Numerical weather analysis KENDA-CH1
MeteoSwiss produces analysis data using the KENDA (Kilometer-scale Ensemble Data Assimilation) system to estimate the atmospheric state at a given time over Switzerland and its surroundings. By combining short-range model forecasts with observations, the analysis provides the most consistent representation of atmospheric conditions. It is updated hourly and serves as the initial condition for numerical weather forecasts (e.g. ICON-CH1-EPS and ICON-CH2-EPS).
Getting started quickly
Example notebooks: From retrieval to visualization
To get started quickly, explore the Jupyter notebook, which shows how to retrieve and visualize maximum wind gusts from analysis data.
Available data
24h availability window
Data made available through this interface is accessible for 24 hours only after its publication. After this window, it is no longer available for retrieval.
If your request returns an empty response or you encounter a 403 error, it likely means the data you are trying to access is older than 24 hours.
Models' specifications
| Attributes | KENDA-CH1 |
|---|---|
| Collection | ch.meteoschweiz.ogd-analysis-kenda-ch1 |
| Horizontal Grid Size | approx. 1 km |
| Ensemble Members | Control Member Only |
| Step | 0 (instant values), 1 (accumulated/max/min values) |
| Grid | Native icosahedral |
| Temporal Output Resolution | 1 hour |
| New Model Run (Initialization) | every hour |
| Output Data Format | GRIB edition 2 |
Analysis data represents the atmospheric state at a specific time, and is therefore provided hourly.
Unlike the forecast datasets, only the deterministic control member is distributed in the analysis data.
Instantaneous analysis fields
Most parameters in the analysis dataset describe the instantaneous atmospheric state at the analysis time.
Examples include:
- temperature
- wind
- humidity
- pressure
- cloud variables
- pollen concentrations
These variables represent the best estimate of the atmosphere at time t and are therefore available at step = 0.
The list of available instantaneous variables can be identified in the Overview of Parameters file in the collection, where the temporal aggregation is labeled as "Instant".
First guess fields
Some meteorological parameters represent values that are accumulated, averaged, minimized, or maximized over a period of time, for example:
- accumulated precipitation
- maximum wind gust
- minimum or maximum temperature
Such quantities cannot represent the instantaneous atmospheric state at a single time and therefore are not available at step = 0.
Instead, these parameters are provided at step = 1, derived from an ensemble of ICON model runs at 1km resolution. They describe values aggregated over a one-hour period.
Parameters to be retrieved at step = 1 can be identified in the Overview of Parameters file by their temporal aggregation, such as "Average", "Sum", "Accumulation", or "Maximum".
Available parameters
To access the parameter overview, open the collection page for the analysis dataset. In the Assets section, download the file labeled Overview of Parameters.
Pollen data
Pollen concentrations are included in the instantaneous KENDA analysis fields of the ch.meteoschweiz.ogd-analysis-kenda-ch1 collection. They are available at step = 0, corresponding to lead_time="P0DT0H" in the Python API, and are updated every hour.
The pollen modelling chain combines information on plant distribution, flowering state and weather conditions with recent observations from the MeteoSwiss real-time pollen observation network. At the beginning of the season for a given pollen type, pollen is initialized in the KENDA-CH1 08:00 UTC cycle.
Pollen data is only available during the active season of the corresponding pollen type. The following pollen variables may be available:
| Pollen type | Variable | Season window* | Day of year |
|---|---|---|---|
| alder | ALNUsnc | Jan 8 – Mar 31 | 8 – 90 |
| ragweed | AMBRsnc | Jul 9 – Sep 30 | 190 – 273 |
| birch | BETUsnc | Mar 18 – May 25 | 77 – 145 |
| hazel | CORYsnc | Jan 8 – Mar 17 | 8 – 76 |
| grasses | POACsnc | Apr 1 – Aug 31 | 91 – 243 |
* For non-leap years; the day of year is used as the reference.
Note that pollen parameters are added dynamically depending on the currently active pollen types. If a specific pollen variable cannot be found, it may simply mean that the corresponding pollen type is not currently active and therefore not present in the dataset.
The pollen variables are provided as specific number concentrations — the number of particles per kg of air. To compare them with observations, which are usually expressed as volume concentrations, multiply the pollen concentration by the total air density DEN. The DEN field is also available for this conversion.
Pollen variables and DEN are available only on the lowest full model level, model_level=80, representing the mean value over approximately the first 20 m above ground.
The same pollen variables are also available in the ICON-CH2-EPS forecast datasets. The ICON-CH2-EPS control runs, initialized at 00, 06, 12 and 18 UTC, provide pollen forecasts up to +120 hours ahead.
For a practical example, see the Jupyter notebook, which demonstrates how to retrieve, convert and visualize pollen data.
Download options
Analysis data can be downloaded using the same methods as forecast data:
When requesting analysis data:
- select the analysis dataset collection
ch.meteoschweiz.ogd-analysis-kenda-ch1 - use step = 0 for instantaneous analysis fields or step = 1 for first-guess fields
- only the control member is available
In practice, choose the step based on the type of parameter:
- step = 0 → instantaneous variables (state of the atmosphere at time t)
- step = 1 → time-aggregated variables (accumulated, minimum, or maximum values over the previous hour)
You can identify the type of a parameter in the Overview of Parameters file (see Temporal Aggregation).
FAQ / Troubleshooting
Why are some accumulated variables not available at step 0?
Accumulated or time-aggregated variables represent values integrated over a time interval (for example precipitation accumulated during the previous hour).
Since the analysis represents the instantaneous atmospheric state, such variables cannot be defined at that time.
They are therefore provided from the first guess at step = 1, which represents the model evolution during the previous hour.
Why can't I find a specific pollen variable?
Pollen parameters are included dynamically depending on which pollen types are currently active.
If a pollen type is not present in the dataset, it likely means that the pollen season for that species is not currently active.