Noise Study at West Oakland BART Station
This research project was undertaken in Spring of 2016, with the aim of gaining a better understanding of the issues of noise related to the BART elevated light rail in the neighborhood of West Oakland. During community meetings for the West Oakland BART Station TOD project, we heard, frequently, that train noise was an issue for residents of this neighborhood. While large-scale, non-specific remedies are often suggested (i.e. “put the whole BART line underground”), we were interested in whether, by pinpointing areas of particular concern, a more manageable solution could be proposed.
At each reporting site shown on the map, a minimum of three trains were recorded (using a hand-held MP3 recorder), and their peak decibel level (A-weighted) was measured. The readings / recordings were taken during weekdays, mostly during the mid-morning / mid-afternoon. This data was then analyzed and used to produce the maps and diagrams shown here, using Grasshopper, a visual programming plugin for Rhinoceros 3D.
In general, the noise of BART trains is an ongoing topic of discussion within the region that it serves (the Bay Area of California). What makes this issue particularly difficult for the West Oakland community is the amount of trains that pass through the area. As shown in the above diagram (Image 1), on any given weekday, trains pass the West Oakland station 570 times. These trains run 21 ½ hours a day, from 4:30 AM to 1 AM. During peak times, as many as 42 trains will pass during a given hour. The average exposure time of samples in this study was 25 seconds overall, with an average of 10 seconds at peak decibel level. Using these estimates, during the busiest two hours of the weekday (from 7-8 am and 5-6 pm), train sounds are present for approximately 30% of the time, with 12% of this at peak decibel level.
The standard unit of measurement for sound is the decibel. Decibel readings can, however, be misleading – as they follow a logarithmic scale (similar to seismic measurements). As shown above (Image 2), for every 10 decibels of increase, the perception of sound doubles. This means that 110 decibels sounds twice as loud as 100 decibels, and so on. For this reason, the color scheme used for diagrams in this study reflects this scale, rather than a linear increase.
The loudest sound level recorded during this study was 108 decibels (DbA), at the corner of 7th Street and Peralta. This is approximately as loud as a very loud rock concert. 100 decibels is approximately as loud as power tools in a wood shop, while 95 decibels is about the level of a motorcycle engine. For more equivalencies, as well as an “Interactive Sound Ruler”, see the National Institute on Deafness and Other Communication Disorders website: https://www.nidcd.nih.gov/health/interactive-sound-ruler-how-loud-too-loud.
Putting aside considerations of “loudness” (decibels) for a moment, it can also be helpful to understand the qualitative aspect of environmental sound.
In talking to community residents – as well as experiencing the sound of BART trains firsthand – we learned that it was not just the loudness (decibels), but also the quality or character of the train sounds that is problematic. In order to understand this, it can be useful to think in terms of the frequency composition of a sound sample. The amount of different frequencies present in a given sound are what give it its character (which is why a violin sounds different than an electric guitar). When adjusting a home stereo, for example, you turn up the amount of 80 or 100 Hz in order to increase the sound of the bass (what this is doing is adding more of that particular frequency to the sound that you are hearing).
As shown in the above diagram (Image 4), sound becomes more or less tonal as frequencies are distributed more or less evenly across the spectrum. In terms of environmental noise, “tonal” noise (e.g. jackhammer, squeaking…) is generally considered to be more problematic than “broadband” noise (wind, road, etc.), as it is distracting, and, if occurring in certain areas of the spectrum, can interfere with speech intelligibility.
Frequency Diagram / Bar Graphs from Site
By charting the relative levels of different frequencies in a 5 second sample of each recording, we were able to pinpoint locations where the sound of the trains had a more “tonal” quality. In particular, the area where the track curves, around the intersection of Peralta and 7th street, shows three distinct bumps in the area of 400 – 450 hz, 800 – 850 hz, and 1300 – 1350 hz. It is interesting to contrast this with the sound of the trains in the area closer to the BART station itself, where the sound is less tonal and the frequency distribution is more even.
The area with the most tonal noise, as heard in the audio clip above, was at Peralta and 7th street.
Tonal Noise Map
What is interesting in looking at the frequency map (Image 6) above is that there are both similarities and differences with the decibel map. Similar to the decibel map (Image 3), the area around 7th Street and Peralta is problematic, and the area around the station itself is less so.
When looking at the tonal map overlaid on the decibel maps (Image 7), it can be seen that there are pockets where the sound is very tonal, but not very loud – for example, the area near Wood and 8th street. While the sound is generally more tonal closer to the tracks, there are also areas where a vertical pattern emerges – such as the area of low tonality in the center. It seems that the tonal quality of the noise is related to both the geometry of the tracks themselves, as well as the speed of the trains, and also the built environment (where there are buildings versus open spaces).
Results / Barrier Design
Perhaps the most interesting finding of this study was the degree to which the experience of BART sound can vary from block to block – both in quantitative (decibel level) and qualitative (spectral analysis) terms. Because of this, we wanted to explore a strategy which could respond to the differences in environmental noise from location to location.
One potential solution is shown [Images 8-10]. The geometry of this barrier is determined parametrically, driven by the decibel level and frequency composition data that was produced through this study. In areas where the decibel level is higher, the barrier height is increases. In areas where there is a highly tonal quality to the noise, the geometry of the barrier becomes more varied. In this way, a specific solution can be tailored to address areas of concern.