← Accuracy Audit  |  Interactive Coverage Dashboard →

National Parcel Coverage — Live

Autonomous discovery + ingest via daedarch.pipeline.state_data_scavenger. Updated 2026-09-09 08:11 UTC. Source of truth: MongoDB gold layer with full lineage.

Per-state coverage

State	Sources	Gold	Valid %	Owner %	Value %	Geom %	Area (sq mi)
NC	7	10,030,689	100%	0%	0%	0%	48,618
NJ	10	3,491,593	100%	0%	0%	0%	7,354
FL	6	1,929,259	100%	0%	0%	0%	53,625
CA	14	1,380,915	100%	0%	0%	0%	155,779
ME	8	704,667	100%	0%	0%	0%	30,843
VA	9	453,022	100%	0%	0%	0%	39,490
AL	3	429,543	100%	0%	0%	0%	50,645
WI	16	416,735	100%	0%	0%	0%	54,158
AK	9	415,827	100%	0%	0%	0%	570,641
GA	7	396,815	100%	0%	0%	0%	57,513
MI	11	356,677	100%	0%	0%	0%	56,539
VT	1	344,323	100%	100%	100%	0%	9,217
KY	5	321,849	100%	0%	0%	0%	39,486
NY	47	270,215	100%	0%	0%	0%	47,126
IA	9	269,231	100%	0%	0%	0%	55,857
IN	3	265,485	100%	0%	0%	0%	35,826
OH	6	200,900	100%	0%	0%	0%	40,861
OK	6	170,044	100%	0%	0%	0%	68,595
PA	7	162,265	100%	0%	0%	0%	44,743
TX	9	144,913	100%	0%	0%	0%	261,232
SC	2	134,112	100%	0%	0%	0%	30,061
AZ	15	124,767	100%	0%	0%	0%	113,594
MT	7	115,560	100%	0%	0%	0%	145,546
NM	6	94,048	100%	0%	0%	0%	121,298
SD	4	86,259	100%	0%	0%	0%	75,811
IL	8	82,422	100%	0%	0%	0%	55,519
MO	4	76,589	100%	0%	0%	0%	68,742
KS	11	65,728	100%	0%	0%	0%	81,759
MS	9	57,553	100%	0%	0%	0%	46,923
AR	6	35,453	100%	0%	0%	0%	52,035
NV	8	34,319	100%	0%	0%	0%	109,781
TN	90	26,024	100%	96%	96%	0%	41,235
MD	9	25,556	100%	0%	0%	0%	9,707
WV	15	21,253	100%	0%	0%	0%	24,038
MA	6	16,555	100%	0%	0%	0%	7,800
RI	6	14,475	100%	0%	0%	0%	1,034
UT	11	12,316	100%	0%	0%	0%	82,170
NE	3	4,740	100%	0%	0%	0%	76,824
ND	10	3,725	100%	0%	0%	0%	69,001
WA	18	3,600	100%	0%	0%	0%	66,455
MN	8	1,287	100%	0%	0%	0%	79,627
DE	6	1,200	100%	0%	0%	0%	1,949
OR	5	1,200	100%	0%	0%	0%	95,988
NH	9	1,200	100%	0%	0%	0%	8,953
HI	7	1,200	100%	0%	0%	0%	6,422
CT	4	900	100%	0%	0%	0%	4,842
CO	6	900	100%	0%	0%	0%	103,642
WY	5	861	100%	0%	0%	0%	97,093
ID	3	300	100%	0%	0%	0%	82,643
LA	4	71	100%	0%	0%	0%	43,204
DC	1	28	100%	0%	0%	0%	61

Scavenger pattern learning (top 8)

AGOL-search patterns dominate. Tavily patterns get deprioritized as their loss count grows.

Backend	Query template	Wins	Losses	Win rate
agol_search	{state_name} standardized parcels	419	0	100%
agol_search	{state} parcels title:parcels	393	1	99%
agol_search	{state_name} construction permits	348	1	99%
agol_search	{state} building permits title:permits	243	7	97%
tavily	{state_name} city open data portal building permits Socrata dataset	164	8	95%
tavily	{state_name} GIS open data hub arcgis rest services root	69	9	87%
tavily	{state_name} statewide tax parcels ArcGIS FeatureServer REST endpoint	49	52	48%
tavily	{state_name} construction permits ArcGIS FeatureServer	35	77	31%

Methodology

• Discovery: daedarch.pipeline.state_data_scavenger — Tavily + AGOL content-search + Data.gov + direct URL patterns. Per-state bbox rejection for cross-state false positives. Pattern win/loss tracked in state_scavenger_patterns for self-improvement.
• Harvest: daedarch.pipeline.harvest_source_to_bronze — paginated FeatureServer pull → trellison_parcels_bronze with provenance (source_key, source_url, fetched_at, discovery_run_id).
• Promote: daedarch.pipeline.promote_bronze_to_gold v1.2 — auto-learned field map per source (regex-scored heuristics across 9 canonical buckets), validation status (valid / warnings / failed), full lineage (bronze_row_id + source_url + license_flag + transformation_version).
• License flag: Auto-detected from source URL host (.gov, .state.us, AGOL public tenants → gov_open_data). Vendor portals → subscription_internal_only.

All data sourced from public government open-data portals and AGOL public tenants. Every gold row carries _source_url, _license_flag, _fetched_at, _promoted_at. To add a new state: POST /api/v4/execute {"tool_id":"daedarch.pipeline.state_data_pipeline","inputs":{"state":"XX"}}