brews / erebusfall

Ice-volume correction to marine-isotope proxy records in Python.

erebusfall

Apply a simple ice-volume correction to marine-isotope δ18O and δD proxy records, in Python.

The correction uses the LR04 benthic stack from Lisiecki and Raymo, 2005. The stack is scaled so that the LGM-to-present change is assumed to be 1.0 ‰ in accordance with the pore-water estimate of Schrag et al. 1996. The package is adapted from the ice-volume correction in Tierney et al. 2017.

Example

Start by importing erebusfall, numpy:

import erebusfall as ef
import numpy as np

...and creating a hypothetical proxy time series...

age_ka = np.arange(0, 20, 1)
proxy = np.random.normal(loc=-2.0, size=len(age_ka))

Now we can plug this into ef.icevol_correction() with a few key options:

proxy_adjusted = ef.icevol_correction(age_ka, proxy, 
                                      proxytype='d18o', 
                                      timeunit='ka')

We first plug in the proxy age and proxy values. The proxytype='d18o' indicates that we're dealing with δ18O. We can set proxytype='dd' for a δD record. The timeunit argument indicates that age_ka is in thousands of years before present. There are also options for for "years BP" and "million years BP". The output from the function, proxy_adjusted, is the corrected isotope proxy as a numpy array.

See help(ef.icevol_correction) for more documentation.

Installation

You can install the package from PyPI with

pip install erebusfall

If you want to use conda:

conda install erebusfall -c sbmalev

Development and Support

Source code is hosted online under an Open Source license. Please feel free to file any bugs and issues you find.

erebusfall's Issues

Benthic stack data is read from path and not package resource

Benthic stack data is read from path and not package resource. Ideally we would read it in as a package resource so that it doesn't depend on getting internal package path correctly with os. This is in erebusfall.benthic_stacks.*.

Passed benthic stack age reference instead of copy leads to bad values

Passed benthic stack age reference instead of copy leads to bad values, especially when working on multiple cores.

Should clarify/correct how input is sorted by age

Should clarify/correct how input is sorted by age. Right now, the docstrings are unclear about this.

Need example data

We need at least one example dataset to use as a compelling example.