.nodeKeep

cremons-et-al-2022

Node Root

Philip Linden

Casey Honniball

Daniel R. Cremons

root

Python reproduction of Cremons, Daniel R., “Software for Simulating Lunar Surface Hydration Measurements for Multispectral Lidar at 3 µm”. Zenodo, Jul. 01, 2022. doi: 10.5281/zenodo.6025377.

Software for Simulating Lunar Surface Hydration Measurements for Multispectral Lidar at 3 µm

Earth and Space Science - 2022 - Cremons.pdf

The DeSci node published as a reproduction of this work using Python instead of Matlab. The goal is to reproduce the figures and results from the data published by Cremons, et. al that accompanied their paper.


The goal of this project is to learn by doing, and to reproduce the figures in a manner than is more broadly accessible. Comments and opinions expressed in the code and supporting text in this notebook are my own.


- Philip Linden, MoonDAO


> Cremons, D. R., &amp; Honniball, C. I.(2022). Simulated lunar surface hydrationmeasurements using multispectral lidarat 3 μm. Earth and Space Science,9, e2022EA002277. https://doi.org/10.1029/2022EA002277



This work was originally published in _Earth and Space Science in 2022_(doi: 10.5281/zenodo.6025377). The manuscript, data, and Matlab code is included unaltered from the originally published materials in the `src` directory.


Figures and algorithms presented in this manuscript by Cremons, et. al. were reproduced with Python code written by Philip Linden. The reproduction is included in the `repro` directory.



Philip Linden independently reproduced the results and adapted code to Python for MoonDAO as a Creative Commons Non Commercial work.


> Figure 1. (a) Regolith and pyroxene spectra as described in Table 1.


This figure was reproduced in Code Cell [4] of the IPy notebook.

[results.ipynb](#/code/3e71b606-f76a-497e-90df-f3d6011ad09a?path=root%2Fcremons-et-al-2022%2Frepro%2Fresults.ipynb)

So far, so good. The imported datasets match what is shown in Figure 1a of the manuscript after limiting the x-axis to the spectrum of interest.

It's not clear to me what the units of reflectance are in this data, as it is unlabeled in the source data files. I looked around briefly in the source dataset ([NASA RELAB](https://sites.brown.edu/relab/lunar-rock-and-mineral-characterization-database/)) and related publications ([source](https://www.researchgate.net/publication/253034096_RELAB_Reflectance_Experiment_Laboratory_A_NASA_multiuser_spectroscopy_facility)) but didn&#39;t see anything obvious. This does not undermine the analysis, though.

I observed the features described in the manuscript in the plotted data.


> Figure 1. (b) Laboratory MORB glass reflectance spectra used to simulate surface hydration in the reflectance simulations. The rawMORB spectra were normalized to each other at a wavelength 2.6 μm.



This figure was reproduced in Code Cell [10] of the IPy notebook.

[results.ipynb](#/code/3e71b606-f76a-497e-90df-f3d6011ad09a?path=root%2Fcremons-et-al-2022%2Frepro%2Fresults.ipynb)

The MORB reflectance over the spectrum of interest (1-4 micron) looks very different from the "processed" traces in Figure 1b. In the authors' code the MORB spectra are normalized to the 22 ppm reflectance spectrum. The "raw" heated MORB spectra also contains strange (to me) features below 1.5 microns.

The manuscript attributes these features to the presence of unquantified hydration and organic doublets in this part of the spectrum, and substitutes parts of the reflectance spectra with linearly fitted data.

The first data processing operation that is performed is to replace the spectra below 1.6 micron with another reflectance spectrum. It's a bit strange to _replace_ a portion of the observations with other observations, but in this case it feels analagous to an HDR composite photo.

Apparent artifacts had to be removed below 2.6 micron for the figure to match the manuscript. This is done by simply replacing the data below 2.6 micron with the least hydrated reflectance spectrum, 22 ppm.

Even the 22 ppm MORB spectrum has odd behavior between 1.0-1.5 microns, though. This region of the spectrum is replaced with the observations that focused on 1.0-1.6 microns (`Morb_D38A_Low_wavelength.txt`), which also has to be normalized to mesh with the 22 ppm reflectance spectrum at the boundary between those datasets.

While there is some pretty heavy processing done on the spectra below 2.6 micron, the focus of this study is the 3-4 micron range. The processing is done to show relative characteristics between spectra without including other features that may distract from the analysis itself.

> Figure 2. A Subset of interpolated MORB single scattering albedo spectra.The interpolated spectra correspond to total water values of (from low tohigh): 0, 150, 300, 450, 600, 750, 900, 1,050, 1,200 ppm, and 1,350, and1,500 ppm.



This figure was reproduced in Code Cell [16] of the IPy notebook.

[results.ipynb](#/code/3e71b606-f76a-497e-90df-f3d6011ad09a?path=root%2Fcremons-et-al-2022%2Frepro%2Fresults.ipynb)


> To create the spectral mixtures, each endmember reflectance spectrum was first converted to single-scattering albedo (SSA) using the methods of Hapke (2012) and least-squares minimization.


The Matlab implementation of this method had to be modified slightly in order to reproduce the author&#39;s results. The Matlab code that reproduced the author&#39;s findings is included here:

[Hydrated_Regolith_Spectra_Generator_and_Retrieval.m](#/code/5acc09bf-f463-42f3-a378-feb197a39379?path=root%2Fcremons-et-al-2022%2Frepro%2FHydrated_Regolith_Spectra_Generator_and_Retrieval.m)


> the SSA in the Hapkemodel by (Lucey, 1998)


This model was successfully reproduced in Python.

[hapke.py](#/code/0b998a5a-abae-4e9e-97df-df66be47e98b?path=repro%2Fhapke.py)


> (b) Comparison of interpolated MORB spectra (obtained from a linear ESPAT fit) and laboratory MORB spectra at the same total watercontent for each. The total water content for each laboratory spectrum is given in the legend.



This figure is reproduced in Code Cell [17] of the IPy notebook.

[results.ipynb](#/code/3e71b606-f76a-497e-90df-f3d6011ad09a?path=root%2Fcremons-et-al-2022%2Frepro%2Fresults.ipynb)


The complete dataset originally published with this manuscript in _Earth and Space Science 2022_ is included in this node in the `data` directory.


Simulated Lunar Surface Hydration Measurements Using Multispectral Lidar at 3 μm (Manuscript)

results.ipynb

Reproduced Results

Hydrated_Regolith_Spectra_Generator_and_Retrieval.m

This node is a reproduction of Software for Simulating Lunar Surface Hydration Measurements for Multispectral Lidar at 3 µm1 using Python instead of Matlab. The goal is to reproduce the figures and results from the data published by Cremons, et. al that accompanied their paper.

Reproduced figures and commentary are found in repro/results.ipynb as a Jupyter Notebook running Python 3. Details about the code and environment are found in pyproject.toml. The reproduction can be produced using a docker image prepared from this repository.

Software for Simulating Lunar Surface Hydration Measurements for Multispectral Lidar at 3 µm

Abstract