B3 · Applications of Artificial Intelligence

AI for the Environment and Wildlife

AI helps protect people and nature by finding patterns in very large amounts of environmental data, such as rainfall, river levels, satellite images and wildlife photos, in order to forecast disasters, monitor forests and track animals.

8 min Beginner Jump to the activity

What you need: Nothing. This page starts from the beginning, with no programming and no formulas.

Definition

Can a computer warn a village about a flood a week before the water arrives?

AI helps protect people and nature by finding patterns in very large amounts of environmental data, such as rainfall, river levels, satellite images and wildlife photos, in order to forecast disasters, monitor forests and track animals.

In simple words

One forest guard cannot watch a whole forest. One weather officer cannot study every river. AI is like giving them thousands of extra eyes that never blink: cameras, satellites and sensors, with a model that tells them where to look first.

Why it matters

Floods, cyclones and heat waves affect millions of people in India every year. A warning that arrives days earlier can save lives, crops and animals. This is AI used for a clear public good, and it also shows that a prediction is only useful when it reaches people in time and in their language.

Key terms

Forecast
A prediction of future conditions, such as weather or river levels.
Early warning system
A way of alerting people to danger in time for them to act.
Satellite imagery
Photos of the Earth taken from space, repeated regularly.
Camera trap
A camera left in the wild that takes a photo when an animal passes.
Biodiversity
The variety of living things in a place.
Carbon footprint
The amount of greenhouse gas released by an activity, including running computers.

Explained in detail

Too much Earth to watch

The natural world produces more data than people can study. Satellites photograph the whole planet every few days. Weather stations, river gauges and ocean buoys report every few minutes. Thousands of automatic cameras and microphones record forests day and night. Hidden in this data are the early signs of floods, fires, illegal logging and disappearing species. AI is useful here for one main reason: it can look at everything, all the time, and point people to the places that need attention.

Faster weather forecasts

Traditional weather forecasting solves the equations of the atmosphere on some of the world's largest computers, a job that takes hours. Since 2023 several research groups have shown AI models that learn from decades of past weather and produce a ten-day forecast in about a minute on a single machine, with accuracy that matches or beats the traditional method for many measures. National weather services are now testing how to combine the two. For a country exposed to cyclones and monsoon floods, faster and more local forecasts mean more time to move people to safety.

Floods: from Patna to the world

One of the best-known projects began in India. In 2018 a pilot project around Patna in Bihar used a model to forecast river flooding and send alerts to phones in the area. The work grew into a public service that now publishes river-flood forecasts for more than a hundred countries, up to about seven days ahead. An important finding from the research was that a model trained on rivers with long records can make useful forecasts for rivers that have few measuring stations, which describes much of Africa and Asia.

Counting tigers and listening to forests

India's national tiger count places camera traps at tens of thousands of locations and collects tens of millions of photos. Sorting them by hand would take years. Software first separates the photos by species, and then identifies individual tigers by their stripe patterns, which are as unique as fingerprints. The same idea is used with sound. Recorders left in forests capture the calls of birds, frogs and insects, and models identify the species present. In some rainforests, solar-powered listening devices recognise the sound of chainsaws and alert rangers to illegal logging as it happens.

ThreatWhat AI watchesAction people take
Forest firesSmoke in camera and satellite imagesSend fire crews early
Illegal loggingSudden loss of tree cover; chainsaw soundsSend forest patrols
Elephants on railway linesVibration and camera sensors near the trackWarn the train driver to slow down
Air pollutionReadings from sensors, traffic and weatherIssue health warnings

AI's own footprint

AI is not free for the environment. The International Energy Agency estimated that data centres used about 415 terawatt-hours of electricity in 2024, roughly one and a half per cent of the world's total, and projected that this could more than double by 2030, with AI the main cause of the growth. Many data centres also use large amounts of water for cooling. Whether AI helps the planet overall depends on what it is used for, how efficiently it is built and where its electricity comes from.

How it works

ProblemData usedWhat AI adds
River floodsRainfall, river levels, the shape of the landForecasts of river flooding up to about a week ahead
CyclonesSatellite images, wind and sea temperatureBetter estimates of the path and strength of a storm
Forest lossSatellite photos taken again and againAlerts when tree cover suddenly disappears
Wildlife countingMillions of camera-trap photosNames the species in each photo in seconds
Animals on railway tracksSensors and cameras near the trackWarns train drivers when elephants are near

How a flood forecast is made

  1. Collect years of records: how much rain fell, and how high the river rose afterwards.
  2. Train a model to learn the link between rain upstream and water level downstream.
  3. Feed in today's weather forecast. The model predicts the river level for the coming days.
  4. Mark on a map which areas the water is likely to reach.
  5. Send alerts through phones, local officials and volunteers.

Try it yourself

A flood warning travels to a village

Heavy rain is falling in the hills, far from a riverside village.

Match the data to the problem it helps solve

Which data would you collect for each problem?

  • Photos from cameras tied to trees
  • Satellite images of the same forest every week
  • River levels and rainfall for many years
  • Sound recordings from a wetland

Common mistakes

  • Thinking a forecast is certain. It gives a likelihood, and it improves as the event gets closer.
  • Thinking the model alone saves lives. The warning must reach people, and they must be able to act on it.
  • Forgetting that AI itself uses energy and water.

Check yourself

Question 1

1. What does a flood model learn from?

Question 2

2. Why are camera-trap photos a good job for AI?

Question 3

3. A perfect forecast is made but nobody in the village hears about it. What happens?

Summary

  • AI finds patterns in rainfall, satellite, sensor and camera data to forecast disasters and monitor nature.
  • Earlier warnings give people time to protect lives, animals and crops.
  • A forecast needs people and communication to be useful, and AI has its own environmental cost.

Finished reading?

Sources: Google Research: flood forecasting · Wildlife Insights: camera-trap data