In the following you will find guidelines for using the Frost API. The API has good capacity and can handle relatively large volumes of requests, but the capacity is not unlimited. We encourage all users to respect the guidelines so that the API will be stable. If we see that there are situations that threaten the operation of our environment, we would consider blocking out the users / services that have high load / bad traffic patterns. If you set up your service in accordance with our policy it will increase the probability that it is not blocked.
Use the information found in any cache headers. For example, check the x-frost section of the header for cache hits, misses, and and noncacheables. You can see how long data will be cached in the cache-control section of the header.
Use a proxy if possible. This is especially important for "slow clients" such as mobile devices.
Do not ask too often. Avoid continuous updating of mobile devices. Applications on mobile devices should download data when explicitly requested by the end user (by pushing a "refresh" button, for example).
You should save the information on your own server if you have heavy traffic. Heavy traffic means more than 20 requests from the API per second. Please contact us if you find that your service generates more traffic than this and you have not been able to reduced the amount of requests by taking into account our guidelines.
If your request is too large, you will get a timeout error. Please split your request into smaller subrequests. For example, a request that specifies data from 1000 weather stations could be split into 10 requests, each specifying 100 stations. Likewise, a long timeseries of observations can be split into requests for N time subranges.
The API requires a unique client ID for identification. Each client ID must be associated with a valid email address. It is possible to register multiple client IDs for each email address.
Data from frost.met.no are licensed and the license is found in every response. Read more about licensing and crediting here.
All weather data available through the API are continually updated. New data is available continuously (once it is in the database, but some types of data go through quality control first). Other products have a varying length of phase-out period, depending on what we consider to be reasonable. There are no guarantees of delivery regarding this service.
The version is specified both in the URL and in the response of each endpoint (service). Whenever the API is changed in a way that is not backwards compatible, the version number will be incremented.