Academic Journal of Science and Technology 
ISSN: 2771-3032 | Vol. 5, No. 2, 2023 

 

56 

Study	on	rainy	Weather	Situation	of	Highway	Based	on	
Historical	Rainfall	Classification	
Xin Zhang1, * 

1 School of Traffic and Transportation, Chongqing Jiaotong University, Chongqing 400074, China 
* Corresponding author: Xin Zhang (Email: zhang_xin_0824@163.com) 

 

Abstract: The traffic system consists of people, vehicles and roads, as well as the environment around the road during driving. 
Weather has always been an important factor that cannot be ignored along with the highway. Good weather will bring good 
vision conditions and psychological conditions to drivers, while bad weather conditions will directly or indirectly affect the 
driver's operation or vehicle power performance. As one of the most common weather, many people neglect the harm of rainy 
days to highway driving safety. When the rain reaches a certain level, it will lead to water on the road surface and reduce the 
friction coefficient of the road surface. If the driver drives too fast at this time, it will be more prone to deviation. When it reaches 
extremely bad conditions, it can even destroy the subgrade and pavement, which directly causes immeasurable safety hazards. 
Based on the rainfall data corresponding to the rainfall events in the traffic environment in the past few years, this paper proposes 
a research method for the characteristics of rainy days, which uses the fuzzy recognition method to classify the rainfall types of 
rainy days, to study the development trend of rainy days. After obtaining the development trend function of the historical type 
of rainy days, we can determine the type of the rainy day event and the specific time period of the type by comparing the closeness 
of the function of the current observation value with time and the historical function, so as to achieve the purpose of forecasting 
the development trend of the current weather event. Taking the change rate of rainfall change as the characteristic function, we 
can further quantify the change range of rainfall, the key indicator representing the intensity of rainy days. This strategy model 
is applied to expressway speed management, which can provide a good reference for traffic managers to make speed limit 
decisions, and can clearly decide when to cancel the speed limit strategy or upgrade to the road closure strategy. 

Keywords: Highway, Rainy weather forecast, Fuzzy recognition. 
 

1. Introduction 
Since the 21st century, with the rapid economic 

development of countries around the world, the construction 
of expressways has also been rapidly constructed. The 
coverage of expressways in some countries has already 
reached half of the national territory. The increase of 
expressway mileage means that the whole road network 
system will also span more areas. For some countries with 
large territory, such a state will inevitably lead to the 
expressway being covered by a variety of meteorological 
environments. As one of the most common weather 
phenomena, rainy days will not affect the traffic of 
expressway when the intensity is light. If it reaches a certain 
intensity, it will lead to a sharp increase in the probability of 
traffic accidents. Especially for China, China has a vast 
territory and a large terrain span. Although it is a country, 
highways in different places face different types of 
meteorological disaster risks. In addition, China's 
construction capacity has become more and more powerful 
with the development of social economy in recent years, and 
the total mileage of expressways across the country has 
continued to rise, reaching 169100 kilometers by 2021. 

How to reduce the impact of rain on highway traffic safety 
is a long-term research content in the field of transportation. 
From a macro perspective, some scholars analyzed and 
explored people's willingness to travel, choice of travel mode, 
expected speed of drivers, and probability of compliance with 
traffic speed limit measures under different weather 
conditions. Some scholars have also carried out the safety 
research of expressway in rainy days from the micro-
perspective of the reduction of friction factors on the 

pavement of expressway in rainy days and the reduction of 
reaction ability caused by the obstruction of driver's vision. 
Aguilar et al. [1] used the collision data of interstate highways 
to explore the factors affecting the collision frequency, with 
the focus on the impact of weather (precipitation, wind, etc.). 
The negative binomial model with random parameters is used 
to check the influence of the above variables on the accident 
frequency. Joshua et al. [2] showed that rainfall intensity has 
a great impact on the traffic capacity of the bottleneck section 
of the expressway. Soyoung et al. [3] verified the impact of 
rainy weather conditions on traffic operation, accident 
frequency and accident severity through microscopic 
simulation modeling. Oluyemi et al. [4] studied the traffic and 
rainfall data collected in four different places in Nigeria for 
eight weeks, and the results showed that the travel speed 
changed significantly between dry and rainy night conditions. 
The average driving speed of light rain, moderate rain and 
heavy rain at night decreased by 3.4%, 6.8% and 10.2% 
respectively. 

Although these studies are carried out on rainy days, they 
rarely serve the safety construction of expressway from the 
perspective of rainy day prediction. This paper will propose a 
method to analyze the rainy weather situation of expressway 
from the perspective of rainfall classification based on the 
rainfall data of historical rainy weather events, mainly using 
the fuzzy recognition method. 

2. Classification of Historical Rainfall 
Events Based on Fuzzy Recognition 

The weather in a region is changeable every day and seems 
irregular. But in fact, it can be seen from the meteorological 



 

57 

factors such as temperature, pressure and humidity that their 
generation, development and dissipation are regular. The 
weather conditions of a region for many years will also 
present certain regularity, which is reflected in the local 
climate to some extent. It must take many years to summarize 
the laws of climate. Therefore, for the research content of this 
paper, it is necessary to analyze the characteristics of rainfall 
data in the process of historical rainy events for many years. 

The method used here is called fuzzy recognition method. 
Fuzzy recognition refers to the identification of useful parts 
in a large number of complex information, that is, to 
recognize the relevance of the received information and past 
memory and experience, and to eliminate irrelevant 
information. It mainly follows two principles, one is the 
principle of maximum subordination, the other is the principle 
of proximity. If any element within the scope of study has a 
number corresponding to it, the number is called the fuzzy set 
within the scope of study, and this relationship is called the 
membership of the element to the number. When the number 
can be changed at will, this will form a membership function, 
and the maximum membership is based on this theory. The 
proximity principle measures the proximity between two sets. 

The principle of maximum membership is expressed by 
mathematical formula as follows: 

 
𝐴 𝑢 ⋁ 𝐴 𝑢             (1) 

 

Where, 𝑢   represents the primary color and 𝐴  
represents the category. 

The principle of proximity is expressed by mathematical 
formula as follows: 

 
𝑁 𝐴 , 𝐵 max

1≤j≤n
𝑁 𝐴 ,𝐵           (2) 

 

Where, n  represents the fuzzy set, 𝐵  represents the 
object to be identified, and 𝑁 𝐴 ,𝐵   represents the 
proximity. When set 𝐴 and set 𝐵 are closest to each other, 
𝐴 and 𝐵 belong to the same class. 

If you want to use python to implement the principle of 
maximum membership, the code example is as follows: 

 

 
Figure 1. The principle of maximum subordination python 

code example diagram 

If you want to use python to implement the proximity 
principle, the code example is as follows: 

 

 
Figure 2. Proximity principle python code example diagram  

 

The idea of using the fuzzy recognition method to deal with 
the rainfall data of the historical rainy day event is to divide 
the actual rainfall process into multiple time periods, and 
establish the pattern matrix of the rainfall event according to 
the percentage of the rainfall in each time period in the total 
rainfall. Then the actual indicators of each rainfall process are 
compared with this matrix, and the general types of rainfall 
events can be classified according to the proximity principle. 
After the rainstorm for many years is divided into several 
categories according to this method, the function of related 
rainfall and time can be further fitted to represent each type 
of rainfall. 

3. Matching Degree Between Current 
Monitoring Data and Historical 
Function 

First of all, it is necessary to judge the position of the 
current detected rainfall index data in the historical function 
curve. First, divide the rainfall of the historical rainfall type 
function curve to which the current rainfall data belongs into 
m time segments. The difference of rainfall in each time 
segment compared with the previous time can be expressed 
as: 

𝑥 𝐻 𝐻 （𝑖=1，2，…，𝑚）    (3) 

 
Where, 𝑥  represents the difference of rainfall in each time 

segment compared with the previous time; 𝐻  refers to the 
rainfall in the ith segment of the predicted rainfall event; 𝐻  
refers to the rainfall in segment 𝑖-1 of the predicted rainfall 
event. 

The change rate of rainfall difference in the two adjacent 
time periods in the history function can be further calculated 
to obtain a set of vector groups of rainfall difference change 
rate corresponding to the predicted value. The actual 
observation value is processed in the same way of time 
section division, and the vector group of rainfall difference 
change rate corresponding to the actual value at present is 
obtained by the same processing. The two can determine the 
actual trend and position of the rainfall time with the type by 
the progress calculation. 

 



 

58 

|𝑋 |
/

（𝑖=1，2，…，𝑚-1）    (4) 

 
Where, A represents the change rate of rainfall difference 

in the period B during the forecast rainfall event; C represents 
the total rainfall duration of the historical rainfall function. 
The change rate of rainfall difference of historical function 
and the change rate of rainfall difference of actual observation 
value are calculated according to the schedule: 

 

𝜎 1
1

𝑚 1
|𝑋′ | |𝑋 |  

（𝑖=1，2，…，𝑚-1）          (5) 
 

Where, 𝜎   represents the closeness between the 
predicted precipitation difference change rate and the actual 
observed precipitation difference change rate in the period 
𝑖+1, the closer to 1 means the closer, and the closer to 0 means 
the less close; 𝑋′   represents the change rate of the 
difference of the actual observed rainfall in the 𝑖+1 period. 
Select 𝑋  corresponding to the maximum closeness as the 
weight value 𝑎 of the current weather trend. A positive value 
indicates that the rainfall trend will increase, and a negative 
value indicates that it will decrease with the trend. For other 
weather types, the corresponding weather situation weight 
factor can be obtained by the same method. 

4. Establishment of Expressway Rainy 
Weather Situation Risk Model 

Next, we start to establish the expressway rainy day risk 
model. It is worth noting that this model is only applicable to 
the situation where rainfall data is detected. Moreover, the 
longer the time of falling rain is, the more conducive it is to 
predict the rainfall trend, because this will, to a certain extent, 
avoid the accidental error caused by too little data from the 
current situation. 

When the rainfall event reaches the local speed limit 
condition, this model starts to apply. The weight coefficient A 
obtained in the previous section can be used to express the 
fluctuation range of the following weather trend clearly and 
quantitatively. 

Before the rainfall reaches the predicted peak value, it is 
necessary to compare and judge the speed limit value of the 
current highway. If the weather trend that the speed limit 
strategy has reached the highest level is still deteriorating, it 

can be considered to take the temporary no-passing strategy 
for the current road instead of continuing to limit the speed. 
Because it is already in a particularly extreme state at this time, 
vehicles are vulnerable to rainstorm weather and safety 
accidents occur. Speed limit is meaningless. 

After the rainfall reaches the predicted peak, it is necessary 
to compare and judge the speed limit value of the current 
expressway. If the speed limit strategy has been reduced to the 
lowest level and the weather trend continues to improve, it 
can be considered to further increase the speed limit threshold 
for rainfall events or cancel the speed limit provisions. 
Because it is close to normal weather at this time, it is difficult 
for the vehicle to receive the influence of rainy weather to 
have a safety accident or the rainy weather has dissipated, and 
the speed limit is no longer meaningful. 

5. Conclusion 
In this paper, a method of expressway rainy weather 

situation prediction based on fuzzy recognition and 
classification of historical rainy weather time rainfall data is 
proposed. Taking into account the fact that the regional 
climate is not easy to change greatly all the year round, this 
method first uses the fuzzy recognition method to analyze the 
historical rainfall data, and obtains the classification of 
historical rainfall types and their corresponding functions. 
Then, the closeness of the current monitoring data to the 
historical data function is calculated to determine the type of 
current rainfall. After determining the rainfall type, the next 
possible trend of current rainfall can be judged. This model 
can effectively serve the safety management of expressway in 
rainy days. 

References 
[1] Aguilar, Cristopher, et al. "Analysis of factors affecting the 

frequency of crashes on interstate freeways by vehicle type 
considering multiple weather variables." Journal of 
Transportation Safety & Security 14.6 (2022): 973-1001. 

[2] Seeherman, Joshua, and Alexander Skabardonis. "Impact of 
Adverse Weather on Freeway Bottleneck Performance." 
Journal of Transportation Engineering, Part A: Systems 146.10 
(2020): 04020120. 

[3] G. O. Young, “Synthetic structure of industrial plastics (Book 
style with paper title and editor),” in Plastics, 2nd ed. vol. 3, J. 
Peters, Ed.  New York: McGraw-Hill, 1964, pp. 15–64. 

[4] Makinde, O. , and  J. Ben-Edigbe . "Effect of Night-Time Rain 
on Travel Speed at Two-Lane Highway without Lights." 
Transportation Research Procedia 48.1(2020):747-755.