Additional features for OpenDAP handling

[Neil Davis]

We have an API that converts a GeoJSON polygon bounding box to the indices used in the OpenDAP queries used to download of the mesoscale time-series. The query will return an XML file that has a complex data element, which contains the indices for the west_east and south_north dimensions. This can then be substituted into the download.

https://wps.neweuropeanwindatlas.eu/cgi-bin/?SERVICE=WPS&VERSION=1.0.0&REQUEST=Execute&IDENTIFIER=opendap_helper&DataInputs=location={"type":"Polygon","coordinates":[[[14,52],[16,52],[16,54],[14,54],[14,52]]]}

Additionally, several users have asked about using the OpenDAP service with Python. We have used XArray for this purpose, using a command like below:

ds = xr.open_dataset("http://opendap.neweuropeanwindatlas.eu/opendap/newa/NEWA_MESOSCALE_ATLAS/2015/NEWA-2015-12-30.nc?WS[0:1:47][2:2][674:1:674][628:1:628],WD[0:1:47][2:2][674:1:674][628:1:628],time[0:1:47],height[2:2],west_east[628:1:628],south_north[674:1:674],XLON[674:1:674][628:1:628],XLAT[674:1:674][628:1:628],Times[0:1:47],crs")

 

[Jonathan Wallin]

On 09/07/2021 at 6:55 AM, Neil Davis said:

Additionally, several users have asked about using the OpenDAP service with Python. We have used XArray for this purpose, using a command like below:

ds = xr.open_dataset("http://opendap.neweuropeanwindatlas.eu/opendap/newa/NEWA_MESOSCALE_ATLAS/2015/NEWA-2015-12-30.nc?WS[0:1:47][2:2][674:1:674][628:1:628],WD[0:1:47][2:2][674:1:674][628:1:628],time[0:1:47],height[2:2],west_east[628:1:628],south_north[674:1:674],XLON[674:1:674][628:1:628],XLAT[674:1:674][628:1:628],Times[0:1:47],crs")

 

 

Hello,

I guess the URL is outdated since the API-adress is now: https://wps.neweuropeanwindatlas.eu/api/mesoscale-ts/v1/get-data-point?... Correct?

I have got it to work to download 30 min. data for specific locations with Python, but have some issues:

- It is slow. Downloading the 4-D variables for (1.) one height, (2.) one location and (3.) one year does take something like 20 min. Is it supposed to take that long?

- Because of that I have opted for downloading two years, and 2-3 heights at the same time, but either a year or a height is left out. Why?

Do you maybe have any new URL for downloading with Python now?

[Neil Davis]

Hi

We are aware of some performance issues. We are currently focusing on processing the remaining 20 years of NEWA data, and then will revisit performance. For now, it seems that best performance is found when downloading data in 6-month intervals, as this matches best with the resources on the server.

You have the correct URL. I am not sure why data would be provided with some data missing, as it should either succeed or fail. Could you provide a list of the variables that you are using so that I can do some testing? I will be on holiday starting Wed, so won't get a chance to look until the new year unfortunately.