IES use acoustic sensor networks for wild animal counting

Institute for Environmental Solutions (IES) for wild animal counting use passive acoustic sensor networks (microphones) as well as other innovative technological solutions – drones, motion-activated camera traps and deer tracking with GPS transmitters. IES test the technologies in cooperation with SIA “Forest Owners’ Consulting Centre” and Latvian State Forest Research Institute "Silava".

Red deer, roe deer, elk and wild boar are four dominant even-toed ungulates species in Latvia. These animals have an important role in the ecosystem, but they can also cause damage to the agricultural and forestry sectors. Furthermore, annually 750 thousand vehicle collisions in Europe are reported with ungulates involvement, thus, indicating a problem related to safety on roads. IES researchers in cooperation with partners are developing innovative data-driven wild animal counting methodology to support decision-making on sustainable wildlife management.

IES researcher Alekss Vecvanags showcased the latest conclusions and described further steps of the research.


IES: In the previous research season (July – September 2020) you carried trials with passive acoustic sensor networks (microphones). How has the data collection using microphone networks progressed?

Alekss Vecvanags (A.V.): By mid-October 2020, we actively collected data using passive acoustic sensors or microphone networks. We already placed microphones in Rāmuļu territory in Latvia during summer, but active recording of animal sound started at the end of the August and continued throughout September and beginning of October. This was the most suitable time for animal sound recordings because of the red deer and elk mating season. Usually, ungulates are silent animals, therefore harder to track, but during mating season animals make sounds to entice females and scare off competitors. We have ended the data gathering campaigns with microphone technologies. Further, we are going to start to develop the automated processing of microphone data.



IES: What types of microphones did you try and how they differ?

A.V.: In the research are we placed tested two types of microphones: 3 SM4TS microphones with triangulation possibilities and 9 simpler microphones in a grid layout.

Triangulation microphones were placed approximately 300 m apart and formed in a triangle, then, synchronized with a GPS watch. Microphones capture animal sounds and if the premiss is that ground surface is a 2D plane by the time difference of the sound detected in each microphone, we can calculate the location of the animal with a potential error of just 10 – 20 meters. Precision of the calculation is affected by different factors – temperature, humidity, terrain.

Microphones in a grid layout we placed randomly  in the research area approximately 200-300 meters apart. Thus, we can cover large territory with the potential to hear animal mating calls. If these sounds can be detected in some microphones but not in others, we can try to determine the approximate location of the animal. This is done based on the sound intensity and location of microphones.



IES: Which of the microphones types has been more successful and why?

A.V.: We have ended the data gathering campaigns. The next step is the data processing. Nevertheless, we can already draw first conclusions. Each type of the microphones has its own pros and cons.

The main pros of the grid layout microphones is the price, they are much cheaper, so we can buy more and cover wider area. On the other hand, the cons are that these microphones are not so precise regarding the location determination of the animals. These microphones provide a general insight into animal activity in the research area, but do not provide information about separate individuals.

The pros of triangulation microphones are the ability to determine the location of animal, yet a higher number of triangulation microphones could provide an even more accurate result. More microphones would provide additional information, therefore, the calculation error would be smaller. Despite the accuracy, offered by triangulation microphones, one of the cons is the high price. These microphones are much more expensive than the other microphones.

IES: How did you choose the locations for placing the microphone networks?

A.V.: During the placement, selection process we had a great help from Gvido Prudņikovičs who is an expert and forest technician of the Rāmuļi territory. He is familiar with specifics of forest ecosystem, the selected research area, and animal activity in it. We needed to find territories with a high animal activity to gather a large amount of sound data with animal mating calls. This was important for two reasons: first, it allowed us to understand how these technologies work; second, provided a great amount of animal-generated sound data needed to develop the automatization of data processing. Data processing automatization is the next step of this research. We are planning to develop sound data sorting algorithm that would automatically recognize noises created by animals in sound spectrograms. Therefore, reducing the need for researcher’s manual work. The algorithm is trained by using a machine-training method that requires a large amount of data.

Spectrogram of red deer mating call in passive acoustic sensor data. Data: Institute for Environmental Solutions.

IES: Why the tests of microphone nets were carried out during the autumn?

A.V.: Microphone networks work differently than other technologies used in this research. For example, camera traps are located in the forest throughout the whole year. The difference is that camera traps are equipped with motion sensors that control when cameras are turned on and off. Thus, cameras do not waste energy unnecessarily. Microphones are not equipped with such sensors, so the data collection is ongoing. This creates the need for manual work – replacement of battery and data storage.

The aim of this research was to develop an automated and remote animal counting approach. We concluded that data gathering with microphones throughout the whole year is not practical, because on-site maintenance is necessary every 2 or 3 weeks. Therefore, we decided that periodic deployment of microphone networks is the most appropriate solution. And the best period for microphone placement is animal mating season when they make sounds to entice females and scare off competitors.



IES: Is it planned to repeat data gathering campaigns with passive acoustic sensor networks?

A.V.: Yeah, of course! This was just the first animal sound data gathering campaign in which we understood how these technologies work. Experience we gathered during year 2020 will help us to repeat a full data gathering campaign during next year's season.

IES: Please share the upcoming plans of this research.

A.V.: We are re-starting drone flights over the research area. Additionally, we are planning to extend camera trap network by placing 30 cameras in different forest types. Therefore, we will concentrate drone flights over these territories to compare the animal counting results for both technologies.

Moreover, drone flights will be carried over the red deer garden in Rāmuļi territory, because we know almost a precise count of animals there. That allows us to compare results of animal counting with drone technology and the total animal count in the fenced area.

The research is a part of the project “ICT-based wild animal census approach for sustainable wildlife management” (No. is part of European Regional Development Fund, 1.1.1 "Improve research and innovation capacity and the ability of Latvian research institutions to attract external funding, by investing in human capital and infrastructure" measure “Support for applied research”.

Find more about this project here.