dartobsgen.ObsGenConfig#

class dartobsgen.ObsGenConfig(*, start=None, first_analysis=None, end, lat_min, lat_max, lon_min, lon_max, obs_types, assimilation_frequency=<factory>, output_dir='.', output_prefix='obs_seq', output_timestamp_format='%Y-%m-%d-{S}', obs_type_map=None)[source]#

Bases: object

Configuration for a dartobsgen run.

Analysis times run from first_analysis through end inclusive, spaced by assimilation_frequency. Each analysis time T produces one obs_seq file named for T and holding the observations in (T - freq/2, T + freq/2], which is DART’s convention.

Give exactly one of start or first_analysis; the other is derived and both are populated after construction. All fields are keyword-only.

Parameters:

  • start (datetime, optional) – Start of the model run — not the first analysis time. The model advances one assimilation period before the first assimilation, so first_analysis = start + assimilation_frequency. Natural when the observations exist independently of the model (real data from CrocoLake or NNJA): pick the time you initialize the model.

  • first_analysis (datetime, optional) – The first analysis time itself. Natural when the observation times are fixed by something else and you need the analysis times to land on them — in particular PerfectModelSource, where synthetic obs are placed at the valid times of the model states being interpolated, so the analysis times must match the model output times.

  • end (datetime) – Last analysis time (inclusive). Must be at or after first_analysis.

  • lat_min (float) – Latitude bounds in degrees.

  • lat_max (float) – Latitude bounds in degrees.

  • lon_min (float) – Longitude bounds in degrees (-180 to 180).

  • lon_max (float) – Longitude bounds in degrees (-180 to 180).

  • obs_types (list[str]) – Observation types to include. Accepts DART compound names (e.g. "ARGO_TEMPERATURE"), DART variable names (e.g. "TEMPERATURE"), or CrocoLake variable names (e.g. "TEMP").

  • assimilation_frequency (timedelta) – Width of each assimilation window, and the spacing between analysis times. Must be an even whole number of seconds so the half-width lands on an integer second. Default timedelta(hours=6).

  • output_dir (str) – Directory where obs_seq files are written. Created if absent.

  • output_prefix (str) – Filename prefix. Files are named {output_prefix}.{timestamp}.out.

  • output_timestamp_format (str) – Format string for the timestamp portion of the filename, applied to the analysis time. Supports all Python strftime codes and the special token {S} which is replaced with the zero-padded seconds-of-day (00000–86400), matching DART’s naming convention. Default: "%Y-%m-%d-{S}" → e.g. 2010-05-01-21600.

  • obs_type_map (dict or None) – Custom mapping that overrides or extends the built-in DEFAULT_OBS_TYPE_MAP in CrocLakeSource. Each key is an obs type name; each value is a dict with keys "crocolake_var" and "db_name" (None = all sources). Pass None to use defaults only.

Raises:

ValueError – If neither or both of start and first_analysis are given, if assimilation_frequency is not a positive even whole number of seconds, or if the run contains no analysis times at all (first_analysis > end).

__init__(*, start=None, first_analysis=None, end, lat_min, lat_max, lon_min, lon_max, obs_types, assimilation_frequency=<factory>, output_dir='.', output_prefix='obs_seq', output_timestamp_format='%Y-%m-%d-{S}', obs_type_map=None)#
Parameters:
Return type:

None

Methods

__init__(*[, start, first_analysis, ...])

Attributes

first_analysis

obs_type_map

output_dir

output_prefix

output_timestamp_format

start

end

lat_min

lat_max

lon_min

lon_max

obs_types

assimilation_frequency

start: datetime | None = None#
first_analysis: datetime | None = None#
end: datetime#
lat_min: float#
lat_max: float#
lon_min: float#
lon_max: float#
obs_types: list[str]#
assimilation_frequency: timedelta#
output_dir: str = '.'#
output_prefix: str = 'obs_seq'#
output_timestamp_format: str = '%Y-%m-%d-{S}'#
obs_type_map: dict | None = None#