the presence of ferrous minerals in the ground, and many more. ArcGIS Pro has many indices available out of the box in its Indices gallery.

To highlight healthy vegetation, there are several indices to choose from. You'll use the Soil-Adjusted Vegetation Index (SAVI), which relies on the Red and NIR bands, and uses the following ratio formula:

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

Most importantly, SAVI measures the gap between the NIR and Red bands (NIR – Red). A higher SAVI value indicates a higher presence of healthy vegetation.

Note:

SAVI is an improvement over the more classic NDVI, which uses a simpler formula without the factor L. The factor L is added to the formula to minimize the influence of soil brightness variation on the output value. L is usually assigned the value of 0.5 for imagery scenes with intermediate vegetation cover. The final SAVI value varies from -1.5 to +1.5 (when L=0.5). Optionally, you can read more about the Soil-adjusted vegetation index.

Next, you'll apply the SAVI index to the Before_storm.tif and After_storm.tif images.

1. In the Contents pane, make sure that the Before_storm.tif image is selected.
2. On the ribbon, on the Imagery tab, in the Tools group, click Indices. In the Indices pane, select the SAVI index.

   

3. In the SAVI window, choose the following values:
   • For the Near Infrared Band Index, choose 4 - Band_4.
   • For the Red Band Index, choose 3 - Band_3.
   • For Soil-brightness correction factor, keep 0.5.

   

4. Click OK.

   A new layer, SAVI_Before_storm.tif, appears.

   Note:

   Unlike the original satellite images, the SAVI raster layer is not multiband. Each raster cell holds exactly one numeric SAVI value that summarizes the healthy vegetation status at this location.

   Also, the SAVI index tool is a raster function, which means that the resulting SAVI layer is computed dynamically and is not saved on disk. Since no intermediate datasets are created, processes can be applied quickly.

5. Similarly, apply the SAVI index to the After_storm.tif image.

   A new layer, SAVI_After_storm.tif, appears. You'll use the Swipe tool again to compare the two SAVI layers.

6. In the Contents pane, drag SAVI_Before_storm.tif above SAVI_After_storm.tif so that it appears first in the list of layers. Make sure that both layers are turned on, and, if necessary, click SAVI_Before_storm.tif to select it.

   

7. On the ribbon, on the Raster Layer tab, in the Compare group, click Swipe.
8. In the map viewer, drag from top to bottom to reveal the SAVI_After_Storm.tif image and compare it to the SAVI_Before_Storm.tif image.

   The highest SAVI values are symbolized in white or light gray tones, and represent the areas with a higher presence of healthy vegetation. You can see that many fields seem to have higher SAVI values before the storm than after. However, it is still difficult to see whether some fields were more affected than others by the storm. Next, you'll compute the difference between the two SAVI layers to measure the change in vegetation more precisely.

9. When you are done exploring, on the ribbon, on the Map tab, in the Navigate group, click the Explore button to exit the swipe mode.
10. Press Ctrl+S to save the project.

Compute the change between the two SAVI layers

To measure the change in vegetation caused by the storm, you'll compute the difference between the two SAVI layers using the raster function Compute Change. For each pixel, the SAVI value in SAVI_After_storm.tif will be subtracted from the SAVI value in SAVI_Before_storm.tif. A resulting positive value will signify a loss of healthy vegetation.

1. On the ribbon, on the Imagery tab, on the Analysis tab, click the Raster Functions button.

   

2. In the Raster Functions pane, type Compute Change in the search box, and click the Compute Change tool to open it.

   

3. In the Compute Change Properties pane, enter the following parameter values:
   • For From Raster, choose SAVI_After_storm.tif.
   • For To Raster, choose SAVI_Before_storm.tif.
   • For Compute Change Method, make sure Difference is selected.

   

4. Click Create New Layer.

   The new raster layer, Compute Change_SAVI_After_storm.tif_SAVI_Before_storm.tif, appears.

   

   The areas in purple tones indicate a loss in healthy vegetation (positive numeric values), the most substantial loss being symbolized in deep purple. A visual inspection indicates quite clearly that the hailstorm crossed this area diagonally in the northwest to southeast axis, damaging the most fields in that diagonal area. The fields in the upper right and lower left side of the image appear much less affected.

   Note:

   You might notice that a few areas symbolized in green (negative numeric values) seem to indicate small gains in vegetation. Since only four days separate the two images, it is unlikely that much agricultural vegetation growth occurred during that time. However, it is possible that some bare soil areas that were dry, because of the August weather, started growing weeds quickly after they got soaked with rain and melted hail during the storm.

   The small gains in vegetation are not relevant to your analysis of crop damage, so, you'll eliminate all the values below 0 in your raster with the Remap raster function.

5. In the Raster Functions pane, look for and open Remap.

   

6. In the Remap properties pane, enter the following parameter values:
   • For Raster, choose Compute Change_SAVI_After_storm.tif_SAVI_Before_storm.tif.
   • For Remap Definition Type, keep List.

   You'll express with a remap rule that all negative values in the raster should be removed, that is, be changed to NoData. As seen in the Contents pane legend, the lowest value of the layer is -0.576915, so you'll use -0.6 as the minimum value.

7. In the first row of the table, enter the following values:
   • For Minimum, enter -0.6.
   • For Maximum, keep 0.
   • For NoData, check the check box.

   

8. Click Create New Layer.

   A new raster layer, Remap_Compute Change_SAVI_After_storm.tif_SAVI_Before_storm.tif, appears. You'll rename it to a shorter, more meaningful name.

9. In the Contents pane, click the Remap_Compute_Change_SAVI_After_storm.tif_SAVI_Before_storm.tif layer two times to edit it. Change the name to Loss of healthy vegetation. and press Enter.
10. In the Contents pane, turn off all layers except the Loss of healthy vegetation raster and the basemaps World Topographic Map and World Hillshade.

    

    The new layer shows only the areas where there has been loss in healthy vegetation. The deep shade of purple indicates the deeper losses.

11. Press Ctrl+S to save the project.

Extract the average vegetation loss for each field

In this last section, you'll compute the average loss of healthy vegetation in each field. To do this, you'll use the Field boundaries feature class, where the boundaries of all the cultivated fields in the area are represented by polygons. Such a layer would be maintained by the farmer organization you work for.

1. In the Contents pane, drag the Field boundaries layer to the top of the layer list. Turn it on.

   

   The polygons representing the cultivated fields appear, symbolized in red.

   

   To compute the average vegetation loss, you'll use the Zonal Statistics as Table tool. For each Field boundaries polygon, the tool will calculate the mean value of all pixels in the Loss of healthy vegetation raster that fall within that polygon.

2. On the ribbon, on the Analysis tab, in the Geoprocessing group, click Tools.

   

   The Geoprocessing pane appears.

3. In the Geoprocessing pane, search for and open Zonal Statistics as Table.

   

4. In the Zonal Statistics as Table tool, enter the following parameter values:
   • For Input raster or feature zone data, choose Field boundaries.
   • For Zone field, keep Field_ID, which was populated automatically.
   • For Input value raster, choose Loss of healthy vegetation.
   • For Output table, type Vegetation_loss_table at the end of the Corn field damage.gdb path.
   • For Ignore NoData in calculations, keep the box checked.
   • For Statistics type, choose Mean.
   • Accept the default for any other parameter.

   

5. Click Run.

   The output of this tool is a table, which you'll now open.

6. In the Contents pane, under Standalone Tables, right-click Vegetation_loss_table, and click Open.

   

   The table opens. It contains one row for each cultivated field polygon. The Mean column gives the mean of healthy vegetation loss for each polygon. You now need to join this table back to the Field boundaries layer using the common Field_ID.

7. Close the Vegetation_loss_table.
8. In the Contents pane, right-click Field boundaries, choose Joins and Relates, and Add Join.

   

9. In the Add Join window, choose the following values:
   • For Input Table, verify that Field boundaries is selected.
   • For Input Join Field, choose Field_ID.
   • For Join Table, choose Vegetation_loss_table.
   • For Join Table Field, choose Field_ID.

   

10. Click OK.

    The Mean value has now been added to each row in the Field boundaries attribute table. Next, you'll symbolize the Field boundaries layer based on that mean value, using a graduated colors scheme.

11. Double-click the Field boundaries symbol to open the Symbology pane.

    

12. In the Symbology pane, if necessary, click the back button.

    

13. Under Primary symbology, choose Graduated Colors.
    • For Field, choose MEAN.
    • For Method, ensure Natural Breaks (Jenks) is selected.
    • For Classes, choose 4.
14. For Color scheme, expand the drop-down list and choose Show names. In the list of color ramps, choose the Yellow to Red color scheme.

    

    The Field boundaries symbology updates.

15. In the Contents pane, turn off the Loss of healthy vegetation layer to simplify the map display.

    You'll rename the Field Boundaries layer and the symbology labels to make them more meaningful.

16. In the Contents pane, click Field boundaries two times to edit it. Change the name to Vegetation loss per field, and press Enter.

    

17. In the Symbology pane, on the Classes tab, click the label value for the yellow class. Type Low and press Enter. Similarly change the other label values to Medium, High, and Very high.

    

    The labels also update in the Contents pane.

18. Observe the final result.

    

    The fields that appear to have experienced High and Very high loss in healthy vegetation are clearly concentrated in the same northwest to southeast axis observed before. The fields with Low loss are mostly concentrated in the upper right and lower left corners.

    You might observe some neighboring fields with different damage levels. This could have many causes. For instance, different crop types could be affected differently by a hailstorm. Also, two fields with the exact same crop, but each at a different level of plant maturity, could also be affected differently.

    The map offers a first assessment of the damage caused by the hailstorm. It could be used to guide a more detailed inspection on the ground.

    Note:

    The next steps could be to publish the map to the web through ArcGIS Online and integrate it into a Field Operations application. This would allow inspectors to interact directly with the map on their mobile devices, and update it with their findings in real time. You can see an example of a similar workflow in the tutorial Inspect hydrants.

19. Press Ctrl+S to save the project.

In this tutorial, you observed before- and after-storm imagery in natural color, and explored the pixels' spectral profiles. You learned about the importance of Red and Near Infrared bands to assess vegetation health, and you learned what the SAVI index is. You applied the SAVI index to both images and computed the difference between the two resulting rasters. Finally, you extracted the average loss of healthy vegetation for each crop field, and created a map that provides a first assessment of the crop damage that was caused by the storm.

You can find more tutorials like this one on the Introduction to Imagery & Remote Sensing page.

Acknowledgements

• The Before_storm and After_storm layers are PlanetScope satellite images, used by courtesy of Planet Labs, Inc. (Copyright © Planet Labs, Inc.).
• The tutorial card image, Corn Cornfield Harvest, is from Carola68 on Pixabay.com.