The last few decades have seen a massive surge in industrialization, which is essential for the success of the global economy. However, it is also no secret that these activities have severely affected the health of the planet and has led to massive stress on our natural resources and environment. Today, our planet is undergoing a major environmental crisis — from air pollution to animal endangerment to global warming. If we do not act swiftly now, we will be jeopardizing not only flora and fauna around us but our entire civilization. The traditional solutions that are being taken to address environmental issues are not enough to tackle problems of such a gigantic scale. What can be useful is the use of technology to speed up the remedial process, and Artificial Intelligence(AI) backed technologies can provide an effective solution for the same.

To say that AI and Machine Learning have revolutionized our world would be an understatement. From healthcare to automobiles, from sports to banking, almost all the sectors have realized the power that AI brings with itself. Artificial Intelligence refers to training computer systems to perform human-like tasks. While AI is the broad science of mimicking human abilities, Machine learning is a subset of AI that trains a machine on how to learn by looking at patterns in data. Deep learning is a specific subfield of machine learning which emphasizes on learning successive layers of increasingly meaningful representations. The field of Satellite Imagery is also undergoing massive development, and machine learning solutions have provided some significant breakthroughs. Today Machine Learning and Deep learning solutions make it possible to analyse vast amounts of data that otherwise would have been impossible for humans.

Satellite imagery, also known as Earth observation imagery, are the images of Earth or other planets collected by imaging satellites. Satellites such as MODIS and Landsat provide real-time high-resolution satellite imaging from the outer space. Tools like Google Earth Engine can pull data from a 5-petabyte database of satellite imagery that is updated in real-time and processes images on the cloud. Such imagery can be of tremendous help in understanding and tracking real-time changes in a region.

Satellite imagery data can be analysed over a period of time to understand the causes of the decline in vegetation for a region. This is done by calculating the vegetation index of an area, which in turn is an indicator of the health of vegetation in that area. One of the most widely used indexes to measure vegetation is the Normalized Difference Vegetation Index (NDVI), whose values range from +1.0 to -1.0.

Satellite sensors capture a variety of data, and one such type of data measures explicitly wavelengths of light absorbed and reflected by green plants. Dense vegetation reflects a lot of near-infrared light(not visible to the human eye) as compared to the visible red light. The reverse happens in the case of sparse vegetation. NDVI compares the reflected near-infrared light to reflected visible red light by the plants and can give a rough estimation of the type, amount, and condition of vegetation at a place. NDVI can also help in identifying areas where there are changes in vegetation due to human activities such as deforestation, natural disturbances such as wildfires, or changes in plants’ phenological stage.

Satellite images are also used to monitor coral reefs. Using AI-backed methods, it is now not only possible to map the location of a reef but also determine whether a reef is healthy or dying. This is done by analysing new algae growth and picking out bleaching events from the satellite data itself. NASA’s Fire Information for Resource Management System(FIRMS) uses near real-time data from MODIS to help guide the rescue missions in times of wildfires. Satellite imagery analysis can also enable us to point out if there has been severe deforestation in any area which might be leading to the effects of global warming. Prediction of hurricanes, droughts, and floods are other areas where analysis of satellite imagery is extensively applied.

The satellite imagery domain is a valuable resource of data, which, when utilized fully, can provide actionable insights for us to analyse. There is no better way to use technology than to work on some real problems threatening the planet, and being able to utilize data from the satellites is a step in that direction.

Image from pxfuel