Evaluating Characteristics of Potential Lunar Secondary Craters Extracted by a Semi-Automated Data Pipeline

Introduction

We are a team of primarily undergraduate student researchers, led by Dr. Martin-Wells, who are creating an IDL-based data analysis pipeline that will combine multiple existing techniques [e.g., 1-5] for classifying craters.  We aim to develop a transparent, accessible tool that can be used to sort primary and secondary craters on the Moon, even by novice crater classifiers.  The pipeline will mimic how an experienced crater classifier sorts primary and secondary craters “by hand,” but in a semi-automated pipeline that tracks the exact criteria used for classification and reduces the time needed to make these decisions.  This work focuses on how craters will be assigned a recommended classification based on the data that the pipeline will collect.

 

Background and Methods

Our IDL pipeline will extract and analyze lunar data obtained from the freely-available JMARS [6] GIS platform [see our previous work, 7-10].  We will use the data extracted by the pipeline in combination to generate automated suggestions for crater classification as either secondary or primary.  Because the characteristics that distinguish primary and secondary craters on the Moon are contextual (with secondary characteristics dependent on factors such as crater age, underlying surface age, and distance from the parent primary), we have developed four different frameworks for assessing each crater.

Framework 1: Total “Points”

In the first framework, we will test each criterion separately, using the same classification characteristics for all craters in the dataset.  Each crater will be assessed in terms of seven secondary criteria: shallow depth-to-diameter ratio (d/D); asymmetric uprange-to-downrange elevation profile slope ratio (Slope Ratio); high degree of clustering (Clustering); cluster orientation toward a parent primary (Orientation); elliptical planform shape (Planform); secondary morphology (Morphology); and the presence of block-rich material downrange (Debris Tail).  In Framework 1, we will assign one point for each secondary criteria the crater exhibits.  We will classify craters with four or more points as secondary crater candidates.

Framework 2: Cluster-Focused

In the second framework, we will evaluate each crater based on the same seven criteria, with an emphasis on clustered craters.  First, we will assess all craters in the dataset for membership in a cluster, using the same clustering criteria for all craters.  In this first step, we aim to identify all craters in the dataset that are even mildly spatially clustered.  We will then remove false-positive secondary craters from this list of craters using the remaining six criteria.  If clustered craters exhibit at least three remaining criteria, we will classify them as potential secondary craters.

Framework 3: Crater-Age-Focused

In the third framework, the cut-offs for each secondary crater criterion will be tuned depending on relative crater age for each crater in the dataset [11].  The process will begin by assessing the degradation state of each crater in the dataset and classifying it as Fresh, Moderately Degraded, or Heavily Degraded.  We will then evaluate all Fresh craters by the same criteria cut-offs.  A differently tuned set of criteria cut-offs will be used for Moderately Degraded craters, and so on.  After we determine the degradation state of each crater, the order that we will assess the remaining criteria will be: Clustering, Orientation, Planform, d/D, Slope Ratio, Morphology, and Debris Tails.

Framework 4: Underlying-Surface-Age-Focused

This framework will follow the same basic structure and logic as Framework 3, except that a proxy for the underlying surface age will first be determined for each crater.  This relative surface age will be based on the number of craters with diameter greater than 1 km per square kilometer (N(>1)) in the surrounding terrain.  After determining this N(>1) value, we will classify each crater in the dataset as being on Sparsely Cratered, Moderately Cratered, or Heavily Cratered terrain.  Using the same order of evaluating the criteria as in Framework 3, we will assess all craters on Sparsely Cratered terrains using the same cut-off criteria, and so on.

Combining the Frameworks

After each crater in the dataset has been assessed using all four frameworks, all craters that are determined to be potential secondary craters by at least two of the four frameworks will be classified as secondaries.  The results for each of the seven criteria in each of the four frameworks will be output in a file for each of the craters in the dataset, along with the final classification of secondary or primary.  This will allow for a quick final decision to be made by an experienced crater classifier, using the recorded data from each crater as a guide in accepting or rejecting the recommendation made by the semi-automated pipeline.  If the decision of the pipeline is overturned, the wealth of data extracted will allow the experienced counter to clearly justify their classification decision.  This will make this method both transparent and repeatable.

 

Conclusions

We will present our results after an eight-week summer research experience of combining the existing pieces of the pipeline into a single, streamlined procedure.  Specifically, we will compare our semi-automated classification of Tycho secondary craters with diameters greater than 1 km, located within 8 crater radii, against existing secondary classifications by various workers [e.g., 12].  

 

References

[1] Dundas, C. M. and McEwen, A. S., (2007), Icarus, 186, 31-40.  [2] Lucchitta, B. K., (1977), Icarus, 30, 80-96.  [3] McEwen, A. S. et al., (2005), Icarus, 176, 351-381.  [4] Robbins, S. J. and Hynek, B. M., (2011), Geophys. Res. Letters, 38, L052201.  [5] Wells, K. S. et al., (2010), J. Geophys. Res., 115, E06008.  [6] Christensen, P. R. et al., (2009), In: American Geophysical Union Conference, Abstract IN22A-06.  [7] Martin-Wells, K. S. et al., (2022), Lunar Planet. Sci. Conf., 53^rd, 2678, id. 2557.  [8] Powers, L. T. et al., (2022), 13^th Planet. Crater Consortium Meeting, 2702, id. 2023.  [9] Powers, L. et al., (2023), Lunar Planet. Sci. Conf., 54^th, 2806, id. 2259.  [10] Martin-Wells, K. S. et al., (2024), Lunar Planet. Sci. Conf., 55^th, 3040, id. 2093.  [11] C. I. Fassett and B. J. Thomson, (2014), J. Geophys. Research: Planets, 119, 2255–2271.  [12] K. N. Singer et al., (2020), J. Geophys. Research: Planets, 125.