Preface

Besides traffic counts, Project Amaterasu also uses weather data from Taiwan’s Central Weather Administration to enhance predictions.

Preparation

Identify Data Source

The CODiS platform provides historical observations. We explore how it retrieves and displays data in order to automate downloads.

Select a station (e.g., Taipei) and inspect network requests when querying hourly data.

A pop‑up window shows the results:

By examining the network panel we find a URL pattern that accepts parameters such as station ID and date. With this information we can programmatically fetch CSV files.

Implementation

Use Python with requests and csv to download and save the data:

1
2
3
4
5
6
7

import requests

URL = "https://e-service.cwb.gov.tw/HistoryDataQuery/csv"
params = {"station": "C0A9C0", "start": "2023-10-01", "end": "2023-10-31"}
resp = requests.get(URL, params=params)
with open('taipei_oct.csv', 'wb') as f:
    f.write(resp.content)

Automate multiple requests by iterating over station IDs and dates.

Summary

By reverse‑engineering CODiS network calls, we can systematically gather weather features for Project Amaterasu.