During my summer internship, I tried to read the station observational data from the Atmospheric Science Research and Application Databank(大氣科學研究與應用資料庫，或稱大氣水文資料庫). However, I found it is difficult to access, so I fix some bug and published the code I used.

By this reader, you can:

• Quick load ASRAD data with mutil-threads
• Useful format for timeseries analyze

Since this class is inherited from pandas.DataFrame, you can use its functions.

• NumPy 1.26.2 or higher
• pandas 2.1.4 or higher
• python 3.12.8 or higher

1. Download the ASRAD_reader.py, and put into same folder with main.py
2. Import this reader as a normal libraries, see below:

   from ASRAD_reader import NanMode
   import ASRAD_reader as Reader

Use example below to input the file.

file_path = "datas/20029999_cwb_hr/20021099.cwb_hr.txt"
df = Reader.DataSet.read_file(file_path)

The mode parameter, is a enumeration named NanMode, which defines what values should be treated as NaN (Not a Number). There are 4 specific mode: ObsEmpty, AllEmpty, NotInObs, and AllValue. Default is NotInObs.

df = Reader.DataSet.read_file(file_path, mode = NanMode.AllValue)

The drop_nan parameter, if set to True, rows containing NaN values will be removed from the DataFrame. Defaults to False.

df = Reader.DataSet.read_file(file_path, drop_nan = True)

The is_utf8 parameter, specifies whether the file is encoded in UTF-8 (True) or Big5 (False).

df = Reader.DataSet.read_file(file_path, is_utf8 = True)

The mode and its special value:

Example below:

folder_path = "datas/"
df = Reader.DataSet.read_folder(folder_path)

Other parameter of read_folder:

• mode, optional, default is NanMode.AllEmpty.
• max_threads, optional, default is 4.
• drop_nan, optional, default is True.

Or only read specific columns:

folder_path = "datas/"
df = Reader.DataSet.read_folder_selected(folder_path)

Other parameter of read_folder_selected:

• mode, optional, default is NanMode.AllEmpty.
• max_threads, optional, default is 4.
• drop_nan, optional, default is True.
• selected_cols, optional, default is ["TX01", "PP01", "PS01", "RH01", "WD01", "WD02"].
• station_number, optional, default is 467490.

1. To normal .csv via DataFrame function:

   df.to_csv("datas.csv")

   For detail, please refer to pandas document.
2. To special .csv, which could load by this reader

   df.to_datasets_csv("datas.csv")

   And use the command below to load back.

   df = Reader.DataSet.read_dataset_csv("datas.csv")