NDVI Index from Satellite Image to Track Crop Area
The Geospatial analysis uses the satellite imagery to help find useful insights which otherwise is not feasible from other data sources. As we started with the agricultural domain, our objective was to find the spread of agricultural fields and to determine the crop yields in a specific region.
NDVI (Normalized Difference Vegetation Index)
In order to achieve this objective, we computed NDVI (Normalized Difference Vegetation Index) which helps with land cover classification. And NDVI visually demonstrates how well a plant is photosynthesizing. These are crucial for evaluating the productivity of crops and proper crop management.
NDVI quantifies vegetation by measuring the difference between near-infrared and red light. Healthy vegetation (chlorophyll) reflects more near-infrared (NIR) and green light compared to other wavelengths. However, it absorbs more red and blue light, that’s why we see vegetation as green color. NDVI value ranges from -1 to 1.
Here LANDSAT-8 satellite images were used to calculate the NDVI of the entire datasets and clipped to a specific region. Normally, the agricultural fields will have NDVI values ranging from 0.2 to 0.5. And for the forest region, it ranges from 0.6 – 0.9.
Once the NDVI is calculated, the pixel values for the continuous range for any dataset could be found and we can determine the total area of croplands present in that district. By calculating the NDVI for a specific region periodically we could see a pattern of farming in that region (i.e.) we would be able to identify the month of sowing and harvesting, the time taken for the crops to completely mature and so on in a particular region.
GIS (Geographic Information System)
The use of GIS software (Geographic Information System) like ArcGIS and QGIS for processing and analyzing the datasets and using the attribute table, we calculate the area of the crop fields and area of the forests. This eventually helps in monitoring the crops and managing it effectively and it also reduces the inconsistent data we have from the ground.
In the below Table, we see the changes in total cropland from a snapshot by every month end indicating the harvesting pattern from June to October in the three districts.
The following two visuals help us see the pattern as it changes from one month to the other for Thoothukudi (left) and Tirunelveli (right).
There is a growing number of such use-cases which emerges by the use of satellite imagery like mapping different types of land, understanding global forest cover, epidemic spread analysis, identifying water bodies and their depletion, weather forecasting and many more.
Credits: This work is first from the list of projects coming out of Probyto’s Data Science Internship programme. This article is contributed by Siddardh, who is an intern with Probyto and a final year student of B.Tech. Computer Science Engineering from PSG Institute of Technology and Applied Research, Coimbatore, Tamilnadu, India.
Please enter a valid email address
Processing your application
There was an error sending the email, please try again
Great!Check your inbox and click the link to confirm your subscription
We're here to help and answer any question you might have.
We look forward to hear from you. 🙂