Checked September 11, 2026. The model’s water category is now grounded across the full retained weather grid. Its zero sea-surface-temperature values are already present in the provincial archive, while model skin temperature and surface heat flux are available. These findings improve the evidence behind the inputs; they do not validate marine dispersion.
We checked all 20 tiles, 2,000 cells and 25 hourly model states supplying the expanded research grid: June 1, 2023 08:00 UTC through June 2 08:00 UTC. The province’s published index and application source led to the original NetCDF objects. Small, guarded byte-range reads recovered the matching surface data; no complete annual file was downloaded.
The files identify USGS land-use encoding with 28 categories, water code 16 and lake code 28. All 228 cells carrying code 16 agree with both explicit water-mask fields throughout the selected period. The separate lake mask adds an important distinction:
| Native model classification | Cells | Cell-hours |
|---|---|---|
| Land-mask land | 1,772 | 44,300 |
| Water code 16, lake mask 0 | 189 | 4,725 |
| Water code 16, lake mask 1 | 39 | 975 |
Every cell’s lake mask is constant across the 25 states. No land-use code 28 remains in this grid, even though 39 cells are lake-marked. The released WRF initializer records the lake mask before a later lake-to-water category merge. This explains a documented way the fields can coexist; the actual June initialization logs remain unavailable.
Model water does not mean verified open ocean. The airport-nearest native point, I33/J84, is code 16 and lake-marked in the provincial archive. These categories do not establish accurate local river, lake, coastal or mixed-cell geography. An initial internal statement generalizing the first tile’s zero lake mask was corrected during final validation, before publication or model use.
All 50,000 SST values are zero in both archives, including every model-water cell. The variable exists; it contains unusable zeros. Separate fields are populated throughout:
| Field over model-water cells | Available range |
|---|---|
| TSK: simulated surface skin temperature | 272.86–288.80 K |
| HFX: native surface heat flux | −122.83 to 42.40 W/m² |
TSK is not observed water temperature or a replacement SST boundary. The released WRF ocean driver uses skin temperature separately. The cause of the archive’s zero SST, and the actual June ocean initialization, remain unresolved. No temperature, heat flux or mask was inserted into the simulation.
The data linkage includes all 500 original tile-hour hashes, exact land-use and SST agreement, and coordinates, terrain, temperature, humidity and pressure agreement within printing precision. Initial pressure checks in 11 tiles exceeded the 0.05 hPa tolerance by at most 0.00005 hPa because the comparison multiplied by 0.01. Following the released CALWRF writer’s single-precision division by 100 resolves every comparison without widening the tolerance. Both sets of checks are retained. CALWRF passes through the supplied land-use code; its hard-coded 25-category header is not the production category count.
The annual NetCDF headers retain the initial December 31, 2022 run’s metadata, including SST update off and ocean physics option 1. They do not prove every June configuration or identify accepted assimilation observations. The NetCDF stores 33 actual vertical levels while the retained weather file has 40; this check covers surface fields only.
The next step is to resolve the production ocean initialization and evaluate any proposed thermal sensitivity under a separate declared protocol. These newly recovered fields have not been adopted as validated marine inputs.
Download the numeric checks and source provenance, 50,000-row surface table, compressed CSV, or compact methods ZIP. The methods contain our scripts, JSON and notes; they exclude third-party source code and annual data files.