Elephant counting from space

For the first time the population of elephants in addo elephant park in South Africa has been counted from space using computer learning.

Counting elephants from space is increasingly possible. Why is this good? it can be done cheaply, and therefore can be done regularly. This allows remote reserve poaching epidemics to be noticed early. As elephants are often the first animal to be targeted, automated satellite counts could give early warning about parks in danger.

The satellite images now taken are detailed enough to show each individual elephant has a separate grey blob. Then by using computer learning, a program can be taught to count these accurately.

This is a fantastic leap forwards, however there is is a level of caution that should be applied. Firstly elephants are far and away the largest land mammal, as such counting them is going to be easier than almost any other species. Secondly, while African Savannah elephants have been hit by waves of poaching, the elephant populations most under pressure include African forest elephants and various Asian elephant species- species which spend most or all of their time under deep cover, making counting from space virtually impossible.

For these and other reasons, this is unlikely to be applicable to many other parts of the world. Having said that, it should make counting elephants far simpler. If it becomes simply a question of taking high-resolution satellite imagery of any elephant habitat in Savannah Africa, then we can keep them more accurate count, and likely spot poaching faster than we currently can. 

One question I would have is, given the low resolution of the grey ‘blobs’ that are counted as elephants, presumably we cannot tell if they are alive- though perhaps this could be dealt with by making a second pass a day later and taking the same shot. Only elephants in identical places 24-hours apart could be assumed to be dead.

While this method cannot be used on significantly smaller species, this method might be usable for counting hippos and open habitats rhino species (presumably animal such as black rhino that prefer dense Bush will once again be largely uncountable from this method.

Furthermore this method may prove more effective than one would think. 

  • Firstly due to the amount of meat and the ivory of an elephant they are often the most profitable species to be hunted. As such, running this process overprotected reserves on a regular basis could work as an alarm bell-if the elephants have started being hunted perhaps the other species are also being hit
  • West African rainforest do not run on and on unbroken forever. There are regular clearings or bai’s, these are incredibly important four elephant populations. By taking pictures of rainforest bai’s at regular intervals, we could once again keep an eye on the rainforest elephants. Certainly this would be a more complicated use and you would be looking at changes rather than actual number of elephants, but this could still make a significant difference.

In conclusion, the prospect of being able to do any significant survey of a wild population nearly by paying for detailed photography from a satellite, is a big change. It is likely to make a particularly big difference in areas that have been in the middle of fighting, as we can keep a closer eye on the impact hitting local species. While satellite surveys will never replace feet on the ground, there are likely to be large areas of conservation for which this could be a significant change.

Leave a Reply

Your email address will not be published. Required fields are marked *