NoisePy Documentation

NoisePy Documentation#

NoisePy is a software to compute large-scale cross correlations for HPC and Cloud infrastructure. The difference in using Noisepy for either infrastructure is the back-end data format that are either file-system (H5) or object-storage (npz/mseed) optimzed.

NoisePy also offers tools for ambient noise monitoring (velocity and attenuation) and for Earth imaging (measuring phase and group velocities).

NoisePy leverages several efforts published, please consider

  • Jiang, C., Denolle, M. 2020. NoisePy: a new high-performance python tool for ambient noise seismology. Seismological Research Letters. 91, 1853-1866. https://doi.10.1785/0220190364.

  • Yuan C, Bryan J, Denolle M. Numerical comparison of time-, frequency-and wavelet-domain methods for coda wave interferometry. Geophysical Journal International. 2021 Aug;226(2):828-46. https://doi.org/10.1093/gji/ggab140

  • Yang X, Bryan J, Okubo K, Jiang C, Clements T, Denolle MA. Optimal stacking of noise cross-correlation functions. Geophysical Journal International. 2023 Mar;232(3):1600-18. https://doi.org/10.1093/gji/ggac410

We gratefully acknowledge support from the Packard Foundation

NoisePy Workflow#

The data processing workflow in NoisePy consists of three steps:

  1. (Optional) Step 0 - Download: The download() function or the noisepy download CLI command can be used to download data from an FDSN web service. Alternatively, data from an S3 bucket can be copied locally using the aws CLI, or streamed directly from S3. For users who want to work entirely locally, this step prepares and organize the data in a DataStore.

  2. Step 1 - Cross Correlation: Computes cross correlaton for pairs of stations/channels. This can done with either the cross_correlate() function or the noisepy cross_correlate CLI command.

  3. Step 2 - Stacking: This steps takes the cross correlation computations across multiple timespans and stacks them for a given station/channel pair. This can done with either the stack_cross_correlations() function or the noisepy stack CLI command.

Data Storage#

NoisePy accesses data through three “DataStore” abstract classes: DataStore, CrossCorrelationDataStore and StackStore. Concrete implementations are provided for ASDF (H5), miniSEED, Zarr, TileDB, npy formats. This part is implemented in a separate IO package noisepy-io.

  1. [optional] data download: for users who want to work entirely locally. This step prepares and organize the data in a RawDataStore or MiniSeeDDataStore

  2. Cross correlations: data may be streamed from the DataStore, which can be hosted on the Cloud, pre-processing and cross correlations are done for each time chunk (e.g., one day for broadband data). Cross-correlations are saved for each time chunck in CrossCorrelationDataStore.

  3. Stacking: Data is aggregated and stacked over all time periods. Stacked data will be stored in StackStore.

Applications#

Monitoring#

NoisePy includes various functions to measure dv/v. The software will read the CrossCorrelationDataStore to aggregate and measure dv/v. The outputs are tabular data in CSV.

Imaging#

NoisePy includes functions to measure phase and group velocity dispersion curve measurements. The software will read the StackStore and ouput curves as tabular data in CSV.