Distinct initiated cells can receive the same `cell_uid`

[RBhupi]

Pipeline stage where the bug occurs

Acquisition — downloading or queuing NEXRAD files

Adapt version

0.1.4

Radar site and data source

KMRX

What happened?

A live KMRX run (2026-08-12 00:14:15 UTC) produced two distinct initiating cells in the same scan with the
same cell_uid:

• cell label 18 → TKR02LX0GX
• cell label 23 → TKR02LX0GX

cell_volume_stats correctly failed because its primary key is
(run_id, scan_time, cell_uid). Earlier tracking persistence can also upsert one
of the colliding cells_by_scan rows, losing an observation before the failure.

Root Cause

New-cell identity uses a quantized v1 birth signature composed of time,
latitude/longitude, reflectivity, ZDR, and area. Distinct cells may share every
quantized value, so they deterministically receive the same UID. This is a
signature collision, not necessarily a digest/hash collision.

Steps to reproduce

(adapt_env) adapt-repo % adapt run-nexrad config.yaml --radar KMRX
ARM Adapt v0.1.3.post39+dirty
Copyright © 2026, UChicago Argonne, LLC.
See the LICENSE (https://arm-doe.github.io/Adapt/license.html) for terms and disclaimer.

2026-08-12 00:12:39,865 - adapt.run - INFO - ── Adapt run 2026AUG12-0012-KMRX ─────────────────────────────
site KMRX · NEXRAD · mode realtime
pipeline aws_nexrad v0.1.3.post39+dirty · commit — · python 3.14.4 · macOS-26.6.1-arm64-arm-64bit-Mach-O
config /Users/bhupendra/Desktop/adapt-repo/collections/KMRX/config_run_2026AUG12-0012-KMRX.json (sha 8fe8194a) · modules: ingest, detection, projection, analysis, tracking, cell_volume_stats
────────────────────────────────────────────────────────────
2026-08-12 00:12:39,865 - adapt.run - INFO - methods:

Full error message or traceback

2026-08-12 00:14:15,208 - adapt.runtime.processor - ERROR - Scan failed | scan=d75eff37114948b6 elapsed=23.87s ValueError: cell_volume_stats: submitted DataFrame has duplicate primary keys
Traceback (most recent call last):
  File "/Users/bhupendra/projects/arm_adaptive_scaning/Adapt/src/adapt/runtime/processor.py", line 419, in _run_scan
    self._router.persist(self._post_modules, ext_result, enrich_meta)
    ~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/bhupendra/projects/arm_adaptive_scaning/Adapt/src/adapt/persistence/output_router.py", line 44, in persist
    self._dispatch(module.name, spec, result, meta)
    ~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/bhupendra/projects/arm_adaptive_scaning/Adapt/src/adapt/persistence/output_router.py", line 143, in _dispatch
    ).write(df)
      ~~~~~^^^^
  File "/Users/bhupendra/projects/arm_adaptive_scaning/Adapt/src/adapt/persistence/module_output.py", line 41, in write
    self._reject_duplicate_primary_keys(frame)
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^
  File "/Users/bhupendra/projects/arm_adaptive_scaning/Adapt/src/adapt/persistence/module_output.py", line 82, in _reject_duplicate_primary_keys
    raise ValueError(f"{self._spec.table}: submitted DataFrame has duplicate primary keys")
ValueError: cell_volume_stats: submitted DataFrame has duplicate primary keys

Configuration file (redact credentials)

Environment

MacOS

Checklist

• I checked the documentation and existing issues
• I can reproduce this consistently
• This occurs on the latest release