How to Analyze 10 Years of Chinese Earthquake Data with Python

Data Acquisition

Download the past 10 years of earthquake data directly from the China Earthquake Network ( https://news.ceic.ac.cn/index.html?time=1637318776 ) and save the file locally.

Data Processing

1. Coordinate Conversion

The original latitude and longitude are in Baidu map coordinates, which need to be converted to Gaode (Amap) coordinates using the free Gaode API.

import requests
import pandas as pd
# Read data
df = pd.read_excel(r'eqList.xlsx')
longitude_list = []
latitude_list = []
# Baidu to Gaode conversion
for i, location in enumerate(df[['经度(°)', '纬度(°)']].values):
    location = str(location[0]) + ',' + str(location[1])
    url = 'https://restapi.amap.com/v3/assistant/coordinate/convert?'
    parames = {
        'locations': location,
        'coordsys': 'baidu',
        'key': 'YOUR_KEY',
    }
    r = eval(requests.get(url, params=parames).json()['locations'])
    longitude_list.append(r[0])
    latitude_list.append(r[1])
    print(f'\r{i+1}', end='')

df['经度(°)'] = longitude_list
df['纬度(°)'] = latitude_list

After execution, the dataset now contains Gaode coordinates.

2. Timestamp Processing

Convert the origin time to a datetime object, then extract month and hour for later analysis.

# Convert to datetime
df['发震时刻'] = pd.to_datetime(df['发震时刻'])
# Extract month (YYYY-MM)
df['月份'] = df['发震时刻'].apply(lambda x: str(x)[:7])
# Extract hour
df['小时'] = df['发震时刻'].dt.hour

3. Province/City Extraction

Use the Gaode reverse‑geocoding API to obtain province and city names from the converted coordinates.

citys = []
provinces = []
for i, location in enumerate(df[['经度(°)', '纬度(°)']].values):
    location = str(location[0]) + ',' + str(location[1])
    url = 'https://restapi.amap.com/v3/geocode/regeo?'
    params = {
        'location': location,
        'key': 'YOUR_KEY',
        'extensions': 'base',
        'batch': 'false',
        'roadlevel': 0,
    }
    r = requests.get(url, params=params)
    data = r.json()['regeocode']
    city = data['addressComponent']['city']
    province = data['addressComponent']['province']
    if len(city) == 0:
        city = province
    citys.append(city)
    provinces.append(province)
    print(f'\r{i+1}', end='')

df['城市'] = citys
df['省'] = provinces

The enriched dataset now includes accurate province and city information.

Statistics and Visualization

Only simple descriptive statistics are performed; no predictive modeling is attempted.

1. Earthquake Counts

Define a magnitude range of 2–4.6 as “moderate” earthquakes. In the last ten years there were 6,188 moderate events, averaging over 600 per year.

Define magnitude ≥4.7 as “destructive”. There were 505 destructive earthquakes in the same period, averaging about 50 per year.

2. Provincial Distribution (Destructive Earthquakes)

Top provinces with the most destructive earthquakes:

Xinjiang – 98

Tibet – 63

Yunnan – 47

Taiwan – 47

Sichuan – 45

Qinghai – 32

Gansu – 10

Jilin – 9

Inner Mongolia – 7

Guangxi – 4

Hubei – 3

Guangdong, Hebei, Guizhou, Chongqing, Heilongjiang – 2 each

Jiangsu, Fujian, Liaoning, Shaanxi – 1 each

3. City‑Level Distribution (All Earthquakes)

Among the 233 cities that experienced earthquakes, the highest counts are in Xinjiang (e.g., Kizilsu Kirgiz Autonomous Prefecture – 491) and Sichuan (e.g., Yibin – 237).

4. Spatial Visualizations

Scatter map of destructive earthquakes (latitude/longitude).

Heatmap of all earthquake locations.

Dynamic heatmap showing monthly distribution over the past ten years.

5. Supplementary Knowledge

Earthquake belt map based on magnitude ≥5.

Feel free to try the code yourself, and join the Python learning group for further discussion.