# Uniform-conductivity comparison

Simultaneous Kh/Kv-family sensitivity; original thickness/storage and lower-layer confined assumption retained. Neither preferred geology nor complete cross-border correction. Repeated 2017 forcing, not historical trajectory.

Measured by `.venv/bin/python docs/records/seepage-storage-aquifer-review-2026-09-28/property_null/compare_conductivity.py`. Compared completed cycles: [7, 8, 9, 10, 11, 12]. All start from the shared original R1000 cycle6 endpoint; upper/lower Kh become 10/5 m/day and Kv becomes 0.1 m/day in both layers. Executed input checks establish unchanged thickness, storage and layer types. These constants are the existing expanded-domain recipe, not newly inferred geology.

At cycle 12, upper/lower p95 year-end head changes are 0.954/0.951 m; periodic-adjustment criterion passed: False. This is a finite-duration conductivity sensitivity, not an equilibrated geological replacement.

## End of repeated-year cycle 12

| Group | Layer | Median head change (m) | 5th–95th percentile change (m) |
|---|---:|---:|---:|
| whole_domain | 1 | 0.000 | -5.579 to 3.332 |
| whole_domain | 2 | 0.000 | -7.945 to 4.783 |
| KAZ | 1 | 0.000 | -0.015 to 0.158 |
| KAZ | 2 | 0.000 | -0.016 to 0.217 |
| KGZ | 1 | -0.231 | -9.002 to 11.101 |
| KGZ | 2 | -1.517 | -10.622 to 10.924 |
| inspection_window | 1 | 0.000 | -5.466 to 0.974 |
| inspection_window | 2 | 0.000 | -10.274 to 0.793 |
| inspection_KAZ | 1 | 0.000 | -0.036 to 0.323 |
| inspection_KAZ | 2 | 0.000 | -0.046 to 0.367 |
| inspection_KGZ | 1 | -0.159 | -9.114 to 1.577 |
| inspection_KGZ | 2 | -4.123 | -12.523 to 1.926 |
| native_zone_0 | 1 | 0.000 | -0.024 to 0.256 |
| native_zone_0 | 2 | 0.000 | -0.025 to 0.336 |
| native_zone_1 | 1 | -1.379 | -9.149 to 13.158 |
| native_zone_1 | 2 | -1.634 | -10.371 to 12.477 |
| native_zone_2 | 1 | -0.209 | -9.645 to 1.896 |
| native_zone_2 | 2 | -2.559 | -11.316 to 4.756 |
| native_zone_3 | 1 | 1.317 | -2.228 to 38.822 |
| native_zone_3 | 2 | 1.836 | -2.426 to 38.982 |

Native zone 0 is expanded default, zones 1–3 are inherited KGZ screening assignments. Country and inspection-window groups overlap; do not sum them. The shared original inspection window is x340–450 km, y4740–4800 km in EPSG:32643.

The maximum layer relative budget residual is 1.39e-11; maximum annual cell relative residual is 4.54e-08. These verify numerical conservation, not physical correctness.

The contrasts show the response to changing horizontal and vertical conductivity together. They cannot attribute a response separately to Kh or Kv, remove retained thickness/storage transitions, validate a material boundary, or prove equilibration. Head-above-terrain counts are pressure diagnostics, not observed inundation area. Review cycle-end change diagnostics alongside these differences: slow deep-fan adjustment can persist even where shallow heads respond quickly.

Original cycles1–6 in inherited results are checkpoint history, not uniform-conductivity runs. Only cycles7 onward use this comparison. Source preservation is recorded separately, and owned raw run trees are removed only after distilled checks.