AIMS Data Provenance

What AIMS is

The Asteroid Institute Moving Source Catalog (AIMS) is the Asteroid Institute’s normalized catalog of moving-source and difference-image detections used by ADAM Precovery, attribution, and related solar-system search workflows. The Precovery methods page explains how these detections are searched.

AIMS is not a replacement for the original survey archives. It selects measurements useful for moving-object searches, preserves source and observatory identity, and translates heterogeneous publisher products into a common observation representation. The resulting data are available as BigQuery tables and as partitioned Parquet products for search services.

Current source families

Each source page documents the publisher product, AIMS selection, normalized values, consumer products, and scientific limitations.

Common observation representation

The most important fields in a normalized AIMS observation are:

Selection and normalization principles

AIMS does not apply one universal source filter. Each publisher exposes a different product, so the source pages define the actual selection function. Examples include an NSC proper-motion and detection-history filter, ZTF alert-quality cuts, ATLAS classifier and duplicate-selection cuts, a Gaia-based SkyMapper static-source mask, LSST alert quality and signal-to-noise cuts, and required-field validation for MPC observations.

After source selection, AIMS applies a small common set of transformations:

• times are represented in UTC MJD and timestamp forms, with dataset-specific time-scale conversion where needed;
• exposure midpoints are retained or derived from source time and duration;
• source identifiers are retained when possible, with deterministic identifiers used otherwise;
• positions are represented in degrees and indexed at nside=128 NESTED HEALPix;
• source filters and photometry are retained or converted only as documented on the source page;
• invalid required coordinates, times, identifiers, and duplicate keys are rejected or handled deterministically by the source-specific transform.

AIMS generally preserves publisher measurements rather than recalibrating every survey onto one astrometric or photometric system.

Source-specific selection at a glance

These differences are scientifically important: dataset_id identifies not just a publisher but also a different AIMS selection function.

Published consumer products

BigQuery

In the AIMS BigQuery dataset, normalized observations are available through provider tables and shared consumer surfaces. AIMS first writes aims_nsc, aims_ztf, aims_atlas, aims_skymapper, aims_lsst, and the hot/archive MPC tables. These feed:

• aims_detections_hot and aims_detections_archive, the physical normalized-observation tables;
• aims_detections_unified, the union view used to query the full observation surface;
• aims_frames_unified and its archive sibling, derived search-frame indexes grouped by source exposure and spatial cell;
• observation_attribution_edges_v2, mpc_obs_identity_v1, and mpc_designation_lookup_v1, which keep observation-level MPC matches and current identity interpretation separate from source detections.

The hot/archive split changes storage cadence, not scientific meaning: recent data are partitioned daily and older sparse data monthly, while aims_detections_unified presents one observation model.

Parquet and runtime catalog

AIMS exports the unified detections and derived frame indexes to Google Cloud Storage as partitioned Parquet files:

• aims_detections_parquet and aims_frames_parquet for daily hot partitions;
• aims_detections_parquet_archive and aims_frames_parquet_archive for monthly archive partitions;
• aims_unified_parquet_catalog/catalog.json, which lists complete detection/frame pairs for consumers such as Precovery.

The Parquet files reorganize the BigQuery products for efficient spatial and temporal search; they do not apply a second scientific calibration. Provider identity remains available through dataset_id. Frame Parquet rows are AIMS-derived search summaries, not copies of publisher image catalogs.

Scientific validity and limitations

AIMS rows are source detections normalized for search, not final object identifications and not an unbiased survey sample.

• Publisher astrometry, photometry, flags, and calibration remain subject to the source survey’s own validity limits.
• AIMS selection intentionally favors moving or transient candidates. Completeness and contamination therefore depend on the source-specific filter and should not be inferred from the normalized schema alone.
• The same physical observation can appear once in a native survey dataset and again as an MPC observation. Consumers should use the attribution products rather than assuming every unified row represents an independent exposure.
• MPC attribution is a separate observation-level match product. It can change as published observations and object identities evolve, and it is not an orbit-linking proof.

Astrometric uncertainty

When populated, ra_sigma_deg and dec_sigma_deg are intended to represent 1-sigma eastward and northward components on the local sky tangent plane. Upstream surveys differ in units, conventions, and calibration, so each source page identifies whether the stored values are publisher measurements, AIMS approximations, or unavailable.

The current common row does not preserve cross-axis correlation or a full 2D source covariance. Consumers therefore have only a diagonal approximation when both values are present. Missing values remain null rather than being treated as zero, and values should not be assumed equally calibrated across surveys.

Documentation scope

These pages describe the current publisher-to-AIMS transformations and public consumer products. They intentionally omit credentials, private delivery locations, operational orchestration, and rapidly changing row counts. Material changes to source access, selection, normalization, products, or scientific interpretation should be reflected on the corresponding source page.