GribStream

GribStream Blog

DMI HARMONIE DINI solar and wind forecasts are now on GribStream

GribStream | Published |
dmidini ifsoper aifsoper gfs era5

GribStream now supports a selective DMI HARMONIE DINI surface forecast dataset for Northwestern Europe, including solar radiation, 10 m and 100 m wind, and near-surface weather context through a coordinate-based API.

GribStream now supports DMI HARMONIE DINI as dmidini, a selective surface forecast dataset from DMI Open Data for Northwestern Europe.

For customers, the useful part is not another raw file feed to manage. It is coordinate-based API access to a high-resolution regional model with solar radiation, 10 m and 100 m wind, cloud, temperature, humidity, visibility, pressure, and accumulated precipitation in one query workflow.

This first GribStream release is intentionally scoped: dmidini is a recent moving-window mirror of the DMI HARMONIE_DINI_SF surface product, not a full copy of every DMI HARMONIE output.

What DMI DINI Adds

DMI describes HARMONIE as a high-resolution regional weather model, with DINI covering Northwestern Europe and IG covering Iceland/Greenland. The DINI surface product adds a 2 km regional option for users who need more local structure over Northwestern Europe than global guidance such as IFS Oper, AIFS Oper, or GFS can provide.

The first dmidini subset focuses on fields that are likely to matter in products:

  • solar and radiation fields such as direct solar radiation, downward shortwave radiation, global radiation, and net shortwave radiation
  • 10 m wind speed, wind direction, wind gust, and 10 m U/V wind components
  • 100 m U/V wind components for workflows that need wind context above the usual near-surface level
  • 2 m temperature, maximum temperature, minimum temperature, dew point, and relative humidity
  • high, medium, low, and fractional cloud cover
  • visibility, pressure, accumulated total precipitation, accumulated rain precipitation, and accumulated snowfall water equivalent

DMI documents deterministic DINI output every third hour, with a 60-hour DINI forecast range. In GribStream, the supported surface product is queryable by coordinate, valid time, forecast cycle, lead time, level, and selector.

Why It Matters

dmidini is a practical specialized dataset for European energy, field operations, transport, and model-comparison workflows.

For solar products, the important addition is regional shortwave radiation guidance that can be joined with the same locations and valid times used in an existing forecast pipeline. For wind products, the 100 m U/V components are useful because many operational questions care about wind above 10 m, even when the rest of the application still needs standard near-surface weather context.

DMI documents the U/V wind components as relative to the Lambert projection. GribStream preserves those source components, so applications should rotate them before deriving Earth-relative wind direction or vectors in another coordinate reference system.

It is not a replacement for global models. A customer comparing forecast systems may still use IFS Oper, AIFS Oper, GFS, or ERA5 for global context, verification, or fallback coverage. DINI is valuable when the question is regional detail over Northwestern Europe.

Instantaneous and Accumulated Series

The selected precipitation fields and most solar-energy fields accumulate from the model run start through each forecast valid time. Downward shortwave radiation is also available as an instantaneous flux. GribStream keeps the instantaneous and accumulated values as separate, coherent series instead of mixing values with different meanings.

For a field such as downward shortwave radiation, use info = "" for the instantaneous series and info = "acc" for the run-to-date accumulated series. The accumulation period grows with lead time: the forecast-hour 24 value covers the first 24 hours of the run, while forecast hour 48 covers the first 48 hours. Use the DMI DINI model page inventory to choose the exact name, level, and info tuple before putting a query into production.

The accumulated dswrf and direct-solar-exposure series begin at forecast hour 1 because DMI does not publish those messages at forecast hour 0. Other selected accumulations, including tp, nswrf, grad, rprate, and tsrwe, do include DMI's zero-valued run-start record.

That distinction matters for dashboards and forecasts. An accumulated radiation value and an instantaneous radiation value answer different questions, even when the source short name is the same.

The inventory follows DMI's source parameter names where DMI publishes them. Statistical distinctions stay in info: 2t uses max and min for preceding-hour extrema, gust uses max, and accumulated fields use acc. This keeps the API selectors traceable to DMI's documentation while preventing distinct series from being coalesced.

What Is Not Included Yet

DMI Open Data publishes more than this first GribStream dataset.

The public DMI bucket includes separate HARMONIE DINI pressure-level and model-level products, DINI EPS means, DINI EPS percentiles, DINI EPS probabilities, and IG products for Iceland/Greenland. Those are separate source products, not hidden members of the current dmidini dataset.

GribStream is starting with the DINI surface solar, wind, and basic meteorological context subset because those fields support immediate energy and operations workflows. We may evaluate the EPS, pressure-level, model-level, or IG products later if customer demand justifies the additional indexing and storage work.

What This Enables

dmidini in GribStream is useful for:

  • solar generation forecasting and irradiance-aware dashboards
  • 10 m and 100 m wind screening over Northwestern Europe
  • joining DMI regional guidance with IFS, AIFS, GFS, or ERA5 at the same points
  • short-range logistics, field operations, and weather-impact products
  • incident review while the source run remains inside the retained recent-data window

Start with the DMI DINI model page, choose the variables from the inventory, and build coordinate, time-range, or grid queries against dmidini.

Sources

Related models

dmidini ifsoper aifsoper gfs era5
Try GribStream