Using the year (int) and fraction of year (float) to create datetime objects [NOTE: assumes that data is of daily frequency!]

calculate_datetime_from_fraction_of_year.py

from typing import Tuple, Any, Union, List
import pandas as pd
import datetime as dt 
from pathlib import Path 
import numpy as np


def find_nearest(array: Union[np.ndarray, List[float]], value: float) -> Tuple[Any, int]:
    # https://stackoverflow.com/a/2566508/9940782
    array = np.asarray(array)
    idx = (np.abs(array - value)).argmin()
    return array[idx], idx


if __name__ == "__main__":
    data_dir = Path("home/tommylees/Downloads/")

    df = pd.read_csv(data_dir / "slr_sla_nrs_free_txj1j2_90.csv", skiprows=5)

    # convert columns into years and frac of years
    years = [int(row[0]) for row in df["year"].astype(str).str.split(".")]
    frac_of_years = [float(f"0.{row[1]}") for row in df["year"].astype(str).str.split(".")]

    # calculate each day-of-year encoded as a fraction, to be used as a lookup list
    #  https://stackoverflow.com/a/56193202/9940782
    reference_fractions = [dt.timedelta(days=i) / dt.timedelta(days=365) for i in range(365)]

    # get the nearest match in the list of reference_fractions
    #  add 1 because of zero based indexing
    doys = [find_nearest(reference_fractions, value=f)[-1] + 1 for f in frac_of_years]

    # convert to datetime
    datetimes = [dt.datetime(year=year, month=1, day=1) + dt.timedelta(days=int(doy)) for year, doy in zip(years, doys)]

    # append back as new columns
    df["time"] = datetimes