Predicting bird migration with radars at offshore wind farms

On behalf of the Offshore Wind Energy program, Rijkswaterstaat Zee en Delta develops ecological guidelines for the construction of offshore wind farms. For example, turbines must stop during major bird migrations. To make this feasible, the University of Amsterdam is developing a predictive model for bird migration. The Maritime Information Provision Service Point (MIVSP) of Rijkswaterstaat  provides the necessary data with a special bird radar. "It's incredibly fun to develop new knowledge this way."

The Maritime Information Provision Service Point (MIVSP) of Rijkswaterstaat places various sensors on transformer platforms of grid operator TenneT and on wind turbines of offshore wind farm owners. This is to collect as much interesting North Sea data as possible and then share it with interested parties. Meteorological sensors help to better predict the weather at sea, nautical sensors promote safe and smooth shipping, and ecological sensors map marine life. For instance, there are microphones that detect the ultrasonic sounds of bats and radars that see passing birds.

A radar image of a bird migration on April 1, 2020.

Ecological Guidelines for Wind Farms

Rijkswaterstaat Zee en Delta is one of the parties interested in these bird radars. "On behalf of the Offshore Wind Energy program of the Ministry of Economic Affairs and Climate Policy, we develop ecological guidelines for the construction of offshore wind farms," explains ecological coordinator Martine Graafland of Rijkswaterstaat Zee en Delta. "Initiators of such a wind farm must meet a number of conditions, described in so-called plot decisions. These conditions include noise production, appearance—such as lighting—and ecology. The impact of the wind turbines on marine mammals, fish, bats, and birds must be minimized."

Good Advice Requires Research

"We provide advice on how to implement mitigating measures for ecology," continues Graafland. "But research is needed first. One of the measures is to stop the wind turbines during massive bird migrations so that these birds do not collide with the rotating blades. To give good guidance for this, we first need to know how many birds migrate over the sea and when." For that research, MIVSP's radars are very useful.

Time Needed to Stop Wind Farm

Initially, the idea was to automatically switch off the wind turbines as soon as birds appear on the radar. But it's not that simple. Graafland: "To keep the energy supply constant, wind farms cannot just be turned off. Wind farm owners and TenneT need to take measures 48 hours in advance to guarantee the energy supply. Therefore, it is useful to know in advance when massive bird migrations will take place."

Developing a Bird Migration Model

To make such predictions, a model of bird migration at rotor height at sea is needed. For this, Rijkswaterstaat Zee en Delta consulted the University of Amsterdam (UvA). They had previously created a bird migration model for the air force: FlySafe. Graafland: "That was made for the high altitude at which planes fly. Unfortunately, the air force's radars cannot look into the wind farms. And since there are few to no places at sea where we can hang radars, we wanted to use radars in the wind farms themselves. Both on the TenneT platforms in the middle of the park and on turbines at the edge. So you can see what is coming."

Those radars are provided by MIVSP. "To detect both the height and location of the birds, we install a horizontal and a vertical radar with a large range: about 6 kilometers," explains technical manager Paul Scharrenburg. "These are radars like you often see rotating on ships, from the Japanese brand Furuno."

Automatically Interpreting Data

"The trick is not in the radar itself, but in the system integration," adds ecological sensors product manager Joris Diehl. "For this, we use software from Robin Radar. This software interprets the radar data. A radar does not see specific birds but moving objects. The software classifies those objects based on their mass and speed. There are different categories, including airplanes, bird flocks, and three sizes of birds: small—like songbirds, medium—like lapwings, and large—like geese. We are still looking into whether we can also record bird sounds with a microphone to determine the exact bird species, but that is very complicated offshore."

Data from sea to land

The software interpretation takes place at sea. The radars collect too much data to send all the raw data to land. Diehl: "That would mean you are streaming videos 24/7. Even now, the radar fills multiple tables with new data every 1.3 seconds. Every 10 minutes, the system makes a copy of this and sends it to Rijkswaterstaat's data center."

Tracking Bird Migration Live

From the data center, the data goes to stakeholders. Diehl: "Currently, this includes the UvA for the bird migration model and Bureau Waardenburg for a study to determine to what extent birds that live in and around the wind farm year-round, such as gulls, avoid wind farms and turbines. But the system is set up so that any interested party can get the data. For example, for other university research programs. Additionally, an application has been developed to track bird migration live—with a 5-second delay. You see the different sizes of birds moving across a map as different colored dots."

Everything is New

Besides the radar data, MIVSP includes the measurements from a weather station in the data for the researchers. The weather conditions at sea influence the birds' flying behavior. "The interesting thing is that this system is completely new," says Diehl. "As a result, we have specific requirements for Robin Radar. For example, managing and sending the data works differently for us than usual. We also continue to develop the system based on the experiences of the UvA researchers."

Testing at the Offshore Expertise Center

"Changes in the system are first implemented at our test location in the Offshore Expertise Center in Stellendam," emphasizes Scharrenburg. "There is a 1-to-1 copy of the mast with sensors as we use on the transformer platforms. This allows us to test everything in advance as if we were offshore. New components are first installed on the test mast, and software updates are thoroughly tested in advance. This way, we know everything works well when we go to sea. You can't just go offshore to adjust something. Moreover, offshore work requires more time and coordination with many different parties."

Diehl adds: "To ensure data quality, we validate each bird radar on-site to make small improvements to the software. We also monitor all servers and incoming data daily. This ensures that the correct data always comes in and goes to the right user."

Surprising Result

The first research results are now known. Graafland: "The surprising thing is that migratory birds fly at different altitudes in autumn and spring. Especially in autumn, birds fly more at rotor height than previously thought. It is challenging and exciting to work with completely different sensors than we are used to at Rijkswaterstaat. I find it especially fun to develop new knowledge this way."