Birds in the city, what do buildings have to do with it?
Methods 1 — Urban Ecologies
Author
Zane Elias
Published
December 12, 2025
Focus
This report will focus on the simple question; Where do bird strike reports cluster in NYC, and how do nearby building forms (height proxy), development intensity, and street-tree density relate to collision intensity? NYC sits on the Atlantic Flyway - It is widely known that Migratory pulses are influenced by artificial lighting, reflective glass, and the structure of the built environment which can create dangerous conditions for birds moving through the city—especially at night and during seasonal migrations. With that in mind, this paper will explore whether or not building height is a predictive factor in bird strikes. If so, what can be done to stop it and where should efforts be focused? If not this report will attempt to further explore the urban fabric and consider alternative explanations to the phenomena of bird strikes around NYC.
Hypothesis
Collision intensity varies by seasons and month according to migration patterns.
Collision intensity will correlate positively with the built environment namely, taller buildings should show more cases of bird strikes.
Data sets used
DBird collision reports (CSV) — gathered from the org directly, partial resolution: point locations (approximate); Temporal resolution: observed date (2014–2025).
NYC Building Centroids (BES) (SHP) — building centroid points with BIN and height proxies
MapPLUTO (GDB) — parcel/building attributes for case study areas (borough-filtered).
2015 NYC Street Tree Census (CSV) — tree points and attributes.
Limitations
DBird data is crowd-sourced/partner-reported, location accuracy varies by a range of 33ft. Although data goes back to 2007, it was only filled in retroactively, and the data is sparse. Dbird was created in 2014, this is when we see an influx of data. So the dataset used is restricted temporally. Further Dbird does not capture the exact time of strike, or collision so a bird could have very well been injured or dead long before it was reported.
Beyond the birds, the tree census although updated in 2024 is from 2015 only highlighting some temporal mismatches between findings. Lastly, parcel/building proxies may not capture glass type, lighting, façade orientation, or interior light practices. While one could use data from NASA for ambient light time restrictions, and hurdles proved to difficult to access. That said, it would be worth exploring further when appropriate.
Figure 1: Reported bird strikes by year (Observed Date).
The above graph shows annual counts of reported bird strikes in NYC. Rather than representing a clear statistical distribution, this plot reflects varied temporal variation. Notably, in 2020 we saw an increase in reported strikes. This may be because there was less interruption to migration by way of human interference and pollution. Accompanied by an increase in outdoor activities like birding during the pandemic this may help explain the stark increase.
Code
fig("fig_dbird_month_by_year_lines.png")
Figure 2: Monthly bird strike reports by observed date. Each year is shown as a separate line.
Where there is no apparent pattern to the annual amount of strikes there seems to be a clear, consistent pattern in when strikes occur. Namely, we see spikes in spring (peaking May) and fall (peaking in October). As the climate shifts in NYC, respectively cooling or heating up we see strikes increase, likely this is tied to the accompanying migratory birds
Code
fig("fig_collisions_by_borough_centroids.png")
Figure 3: DBird collision reports summarized by borough using nearest building centroid.
Of all boroughs, Manhattan is by far and away responsible for the most bird strikes. This data intuits that it must be building height and density that is the causing lead of strikes, yet it is important to dig deeper.
Code
fig("fig_top_buildings_50ft_buffer.png")
Figure 4: Buildings with the highest number of bird strike reports within a 50-foot buffer.
The above graph displays the bird strike counts (within a buffer of 50 ft). Interestingly, while notable tall buildings like the one world trade center make the top 5 leaders in this class are not explicitly tall or towering buildings.
Code
fig("fig_spearman_height_vs_strikes_50ft.png")
Figure 5: Relationship between building height proxy (z_floor) and bird strike counts within 50 feet.
This chart shows the relationship between reported bird strikes within a 50-foot buffer of building centroids and a building height proxy (z_floor). Each point represents a building, with the y-axis displaying the number of reported bird strike incidents on a logarithmic scale to account for skewed count data. The distribution highlights a wide range of building heights with generally low strike counts across most buildings.
Code
fig("fig_hero_map_strikes_50ft_height_zoom.png")
Figure 6: Buildings with ≥3 nearby bird strikes. Color indicates strike count within 50 ft; size indicates height proxy.
This map displays buildings in New York City that are associated with reported bird strikes, measured within a 50-foot buffer around each building centroid. Color intensity reflects the number of recorded strike reports, while symbol size represents a proxy for building height (z_floor). Clusters of higher strike counts appear in specific areas of the city, suggesting localized concentrations of bird–building interactions that warrant closer examination.
(a)
(b)
(c)
(d)
(e)
(f)
(g)
Figure 7: Case study maps: trees (green), DBird strikes (red), MapPLUTO parcels (grey)
These case study maps display DBird collision points (red) overlaid on MapPLUTO parcel outlines (grey) within a buffered hotspot area. Street trees from the NYC Tree Census are shown in green, providing local environmental context around the collision locations. The circular buffer defines the spatial extent used for case-level analysis and comparison across sites.
Code
fig("fig_final_case_strikes_height_trees.png")
Figure 8: Case-level bird strike intensity vs typical building height, with tree density and built intensity context.
This figure presents a case level comparison of bird strike intensity normalized by area against typical building height, with tree density shown as a continuous color scale and built density represented by point size. Each point corresponds to a buffered case-study area, integrating multiple characteristics of the surrounding built and ecological environment into a single view. Critically, the pearson correlation reveals that both tree density, and building height together cannot predict whether or not strikes will happen. While the sample size is small, it could be used as an inference for later assumptions.
Discussion & Interpretation
Main findings
Across citywide and case-study scales, bird strike reports are highly unevenly distributed in space and time, with clear clustering around specific buildings and corridors rather than a uniform relationship to building height or tree density. Exploratory and case-level correlation analyses indicate that neither building height (measured by floor count proxies) nor street tree density shows a strong or consistent association with bird strike rates when normalized by area. Instead, bird strikes appear to concentrate in localized hotspots, suggesting that site-specific conditions and building-level features play a more important role than broad measures of urban form alone.
Unexpected results
A notable and unexpected result is the near-zero correlation between building height and bird strike counts, even when accounting for a 50-foot buffer around building centroids. Extremely tall buildings do not consistently correspond to higher strike counts, and many high-rise structures exhibit few or no recorded collisions. Similarly, areas with higher tree densities do not uniformly experience elevated strike rates, challenging the assumption that vegetation presence alone increases collision risk at the neighborhood scale. On the contrary, the data reveals that lower laying buildings, which also may be hidden behind vegetation, could lead to increased strike hot zones. That said, Queens, and Staten island - notoriously lower laying communities - recorded some of the lowest amount of strikes between 2017-2024. Whether this is an inconsistency in community gathered data is a persistent question. Another unexpected result in the data was the significant drop in bird strikes between 2020 and 2021. This could be for a multitude of reasons beyond the capacity of this report to discuss, but it was notable and worth looking into further.
Implications and meaning
These findings suggest that bird strike risk in New York City is less a function of generalized density or height and more likely the amalgmation of context specific architectural, material, and contextual factors such as glass treatments, noise, lighting conditions, façade orientation, and proximity to migratory pathways. From a planning perspective, this implies that targeted, building level, or community specific interventions may be more effective than broad zoning- or greening-based approaches. More broadly, the analysis highlights the importance of integrating fine-grained spatial data, multi-variate analysis and ecological context when evaluating human–wildlife interactions in dense urban environments.
Conclusion
This project examined patterns of bird–building collisions in New York City by combining DBird reports with spatial data on building form, parcel structure, and street trees. Through exploratory analysis, spatial joins, and case-study mapping, the analysis revealed that bird strikes are clustered in specific locations rather than evenly distributed across the urban landscape. The initial hypotheses followed intuitive logic; that taller buildings would definitively correspond with increased collision risk, multivariate and case-level analyses showed no strong or consistent relationship between bird strike rates and either building height or street tree density when normalized by area.
These findings matter because they challenge assumptions about urban form and wildlife risk, highlighting instead the importance of site-specific conditions and context aware, ecological design. The concentration of strikes around particular buildings suggests that architectural features, façade materials, lighting, and local ecological context may also play a more significant role than generalized measures of density or height.
From an urban planning, architectural and policy perspective, this implies that targeted mitigation strategies, site specific environmental design and community efforts (such as basic vetrenatrian and animal stewardship programs) at the building level may be more effective than broad, citywide interventions as was initially hypothesized. That is not to say that policy interventions which have taken place are worthless. All in all any interventions taken up to this point are win-win, unless austerity finds this spending unnecessary. Placing dots on a window for example is neither a loss for a tenet and its most definitely a win for a bird, not to mention a superstitious tenant afraid of birds hitting their windows. In all seriousness, the data reveals that bird strikes still occur, and there are seasonal patterns worth considering. For example, could the recognition of the seasonal strike patterns lead to New York’s implement of seasonal or outright bans on disruptive sounds, lighting and other pollution(s)?
Future work could be extended into other less dense yet still developed Ecologies, especially those edge environments such as urban wetlands, fields and shorelines which similarly show notable patterns as strike hotspots as revealed in my prior research at TidMarsh wildlife sanctuary. I would like to also extend this work to other countries where I am pursuing work on flora and fauna preservation and restoration of habitats. For example one can make a strong case for environmental interventions, if you can bolster already existing benefits of an urban farm with the fact that as green space is restored and increase bird strikes decrease. That said, if this report and work revealed anything it is that there is no one size fits all panacea of a solution.