Skip to content

time_utils

Time parsing utilities.

parse_ioapi_times(yyyymmdd, hhmmss)

Vectorized parsing of IOAPI (CMAQ/CAMx) TFLAG dates and times.

Parameters:

Name Type Description Default
yyyymmdd ndarray or array - like

Array of dates in YYYYDDD format (Julian day).

 required
hhmmss ndarray or array - like

Array of times in HHMMSS format.

 required

Returns:

Type Description
ndarray

Array of datetime64[ns] objects.

Examples:

>>> parse_ioapi_times([2023001], [120000])
array(['2023-01-01T12:00:00.000000000'], dtype='datetime64[ns]')
Source code in monetio/readers/time_utils.py 
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
def parse_ioapi_times(yyyymmdd, hhmmss):
    """
    Vectorized parsing of IOAPI (CMAQ/CAMx) TFLAG dates and times.

    Parameters
    ----------
    yyyymmdd : np.ndarray or array-like
        Array of dates in YYYYDDD format (Julian day).
    hhmmss : np.ndarray or array-like
        Array of times in HHMMSS format.

    Returns
    -------
    np.ndarray
        Array of datetime64[ns] objects.

    Examples
    --------
    >>> parse_ioapi_times([2023001], [120000])
    array(['2023-01-01T12:00:00.000000000'], dtype='datetime64[ns]')
    """
    y = np.asanyarray(yyyymmdd)
    h = np.asanyarray(hhmmss)

    # 1. Extract components using math
    years = (y // 1000).astype(int)
    days = (y % 1000).astype(int)

    hours = (h // 10000).astype(int)
    minutes = ((h // 100) % 100).astype(int)
    seconds = (h % 100).astype(int)

    # 2. Convert to nanoseconds since epoch
    # We use pandas to handle the variable year starts (leap years)
    unique_years = np.unique(years)
    year_to_start = {
        y_val: pd.to_datetime(str(y_val), format="%Y").to_datetime64() for y_val in unique_years
    }

    year_starts = np.array(
        [year_to_start[y_val] for y_val in years.ravel()], dtype="datetime64[ns]"
    ).reshape(y.shape)

    # 3. Combine using datetime arithmetic
    # Julian day 1 is the first day of the year (0 offset)
    day_offset = (days - 1).astype("timedelta64[D]")
    time_offset = (
        hours.astype("timedelta64[h]")
        + minutes.astype("timedelta64[m]")
        + seconds.astype("timedelta64[s]")
    )

    # Force ns to match project expectations and avoid discrepancies with Pandas 3.0+
    return (year_starts + day_offset + time_offset).astype("datetime64[ns]")

parse_wrf_times(times_arr)

Vectorized parsing of WRF/RAQMS character array or string times.

Parameters:

Name Type Description Default
times_arr ndarray or array - like

Array of times as strings or character arrays.

 required

Returns:

Type Description
ndarray

Array of datetime64[ns] objects.

Examples:

>>> parse_wrf_times(["2023-01-01_12:00:00"])
array(['2023-01-01T12:00:00.000000000'], dtype='datetime64[ns]')
Source code in monetio/readers/time_utils.py 
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
def parse_wrf_times(times_arr):
    """
    Vectorized parsing of WRF/RAQMS character array or string times.

    Parameters
    ----------
    times_arr : np.ndarray or array-like
        Array of times as strings or character arrays.

    Returns
    -------
    np.ndarray
        Array of datetime64[ns] objects.

    Examples
    --------
    >>> parse_wrf_times(["2023-01-01_12:00:00"])
    array(['2023-01-01T12:00:00.000000000'], dtype='datetime64[ns]')
    """
    t = np.asanyarray(times_arr)

    if t.ndim > 1:
        # It's likely a character array (..., DateStrLen)
        last_dim = t.shape[-1]
        if t.dtype.kind == "U":
            # Unicode: join along the last axis
            orig_shape = t.shape
            flat_times = t.reshape(-1, last_dim)
            s = np.array(["".join(row) for row in flat_times])
            s = s.reshape(orig_shape[:-1])
        else:
            # Bytes: use view if C-contiguous
            if t.flags.c_contiguous:
                s = t.view(f"S{last_dim}").squeeze(-1)
            else:
                orig_shape = t.shape
                flat_times = t.reshape(-1, last_dim)
                s = np.array([b"".join(row) for row in flat_times])
                s = s.reshape(orig_shape[:-1])
    else:
        s = t

    # Replace '_' with ' ' for pandas compatibility (WRF format: YYYY-MM-DD_HH:MM:SS)
    if s.dtype.kind in {"S", "a"}:
        s = np.char.replace(s, b"_", b" ")
        s = s.astype(str)
    else:
        s = np.char.replace(s.astype(str), "_", " ")

    # Force ns to match project expectations and avoid discrepancies with Pandas 3.0+
    return pd.to_datetime(s.ravel()).values.astype("datetime64[ns]").reshape(s.shape)

parse_yyyymmdd_hhmm(yyyymmdd, hhmm)

Vectorized parsing of YYYYMMDD and HHMM (or HHMMSS) dates and times.

Parameters:

Name Type Description Default
yyyymmdd ndarray or array - like

Array of dates in YYYYMMDD format.

 required
hhmm ndarray or array - like

Array of times in HHMM or HHMMSS format.

 required

Returns:

Type Description
ndarray

Array of datetime64[ns] objects.

Examples:

>>> parse_yyyymmdd_hhmm([20230101], [1200])
array(['2023-01-01T12:00:00.000000000'], dtype='datetime64[ns]')
Source code in monetio/readers/time_utils.py 
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
def parse_yyyymmdd_hhmm(yyyymmdd, hhmm):
    """
    Vectorized parsing of YYYYMMDD and HHMM (or HHMMSS) dates and times.

    Parameters
    ----------
    yyyymmdd : np.ndarray or array-like
        Array of dates in YYYYMMDD format.
    hhmm : np.ndarray or array-like
        Array of times in HHMM or HHMMSS format.

    Returns
    -------
    np.ndarray
        Array of datetime64[ns] objects.

    Examples
    --------
    >>> parse_yyyymmdd_hhmm([20230101], [1200])
    array(['2023-01-01T12:00:00.000000000'], dtype='datetime64[ns]')
    """
    y = np.asanyarray(yyyymmdd)
    h = np.asanyarray(hhmm)

    # Use float for intermediate to handle NaNs if present
    years = (y // 10000).astype(float)
    months = ((y // 100) % 100).astype(float)
    days = (y % 100).astype(float)

    # HHMMSS check: 2400 is the threshold (HHMM max is 2359)
    try:
        # Use nanmax safely for arrays, or standard max for scalars
        h_max = np.nanmax(h) if h.size > 0 else 0
        is_hhmmss = h_max >= 10000
    except (ValueError, TypeError):
        is_hhmmss = False

    if is_hhmmss:
        hours = (h // 10000).astype(float)
        minutes = ((h // 100) % 100).astype(float)
        seconds = (h % 100).astype(float)
    else:
        hours = (h // 100).astype(float)
        minutes = (h % 100).astype(float)
        seconds = np.zeros_like(h, dtype=float)

    df_dict = {
        "year": years.ravel(),
        "month": months.ravel(),
        "day": days.ravel(),
        "hour": hours.ravel(),
        "minute": minutes.ravel(),
        "second": seconds.ravel(),
    }

    # Use coerce to handle any invalid dates produced by math on NaNs
    res = pd.to_datetime(pd.DataFrame(df_dict), errors="coerce")

    # Return with original shape
    return res.values.astype("datetime64[ns]").reshape(y.shape)