An open, fast and flexible framework for analysing seismic data in Julia
Author anowacki
20 Stars
Updated Last
5 Months Ago
Started In
June 2018


Build Status Coverage status

Seis.jl is an open, fast and flexible framework for analysing seismic data in Julia.

For detailed instructions on using Seis, see the documentation via the links below.

Documentation Development branch documentation

This readme file gives a brief introduction to installing and using Seis.


At present, Seis is unregistered and it and its dependencies must be installed manually. This will change once Seis is registered in the General registry.

To install Seis, do:

julia> ] # As soon as you press ']', you enter Pkg mode...

(v1.3) pkg> add

Basic introduction

Seis is based around single traces of evenly-sampled continuous data where time is the independent variable. (In future we may support unevenly-sampled data and traces with gaps, and we plan to soon support frequency-domain traces.) Each trace holds its sampling interval (the property .delta) and a start time (.b). Traces have the type Trace.

Traces also optionally hold information about the event (.evt) and station (.sta) associated with this recording. Events and Stations are the corresponding types, which also contain useful properties.

For more information, see the docstrings for Trace, Event and Station. This is easily done in the Julia REPL like so:

julia> using Seis

julia> ? # As soon as you press '?', the prompt changes to 'help?>'

help?> Trace
search: Trace trace backtrace AbstractSet AbstractVector AbstractVecOrMat


  Evenly-sampled time series recorded at a single seismic station. The start
  time of the trace, in s, is in the b field, whilst the sampling interval, in
  s, is delta. The trace itself is accessed using the trace method, like


All three types above also hold a .meta property, which contains any extra metadata you want to associate with the trace, event or station.

Common conventions

f vs f!

Traces are mutable structs, and therefore for all functions which potentially modify a trace, there are two versions. Firstly, an in-place function (e.g., bandpass!) which as per Julia convention ends with an exclamation mark, and modifies the trace. Secondly, for convenience, there is always a copying version without the exclamation mark (e.g., bandpass) which returns a modified copy of the input trace.

Arrays of traces

There is no special type for holding multiple traces. Instead, we operate on arrays of traces. For instance, reading multiple traces from the same event, we can access all the station names like so:

julia> t = sample_data(:array);

julia> t.sta.sta
60-element Array{String,1}:

Or to get the full channel code from available header information:

julia> channel_code.(t)
60-element Array{String,1}:

Note the use of the broadcasted . operation (channel_code.(t)) which applied the 'scalar' function (channel_code) to each trace in the array t.


Currently, SAC data are read or written (and may be either bigendian (SAC/BRIS convention) or little-endian (usual IRIS SAC convention)), and miniSEED files are read.

Use the read_sac and write_sac functions for SAC-formatted IO, and read_mseed function for miniSEED reading.

Future work will add support for reading many more formats and format auto-detection.


Plotting is available when using Plots. Seis.jl makes use of RecipesBase.jl so that if you do not need plotting, the package does not introduce this dependency. If you do want plots, simply install Plots.jl by import Pkg; Pkg.add("Plots"). You then need to do using Plots when you want to start using Seis.jl's plotting routines.

By default, plotting functionality is not exported, so you must also do using Seis.Plot before the routines section (plot record sections) and hodogram (particle motion plots) are available. (plot to show traces is available as soon as Plots is loaded.)

As an example:

julia> using Seis, Seis.Plot, Plots

julia> t = sample_data(:array);

julia> section(t)


Record section

See the help for the Seis.Plot module for more information.

Basic processing

Basic time series processing of traces is possible using functions such as integrate, bandpass, remove_trend. See the online documentation for a full description of the functions available and how to use them.


At present, Andy Nowacki (@anowacki) is the primary maintainer of Seis.jl.


Contributions to Seis are most welcome. Please open issues in this repo with bug reports or feature requests. Pull requests for the same are also very welcome.