kacpnowak/hclimrep-ocean-oceanbench

HCLIMREP Ocean — OceanBench challenger forecasts

OceanBench challenger submissions produced with the WeatherGenerator machine-learning Earth system model, together with the executed evaluation notebooks that turn each forecast store into an OceanBench report.

Each subdirectory is one evaluated model. Scores are tied to a specific OceanBench version and are not comparable across versions.

Model	OceanBench	Store	Report
ft0818	0.4.0	ft0818/wg_ft0818_B1.zarr	ft0818/challenger_ft0818_global_report.ipynb

---

ft0818

Ten-day global ocean forecasts from WeatherGenerator run glory_finetuning_20260818_105654, initialised weekly through 2024 and evaluated with OceanBench 0.4.0 over the global region.

Contents

ft0818/
  wg_ft0818_B1.zarr/                        forecast store (Zarr v2, ~39 GB on disk)
  challenger_ft0818.py                      opens the store as `challenger_dataset`
  challenger_ft0818_global_report.ipynb     executed OceanBench report, region "global"

Store schema

Dimensions:  first_day_datetime: 52, lead_day_index: 10, depth: 9,
             latitude: 672, longitude: 1440

Data variables (float32)
  thetao  sea_water_potential_temperature   degrees_C   (time, lead, depth, lat, lon)
  so      sea_water_salinity                PSU         (time, lead, depth, lat, lon)
  uo      eastward_sea_water_velocity       m/s         (time, lead, depth, lat, lon)
  vo      northward_sea_water_velocity      m/s         (time, lead, depth, lat, lon)
  zos     sea_surface_height_above_geoid    m           (time, lead, lat, lon)

• Forecast starts: 52 weekly Wednesdays, 2024-01-03 through 2024-12-25. Each forecast is initialised from the GLORYS nowcast of the preceding Tuesday.
• Lead days: lead_day_index 0–9, one daily mean per step, ten days total.
• Grid: 0.25° regular, latitude −78.0 to 89.75, longitude −180.0 to 179.75.
• Depths (m): 0.494, 47.374, 92.326, 155.851, 222.475, 318.127, 380.213, 453.938, 541.089. These are the nine OceanBench reference levels, taken directly from the model's own GLORYS depth axis, so no vertical interpolation or nearest-neighbour substitution is involved.
• Chunking: one forecast start per chunk, all leads and depths in that chunk.
• Written with Zarr v2. OceanBench pins zarr==2.18.4 and cannot read v3 stores. Consolidated metadata is present, so the store opens either way; the shipped challenger script passes consolidated=False so that it keeps working if the store is ever extended with further forecast starts.

Usage

hf download kacpnowak/hclimrep-ocean-oceanbench --repo-type dataset \
  --include "ft0818/*" --local-dir ./hclimrep-ocean-oceanbench

import xarray as xr

challenger_dataset = xr.open_zarr(
    "hclimrep-ocean-oceanbench/ft0818/wg_ft0818_B1.zarr", consolidated=False
)

challenger_ft0818.py reads the store path from the WG_CHALLENGER_STORE environment variable, so the shipped file works unchanged once that points at your local copy:

export WG_CHALLENGER_STORE=$PWD/hclimrep-ocean-oceanbench/ft0818/wg_ft0818_B1.zarr

To reproduce the report, drop that script into the OceanBench challenger notebook template and execute it, or re-run the shipped notebook in an environment with oceanbench==0.4.0 and the OceanBench reference data staged locally.

Results (OceanBench 0.4.0, region global)

RMSD against GLORYS reanalysis, surface fields:

Metric	Lead day 1	Lead day 5	Lead day 10
Sea surface height (m)	0.0574	0.0569	0.0614
Temperature (°C)	0.5692	0.5837	0.6416
Salinity (PSU)	0.4080	0.3925	0.3881
Zonal current (m/s)	0.1136	0.1222	0.1288
Meridional current (m/s)	0.1133	0.1202	0.1257
Mixed layer depth (m)	49.11	49.49	50.68

RMSD against GLORYS reanalysis, temperature and salinity at depth:

Metric	Lead day 1	Lead day 5	Lead day 10
Temperature 50 m (°C)	0.8030	0.7764	0.8019
Temperature 100 m (°C)	0.9491	0.9237	0.9337
Temperature 200 m (°C)	0.7709	0.7583	0.7572
Temperature 500 m (°C)	0.4948	0.4849	0.4872
Salinity 50 m (PSU)	0.2047	0.1997	0.1988
Salinity 100 m (PSU)	0.1670	0.1636	0.1621
Salinity 200 m (PSU)	0.1374	0.1344	0.1314
Salinity 500 m (PSU)	0.0846	0.0827	0.0805

RMSD against observations (Class-4):

Metric	Lead day 1	Lead day 5	Lead day 10
Sea level anomaly (m)	0.0578	0.0648	0.0717
Temperature 0–5 m (°C)	0.8235	0.8840	0.9307
Temperature 5–100 m (°C)	1.0374	1.0485	1.0924
Temperature 100–300 m (°C)	0.8354	0.8647	0.9112
Temperature 300–600 m (°C)	0.5459	0.5637	0.5978
Salinity 0–5 m (PSU)	0.2694	0.2658	0.2896
Salinity 5–100 m (PSU)	0.3088	0.3351	0.3168
Zonal current 15 m (m/s)	0.2023	0.2074	0.2150
Meridional current 15 m (m/s)	0.1853	0.1931	0.1976

Geostrophic currents against GLORYS, and Lagrangian trajectory deviation:

Metric	Lead day 1	Lead day 5	Lead day 10
Zonal geostrophic current (m/s)	0.2088	0.1913	0.1851
Meridional geostrophic current (m/s)	0.1828	0.1685	0.1635
Lagrangian trajectory deviation (km)	—	36.16	—

Lagrangian deviation is defined for lead days 2 through 9; it grows from 9.70 km at lead day 2 to 67.54 km at lead day 9.

The notebook also contains the GLO12-analysis counterparts of the GLORYS-reanalysis metrics. See the executed report for every row.

Provenance

• Model: WeatherGenerator run glory_finetuning_20260818_105654, an aggregation-only, ten-step finetune of the preceding ft0808 checkpoint.
• Inference: 52 single-forecast runs driven by config/config_forecasting_glorys_obench_v2.yml, using the 133-channel GLORYS v2 ocean stream and an ERA5 atmospheric stream. Atmospheric forcing is supplied only over the initial-condition window; the model sees no future atmosphere during the ten-day rollout.
• Export: export_glorys_nc.py, selecting the nine reference depth labels from the model's native output levels.
• Evaluation: OceanBench 0.4.0, region global, no post-processing or bias correction applied to the forecasts.

Caveats

• Mixed layer depth is computed from nine output levels rather than a full vertical profile, which discretises the diagnostic and inflates its RMSD.
• Scores are specific to OceanBench 0.4.0. Do not compare them against numbers produced by other versions of the benchmark.
• The store is Zarr v2 for compatibility with the zarr==2.18.4 pin in OceanBench. It cannot be read by a Zarr v3-only reader.
• Values are raw model output with no clipping. A small number of salinity points fall marginally outside physically plausible bounds, which the scores already reflect.