Description
|
Motivation Our motivation for this analysis is to create an efficient and transparent system to categorize anthropogenic space objects (ASO's) based on orbital behavior, namely angular momentum. Doing so, we hope to lay the groundwork for future studies that involve trackability, detection, and identification. Through the creation of "orbital zip-codes", and characterization of these clusters of ASOs in angular momentum space, we can better understand their orbital behavior to a greater extent and identify migration patterns. Objects that moved from one cluster to another were considered to be "nomadic" objects. In theory, ASOs should have a constant or near-constant angular momentum value. Objects that exhibit significant movement between clusters require keen investigation as to why they exhibit erratic orbital migration patterns. (2021-08-02)
The Data The 12 datasets included in this submission are JSON files that contain orbital information regarding ASOs over a 12-month period in 2019. Each dataset is a snapshot of the objects on the first day of each month. The orbital information includes angular momentum, semi-major axis, eccentricity, mean anomaly, longitude of the ascending node, equinoctial elements, and inclination. The data was acquired from ASTRIAGraph. Python scripts were written to convert the data from JSON format to a clean, manipulatable Pandas data frame. The angular momentum vectors in the x, y, and z axes were isolated and fed into K-means and spectral clustering models were implemented after conversion to polar coordinates. From here, plotting can be done if one wishes to observe the created clusters. The elbow method was used to determine the optimal number of clusters for the month of January. This number was kept constant for each run of the model on a different dataset to maintain consistency. One issue encountered was the redundant naming of the ASO objects inherent in the dataset. To discover and track nomadic objects over the one-year period, it may be helpful to handle these redundancies through renaming. Additionally, the month of January was instantiated as a reference point. The data in other months was compared to that of January to identify nomadic objects. The clusters can be characterized through object density or population. The 3-dimensional volume of a cluster could be roughly calculated by using 3D shape formulas. The object density of a cluster can be obtained by sampling points iteratively from one cluster in a constrained volume and dividing by said volume.
Usage These datasets can be reused by policymakers, by industry, by researchers and by the public interested in clustering of ASOs
|
Related Publication
| Riley Steindl, Vishnu Nair, Maya Slavin, Mauricio Barba, Danielle Wood, Moriba Jah (2021). Developing Detectability, Identifiability, and Trackability Analyses for the Space Sustainability Rating
|