Avian Radar Project
Making Wind Power Wildlife Friendly

Avian Radar Project Home

Why Migration Is Important

Wind Power and Migrants

Avian Radar Units

Acoustic Monitors

Research Results

Current Research

Interactive Radar Location Map

Reports, Facts, and Accomplishments

Evaluating Radar Studies

Radar Team


USFWS National
Wind Energy Site

Midwest Wind Energy Home


Funding provided by GLRI Logo

Recommendations for Radar Studies

What should researchers know before conducting surveys


The purpose of our study is to examine migration patterns along the U.S. shorelines of the Great Lakes. The U.S. portion of the Great Lakes has approximately 5,000 miles of shoreline and with only two avian radar units, the USFWS cannot monitor activity at every location. The choices of where we monitored were based on many factors, including the ability to extrapolate our data to additional locations, suitability of sites in an area for the radar units (open areas free of buildings), and areas that needed specific migration questions answered. Early in the project, we gathered information for a full season at a site to establish a baseline but in later seasons we moved to multiple locations to cover more area and answer more questions.


We do not intentionally place our radar units to monitor specific proposed or existing wind projects. We believe that it is the responsibility of a developer to determine the site specific conditions for their proposed development and if it comes to our attention that this was not done adequately, we inform the developer and the relevant regulating agencies of our concerns. Our data can also be used by developers to investigate areas nearby locations they are examining in order to have a better idea of the biological activity in the area, before they are heavily invested in a location.


Our data from along the shorelines of the Great Lakes sampled up to approximately 2 miles offshore in our early seasons. With additional modifications, one of our radar units can now partially sample about 4-6 miles offshore when located 3/4 of a mile inland. We feel that we can extrapolate our data to other shoreline areas but it would be more difficult to extrapolate our data to sites that are offshore beyond these distances.


The radar team has often been asked to evaluate other radar studies conducted by consulting companies, wind developers, and other scientists. As a result, we have put together a list of a few major points that we think help to make credible conclusions from the data. Each individual study is different, but these points should apply to most studies. Working through the Service's local Ecological Services Field Office, our team may be available to to review your radar study, at any stage of the process from initial design to final reporting, depending upon staff resources.


Sampling Schedule

Migration occurs in pulses, occurring strongly on some nights and very little to not at all on other nights, sometimes right after one another. Due to this pulsed pattern, if sampling occurs infrequently, such as once a week, pulses can be missed entirely and this can lead to drawing the wrong conclusion about what is occurring in an area. Below on the left is a sampling regime that sampled with radar once a week for a full day. The results indicate that little migration was occurring during this time period. However, we actually sampled at that location for the entire time period. On the right is what actually occurred with the dates surveyed by the once a week method circled.


Intermittent SamplingContinuous sampling


This sampling schedule can apply to all studies of migration, from visual bird surveys, to banding stations, to radar surveys. The pulsed nature of migration means that numbers, species composition, and risk may all change drastically. Continuous sampling at one site may provide a better picture of migration than intermittent sampling at multiple sites.


Volume Correction

VSR Beam ShapeSampling using radar is biased due to the structure of the radar beam. The vertical radar beam is used for standardized counts as well as estimating flight heights of targets and thus the amount of potential risk within the rotor swept zone (RSZ; the height band through which the wind turbine rotors spin). As the beam emanates from the radar unit it expands and results in a survey volume near the radar unit that is much smaller than the sampled volume that is further away from the radar unit. The volume sampled below 200 m (approximate height of the RSZ) within the 1 km standard front is about 1% of the total volume sampled by the radar unit. Ignoring this difference in sampling effort can reduce the estimated risk at a site. The smaller volume can mean that fewer targets are counted at low altitudes, such as in the RSZ, due to reduced sampling effort (Schmaljohann et al. 2008). If a correction is not used to account for this difference, the data can suggest that low altitudes are not used when in reality, target density is as great or greater than it is at higher altitudes. Similarly, reporting the percent of targets that are below, in, or above the RSZ without accounting for the difference in
sampling effort among these categories could lead to underestimating
the risk to migrants.


Avoid Using Seasonal Metrics Alone

Migrant activity differs greatly between day and night, on different nights of the year, and even between the beginning of one night and the end of that same night. By using seasonal metrics such as the mean and median for measures such as target passage rates and flight height estimates the times with high risk can be masked by times with much lower risk. By examining the data in smaller blocks than the entire season, these times of higher risk can emerge and action can be taken to attempt to reduce this risk through curtailing turbines at certain times, under certain conditions, or ultimately not building them at that location. Seasonal metrics can be useful for some measures but often they need to be accompanied by metrics that examine the data at a finer scale.


Last updated: October 31, 2017