Pre-processing¶

Before the image function annotates the waveforms, Qseek pre-processes them. pre_processing is a list of steps, applied in order. By default, the waveforms are resampled to 100 Hz and bandpass filtered between 0.5 and 30 Hz:

Default pre-processing

"pre_processing": [
  {"process": "resample", "sampling_frequency": 100.0},
  {"process": "bandpass", "bandpass": [0.5, 30.0]}
]

Each step applies to all stations, or only to the station codes in its stations list.

Resampling¶

The image function needs one sampling rate for all stations. If your stations record at different rates, resample them.

resample uses polyphase filtering and reaches any target rate, higher or lower.
downsample decimates by integer factors. It only lowers the sampling rate.

Resample

{
  "process": "resample",
  "stations": [],
  "sampling_frequency": 100.0,
  "n_threads": 8
}

Resample `pydantic-model` ¶

Bases: BatchPreProcessing

Resample the traces to a new sampling frequency.

Fields:

stations (set[NSL])
process (Literal['resample'])
sampling_frequency (PositiveFloat)
n_threads (int)
_thread_pool (ThreadPoolExecutor)

Validators:

validate_stations → stations

sampling_frequency `pydantic-field` ¶

sampling_frequency: PositiveFloat = 100.0

The new sampling frequency in Hz.

n_threads `pydantic-field` ¶

n_threads: int = 8

The number of threads to use for resampling.

stations `pydantic-field` ¶

stations: set[NSL] = set()

List of station codes to process. E.g. ['6E.BFO', '6E.BHZ']. If empty, all stations are processed.

get_subclasses `classmethod` ¶

get_subclasses() -> tuple[type[BatchPreProcessing], ...]

Returns a tuple of all the subclasses of BasePreProcessing.

filter_traces ¶

filter_traces(batch: WaveformBatch) -> list[Trace]

Selects traces from the given list based on the stations specified.

Parameters:

Name	Type	Description	Default
`batch`	`WaveformBatch`	The batch of traces to select from.	required

Returns:

Type	Description
`list[Trace]`	list[Trace]: The selected traces.

prepare `async` ¶

prepare() -> None

Prepare the pre-processing module.

Downsample

{
  "process": "downsample",
  "stations": [],
  "sampling_frequency": 100.0,
  "n_threads": 8
}

Downsample `pydantic-model` ¶

Bases: BatchPreProcessing

Downsample the traces to a new sampling frequency.

Fields:

stations (set[NSL])
process (Literal['downsample'])
sampling_frequency (PositiveFloat)
n_threads (int)
_thread_pool (ThreadPoolExecutor)

Validators:

validate_stations → stations

sampling_frequency `pydantic-field` ¶

sampling_frequency: PositiveFloat = 100.0

The new sampling frequency in Hz.

n_threads `pydantic-field` ¶

n_threads: int = 8

The number of threads to use for downsampling.

stations `pydantic-field` ¶

stations: set[NSL] = set()

List of station codes to process. E.g. ['6E.BFO', '6E.BHZ']. If empty, all stations are processed.

get_subclasses `classmethod` ¶

get_subclasses() -> tuple[type[BatchPreProcessing], ...]

Returns a tuple of all the subclasses of BasePreProcessing.

filter_traces ¶

filter_traces(batch: WaveformBatch) -> list[Trace]

Selects traces from the given list based on the stations specified.

Parameters:

Name	Type	Description	Default
`batch`	`WaveformBatch`	The batch of traces to select from.	required

Returns:

Type	Description
`list[Trace]`	list[Trace]: The selected traces.

prepare `async` ¶

prepare() -> None

Prepare the pre-processing module.

Frequency filters¶

Butterworth filters remove noise outside the frequency band of the earthquakes. With zero_phase, the bandpass does not shift the phase arrivals.

Bandpass

{
  "process": "bandpass",
  "stations": [],
  "corners": 4,
  "bandpass": [
    0.5,
    30.0
  ],
  "demean": true,
  "zero_phase": true
}

Bandpass `pydantic-model` ¶

Bases: BatchPreProcessing

Bandpass filter waveform data.

Fields:

stations (set[NSL])
process (Literal['bandpass'])
corners (int)
bandpass (Range)
demean (bool)
zero_phase (bool)

Validators:

validate_stations → stations
_check_bandpass → bandpass

corners `pydantic-field` ¶

corners: int = 4

Number of corners for the filter.

bandpass `pydantic-field` ¶

bandpass: Range = Range(0.5, 30.0)

Lower and upper corner frequency of the bandpass in Hz.

demean `pydantic-field` ¶

demean: bool = True

If True, demean the trace before filtering.

zero_phase `pydantic-field` ¶

zero_phase: bool = True

If True, apply zero-phase filtering.

stations `pydantic-field` ¶

stations: set[NSL] = set()

List of station codes to process. E.g. ['6E.BFO', '6E.BHZ']. If empty, all stations are processed.

get_subclasses `classmethod` ¶

get_subclasses() -> tuple[type[BatchPreProcessing], ...]

Returns a tuple of all the subclasses of BasePreProcessing.

filter_traces ¶

filter_traces(batch: WaveformBatch) -> list[Trace]

Selects traces from the given list based on the stations specified.

Parameters:

Name	Type	Description	Default
`batch`	`WaveformBatch`	The batch of traces to select from.	required

Returns:

Type	Description
`list[Trace]`	list[Trace]: The selected traces.

prepare `async` ¶

prepare() -> None

Prepare the pre-processing module.

Highpass

{
  "process": "highpass",
  "stations": [],
  "corners": 4,
  "frequency": 0.1,
  "demean": true
}

Highpass `pydantic-model` ¶

Bases: BatchPreProcessing

Highpass filter waveform data.

Fields:

stations (set[NSL])
process (Literal['highpass'])
corners (int)
frequency (PositiveFloat)
demean (bool)

Validators:

validate_stations → stations

corners `pydantic-field` ¶

corners: int = 4

Number of corners for the filter.

frequency `pydantic-field` ¶

frequency: PositiveFloat = 0.1

Corner frequency of the highpass in Hz.

demean `pydantic-field` ¶

demean: bool = True

If True, demean the trace before filtering.

stations `pydantic-field` ¶

stations: set[NSL] = set()

List of station codes to process. E.g. ['6E.BFO', '6E.BHZ']. If empty, all stations are processed.

get_subclasses `classmethod` ¶

get_subclasses() -> tuple[type[BatchPreProcessing], ...]

Returns a tuple of all the subclasses of BasePreProcessing.

filter_traces ¶

filter_traces(batch: WaveformBatch) -> list[Trace]

Selects traces from the given list based on the stations specified.

Parameters:

Name	Type	Description	Default
`batch`	`WaveformBatch`	The batch of traces to select from.	required

Returns:

Type	Description
`list[Trace]`	list[Trace]: The selected traces.

prepare `async` ¶

prepare() -> None

Prepare the pre-processing module.

Lowpass

{
  "process": "lowpass",
  "stations": [],
  "corners": 4,
  "frequency": 0.1,
  "demean": true
}

Lowpass `pydantic-model` ¶

Bases: BatchPreProcessing

Lowpass filter waveform data.

Fields:

stations (set[NSL])
process (Literal['lowpass'])
corners (int)
frequency (PositiveFloat)
demean (bool)

Validators:

validate_stations → stations

corners `pydantic-field` ¶

corners: int = 4

Number of corners for the filter.

frequency `pydantic-field` ¶

frequency: PositiveFloat = 0.1

Corner frequency of the lowpass in Hz.

demean `pydantic-field` ¶

demean: bool = True

If True, demean the trace before filtering.

stations `pydantic-field` ¶

stations: set[NSL] = set()

List of station codes to process. E.g. ['6E.BFO', '6E.BHZ']. If empty, all stations are processed.

get_subclasses `classmethod` ¶

get_subclasses() -> tuple[type[BatchPreProcessing], ...]

Returns a tuple of all the subclasses of BasePreProcessing.

filter_traces ¶

filter_traces(batch: WaveformBatch) -> list[Trace]

Selects traces from the given list based on the stations specified.

Parameters:

Name	Type	Description	Default
`batch`	`WaveformBatch`	The batch of traces to select from.	required

Returns:

Type	Description
`list[Trace]`	list[Trace]: The selected traces.

prepare `async` ¶

prepare() -> None

Prepare the pre-processing module.

Denoising¶

The DeepDenoiser neural network removes noise from the waveforms. It is slow; use it for noisy stations and run it on a GPU.

DeepDenoiser

{
  "process": "deep-denoiser",
  "stations": [],
  "model": "original",
  "torch_use_cuda": false
}

DeepDenoiser `pydantic-model` ¶

Bases: BatchPreProcessing

De-noise the traces using the DeepDenoiser neural network (slow).

Fields:

stations (set[NSL])
process (Literal['deep-denoiser'])
model (DenoiserModels)
torch_use_cuda (bool | str)

Validators:

validate_stations → stations

model `pydantic-field` ¶

model: DenoiserModels = 'original'

The model to use for denoising.

torch_use_cuda `pydantic-field` ¶

torch_use_cuda: bool | str = False

Use CUDA for the PyTorch model. A string selects the device.

stations `pydantic-field` ¶

stations: set[NSL] = set()

List of station codes to process. E.g. ['6E.BFO', '6E.BHZ']. If empty, all stations are processed.

get_subclasses `classmethod` ¶

get_subclasses() -> tuple[type[BatchPreProcessing], ...]

Returns a tuple of all the subclasses of BasePreProcessing.

filter_traces ¶

filter_traces(batch: WaveformBatch) -> list[Trace]

Selects traces from the given list based on the stations specified.

Parameters:

Name	Type	Description	Default
`batch`	`WaveformBatch`	The batch of traces to select from.	required

Returns:

Type	Description
`list[Trace]`	list[Trace]: The selected traces.

Pre-processing¶

Resampling¶

Resample pydantic-model ¶

sampling_frequency pydantic-field ¶

n_threads pydantic-field ¶

stations pydantic-field ¶

get_subclasses classmethod ¶

filter_traces ¶

prepare async ¶

Downsample pydantic-model ¶

sampling_frequency pydantic-field ¶

n_threads pydantic-field ¶

stations pydantic-field ¶

get_subclasses classmethod ¶

filter_traces ¶

prepare async ¶

Frequency filters¶

Bandpass pydantic-model ¶

corners pydantic-field ¶

bandpass pydantic-field ¶

demean pydantic-field ¶

zero_phase pydantic-field ¶

stations pydantic-field ¶

get_subclasses classmethod ¶

filter_traces ¶

prepare async ¶

Highpass pydantic-model ¶

corners pydantic-field ¶

frequency pydantic-field ¶

demean pydantic-field ¶

stations pydantic-field ¶

get_subclasses classmethod ¶

filter_traces ¶

prepare async ¶

Lowpass pydantic-model ¶

corners pydantic-field ¶

frequency pydantic-field ¶

demean pydantic-field ¶

stations pydantic-field ¶

get_subclasses classmethod ¶

filter_traces ¶

prepare async ¶

Denoising¶

DeepDenoiser pydantic-model ¶

model pydantic-field ¶

torch_use_cuda pydantic-field ¶

stations pydantic-field ¶

get_subclasses classmethod ¶

filter_traces ¶

Resample `pydantic-model` ¶

sampling_frequency `pydantic-field` ¶

n_threads `pydantic-field` ¶

stations `pydantic-field` ¶

get_subclasses `classmethod` ¶

prepare `async` ¶

Downsample `pydantic-model` ¶

sampling_frequency `pydantic-field` ¶

n_threads `pydantic-field` ¶

stations `pydantic-field` ¶

get_subclasses `classmethod` ¶

prepare `async` ¶

Bandpass `pydantic-model` ¶

corners `pydantic-field` ¶

bandpass `pydantic-field` ¶

demean `pydantic-field` ¶

zero_phase `pydantic-field` ¶

stations `pydantic-field` ¶

get_subclasses `classmethod` ¶

prepare `async` ¶

Highpass `pydantic-model` ¶

corners `pydantic-field` ¶

frequency `pydantic-field` ¶

demean `pydantic-field` ¶

stations `pydantic-field` ¶

get_subclasses `classmethod` ¶

prepare `async` ¶

Lowpass `pydantic-model` ¶

corners `pydantic-field` ¶

frequency `pydantic-field` ¶

demean `pydantic-field` ¶

stations `pydantic-field` ¶

get_subclasses `classmethod` ¶

prepare `async` ¶

DeepDenoiser `pydantic-model` ¶

model `pydantic-field` ¶

torch_use_cuda `pydantic-field` ¶

stations `pydantic-field` ¶

get_subclasses `classmethod` ¶