Minimizing Road Accidents on Rainy Days

A research that investigates the impact of rainy days on road accidents and how to minimize it

Introduction

On a global scale, Singapore ranked fifth in the estimated road traffic death rate per 100,000 vehicles (OECD, 2019).

This ranking is alarming, considering Singapore's figure is much higher than that of geographically enormous countries like the USA, Canada and Australia.

Within Singapore, 50% of overall road fatalities are motorcyclists and pillion riders, and 50% of pedestrian fatalities are the elderly.

These statistics emphasize the urgent need for Singapore to address road safety comprehensively.

Problem Statement

Rainy weather can impede our navigation skills on the road, either due to reduced visibility or more slippery surfaces.

To what extent does weather play a part in road safety? How can we minimize road accidents stemmed from rainy weather?

Purpose

As a leader in the climate research space, DSI 39 spearheaded this research to explore the relationship between rainy days and the number of road accident casualties. Recommendations are proposed to the relevant goverment organizations based on the findings of the research. This research aims to discover:

The effect of rainy days on the number of road accident casualties (correlation between number of rainy days vs casualties per month)
Whether rainy weather aggravates road accidents (correlation between number of rainy days vs fatalities, number of rainy days vs injured per month)
The vehicle types that are more prone to road accidents (rank number of casualties by vehicle types)

Methodology

The research analysis spanned a ten-year period from January 2010 to December 2019.
Data sets were aggregated on a monthly basis over this ten-year period to neutralize the effect of irregular weather phenomena, such as the El Niño event in 2015-2016, and instances of intense rainfall/flooding in Jun 2010, Jan 2011, and Jan 2018.
Data from 2020 onwards were excluded due to the effects of the COVID-19 pandemic.
Data related to Personal Mobile Devices (PMD) were excluded since PMDs came into existence in Singapore only from 2017 onwards.

Data sets used

To investigate the correlation between rainy days and road accidents, we utilized the following datasets (click on the hyperlinks to access):

Number of Rainy Days per Month (Jan 2010 - Dec 2019)

121 rows, 2 columns; illustrates the no. of rainy days per month from Jan 2010 to Dec 2019
We chose to use the number of rainy days rather than the specific amount of rainfall because it provides a better indication of how road accidents may be affected by the frequency of rain.

Number of Casualties: Fatalities and Injuries (Jan 2010 - Dec 2019)

120 rows, 29 columns; illustrates the no. of road accident casualties fatalities and injured, as well as a breakdown of those by vehicle type
The total number of casualties (the sum of fatalities and injuries) was used as a relative measure of road accidents.
We also examined the total fatalities and total injuries to assess the severity of accidents.
To identify which vehicle types are more susceptible to road accidents, we tabulated the total number of casualties across each vehicle type.

Data dictionary

Feature	Type	Dataset	Description
total_rainfall	float	rainfall-monthly-total	Total rainfall in mm
date	object	raindays_by_mth & road_acc_by_mth	Date in YYYY-MM format
yr	integer	raindays_by_mth & road_acc_by_mth	Year in YYYY format
mth	integer	raindays_by_mth & road_acc_by_mth	Month in MM format
rain_days	integer	raindays_by_mth	No. of rainy days in a month
log_raindays	float	raindays_by_mth	Log (base-10) of rain_days
total_deaths	integer	road_acc_by_mth	Total road accident casualties fatalities
total_injured	integer	road_acc_by_mth	Total road accident casualties injured
total_cas	integer	road_acc_by_mth	Sum of total_deaths and total_injured
log_deaths	float	road_acc_by_mth	Log (base-10) of total_deaths
log_injured	float	road_acc_by_mth	Log (base-10) of total_injured
log_cas	float	road_acc_by_mth	Log (base-10) of total_cas
buses, cyclists_&_ pillions, goods_vans, lorries, motor_cars_and_station_wagons, motor_cyclists_&_ pillion_riders, pedestrians, pick_ups, tipper_trucks, trailers, others	integer	road_acc_by_mth	Number of casualties of the respective vehicle type

Conclusion

The correlation between the number of rainy days and total casualties was 0.73 (after removing the outlier, Feb 2014). This implies that rainy days have a significant impact on the number of road accident casualties.
The correlation of rainy days vs fatalities, and rainy days vs injured were 0.58 and 0.73 respectively. This shows that rainy days have a greater effect on injuries but not fatalities.
Fatalities were less likely affected by other factors, such as drink driving, failure to adhere to traffic rules, jaywalking etc (Motorist.sg, 2020).
Jan, Mar and Apr (Northeast Monsoon season with the exception of Feb being the dry period) consistently had high number of casualties with relatively high number of rainy days. Though articles on Singapore road accident causes stated only human error reasons (Direct Asia, 2022; Motorist.sg, 2018), the strong correlation suggests that rainy weather might have elevated those human error and resulted in road accident injuries.
The top three vehicle types that had apparently higher number of casualties were 1. motorcyclists and pillion riders, 2. motor cars and station wagons and 3. pedestrians.
During the Northeast Monsoon, the respective government organizations should practise more stringent road safety measures tailored to motorcyclists and pillion riders, motor cars and station wagons and pedestrians.

Recommendations

The table below proposes some targeted road safety measures:

Limitations

1. Accuracy of research
As this research uses an aggregated number of rainy days per month, there was no clear indication of which road accidents occurred on a rainy day. For a more accurate analysis, the total number of casualties on rainy days should be used.

2. Limited analysis on other factors
With a focus on analysing the impact of wet weather on road accidents, this research does not take into account other factors that could be the primary reason of road accidents. A seperate research looking into the common human errors should be conducted in order to implement more comprehensive road safety measures.

3. Lack of geolocation data
There is a lack of updated geolocation data that illustrates areas in Singapore that are more susceptible to road accidents. This is another variable worth investigating into to refine those road safety measures.

References

About

Minimizing Road Accidents on Rainy Days

accidents casualties rain road

Languages

Language:Jupyter Notebook 100.0%