Analyzing Speed Limit Changes in the City of Los Angeles

This presentation is available at  https://arcg.is/11jqrO 

This study, prepared for the  Conference on Advancing Transportation Equity (CATE) , examines the geographic distribution and effects of recent speed limit changes in the city of Los Angeles. These were established by city ordinance 185922 in 2018​ ​. It defines just over 100 miles of ​increased​ and only around 10 miles of decreased speed segments. With Los Angeles committed to eliminating deaths in traffic by 2025, within its  Vision Zero program , these mainly increased speed changes might appear (and might be) counter-productive. The changes, however, were related to local speed enforcement laws and the arcane California vehicle code regarding road speeds and enforcement (see  this article  for a longer explanation).

Our analysis compares the location of the changed road segments in relation to disadvantaged communities (SB 535), and other socioeconomic variables, in a series of maps and graphs. It is based on publicly available data sources including the CA OEHHA (CalEnviroScreen), the Census Bureau and OpenStreetMap.

Additionally, the analysis establishes a baseline for before and after studies of the occurrence of traffic crashes along the impacted roads segments compared to the general trend in the city. This analysis includes comparisons of the rate and severity of incidents and number of pedestrians and bicyclists involved, and the proximity to the Los Angeles High-Injury Network and bike lanes (HIN - the LA Dept. of Transportation has identified streets with a high concentration of severe traffic collisions, with an emphasis with those involving pedestrians and bicyclists). However, it only gives an indication of any potential impact as the data available is limited to the timeframe 2019-2020, which is soon after the planned speed changes were implemented.

Speed limits, in California, are established by an Engineering and Traffic Survey (E&TS) according to vehicle code section 223493​ ​, considering: a) prevailing speeds, b) collision history and c) highway, traffic, and roadside conditions not readily apparent to the driver. 

The speed change listings, as set by  Council File No. 15-1006 Attachment A (p. 7-10) , were extracted and then geocoded using the Google Directions API via Python scripts.

• The resulting start/end points were used to identify the area of candidate roads in OpenStreetMap (OSM).
• Matching road features were selected based on road names. Some short segments were missed and had to be added manually.
• Additionally, Google Street View was used to verify if the planned change had actually been implemented by new signage.

The following illustrates the process for identifying the Amalfi Dr segment.

The distribution of changes per speed categories are shown in the following graph:

This matching process and the remaining analysis was performed using the Google BigQuery cloud platform, which provides standard SQL geography functions as well as a direct access to OSM.

 CalEnviroScreen 3.0  (CE) identifies California communities that are most affected by many sources of pollution, and where people are often particularly vulnerable to pollution’s effects. It uses environmental, health, and socioeconomic information to produce scores for every census tract in the state.

• A set of 11 relevant  CE indicators  and other demographic variables were chosen as the basis for comparing the speed change locations in relation to the city as a whole.
• The length of each road segment intersecting with the corresponding CE polygon features was calculated.
• The resulting weighted scores/values were calculated using the following SQL query (simplified):

SELECT Sum(ce.CIscore * ST_Length(ST_Intersection(rd.geometry, ce.geog)) / rd.seg_totlen) FROM `calenviro` ce `roads` rd WHERE ST_Intersects(rd.geometry, ce.geometry) GROUP BY segment_id

The following illustrates the process for calculating a weighted score for the Amalfi Dr segment covering two tracts in roughly equal parts:

The  Transportation Injury Mapping System (TIMS)  collects and provides crash location data for the state of California.

• Crashes from 2015-2020 were used to calculate the number of crashes within 25 meters along the affected road segments.
• Crash frequency per mile was calculated using the following SQL query (simplified):

SELECT Sum(c.number_killed) / (ST_Length(r.geometry) / 1609) FROM `tims` c JOIN `roads_major` r ON ST_DWITHIN(c.geog, r.geometry, 25) GROUP BY c.ACCIDENT_YEAR

Similar frequencies were calculated for the city as a whole. For a relevant comparison based on the character of the involved segments, freeways and minor (residential) streets were not included in this tabulation. Instead only the following OSM highway tags were included: "trunk", "primary", "secondary", "tertiary", "trunk_link", "primary_link", "secondary_link", "tertiary_link".

The following graphs show the distribution for each indicator using box plots ( https://en.wikipedia.org/wiki/Box_plot ). Each dot represents a specific segment.

Click the arrows to step through the indicators:

Click the arrows to step through the indicators

The segments overlapping with  SB 535 Disadvantaged Communities  are:

• 24 of the 67 speed increases (36%)
• 7 of 12 speed decreases (58%)

The following chart shows the percentage difference for the indicators compared to the city median, where a value of 1 means no difference (0.5 means 50% lower), for the segments.

For example: the 2nd green bar shows that speed increase segments are located in white (non-hispanic) areas at a rate of nearly two times (200%) compared to the city overall.

Click on the top labels to add/remove bars (Average Decrease, Average Increase, ..) for clarity:

 NOTICE from the Transportation Injury Mapping System (TIMS) on 6/15/21 :

"The provisional 2019 and 2020 collisions has been updated. The provisional data only contains records added in I-SWITRS website until June 11th, 2020 by CHP, and therefore should not be directly compared to the number of collisions in previous years."

Of the proposed changes, only the segments verified to have actual signage implemented by the latest available Google Street View imagery were included:

• 59 of the 67 (88%) speed increases
• 8 of 12 (67%) speed decreases

The following graph compares the frequency of crashes per mile for the selected categories, locations and year. For example: the highest crash rate was for "Pedestrians injured" in areas with speed decreases (middle section) at a rate of nearly 9 occurrences/mile in 2019.

Segments aligned with the  High-Injury Network (HIN)  road segments are:

• 5 of the 67 speed increases (7 miles)
• 4 of 12 speed decreases (5 miles)

Segments aligned with Bikeways (class 2 and 3) are:

• 22 of the 67 speed increases (29 miles)
• 2 of 12 speed decreases (3 miles)

From an equity perspective, speed limit changes should be made to improve traffic flow as well as safety for all transportation modes. In line with the goals of Vision Zero, severe crashes should be eliminated. Including speed changes (mainly reductions), other measures such as traffic calming and road safety improvements, such as separated paths for bikes and pedestrians, have a positive impact on safety, health and general mobility for all people.

Given the  dramatic increase of traffic deaths in Los Angeles  over the past decade, raising speed limits may seem strange and the path to decision rather opaque. A series of legislation changes need to happen to make it easier to implement multi-modal traffic safety projects, such as those mentioned above and promoted by agencies such the  Los Angeles Department of Transportation .

The characteristics of the affected census tracts, especially for speed decreases, are fairly well distributed among disadvantaged communities with the exception of predominantly African American areas being clearly underrepresented.

• The main pattern standing out is the occurrence of speed increases in areas, skewing towards more affluent white population areas. Also, the speed decrease locations show a similar albeit less strong pattern.
• As mentioned above, areas with a high African American populations are strongly underrepresented for both speed change categories.
• The hispanic population is also underrepresented when it comes to speed increases.

One could question if it is a bad thing to be underrepresented within the speed increase category. But this could mean that speeds are already set high in these areas, or that traffic density is too high for changes to be made. This would be an interesting detail to examine for further analysis, but hindered by the lack of current and historical speed and traffic volume data in the open data realm.

The location of segments relative to the HIN corresponds to speed increase / route avoidance and speed decrease / preferred route segments respectively. However, unprotected bicycle lanes are aligned with speed increase segments in some notable instances including: Venice Bl, Chandler Bl, Winnetka Av, Wilbur Av, S Brand Bl, Huntington Dr and S Sepulveda Bl for a total of approximately 29 miles or approximately 30% of the total length.

Given the incomplete TIMS crash data for 2019 and 2020, and the short time span since implementation, it is premature to make any generalizations regarding the potential effect of the speed changes that otherwise generally show a decreasing trend.

Yet, the higher crash frequency for the speed decrease segments stand out. However, given the limited set of segments (9 miles) and the potential for the results being skewed by the four segments (6 miles) in the downtown area, this is possibly misleading.

Curiously, speed increase segments, on the other hand, show a slightly lower rate of crashes compared to the city overall.

Going forward, as more accurate crash data becomes available, we recommend that more study is done to examine changes for speed change segments, particularly for the timeframe around 2019 when the changes occurred. This methodology could also be adapted to study the effects of other road safety implementations, such as separated paths for bikes and pedestrians as well as safer road crossings and intersections.

Please check our website  https://dcrdesign.net   for any updates on this subject.