meteohub.de

Hi,

Last weeks my WS2350 everyday records extremely low temperature values, so I have to delete these wrong data from the "Inspect Data" panel. The problem is that, although the wrong data are deleted and the HTML variables [e.g. day1_th0_ temp_max] produce the right values, the graphs on the other hand are still displaying the false values. The only way to make the graphs display the right data is to start a recomputation, which unfortunately takes very long time. So, my question is: is there any fast, immediate way (after deleting some data) to update the graphs without making a whole data recomputation?

Thanks

Hi,

Short answer: No

Long answer with background information:
The re-computation of weather data is something that is called in database terminology "aggregates". The lengthy process analyzes all data, and builds summary data. This is the reason that generating a graph goes very quickly, as it only has to go through the summary data.

So long story short: if you remove some of the data using the inspect function in the gui, you are effectively only altering the "base" data. The aggregates are untouched. Since the graphs are using the aggregates, you will still get the --wrong-- results based on the data before you deleted the erroneous records.

Ysbrand

Thank you YJB for the instant reply.
So it would be very useful if there was the ability to generate a graph immediately from the "base data", not only from the "aggregated". The strange to me was that, even though the graphs are updated on a scheduled interval (e.g. every 5 minutes), they cannot integrate any changes that may have been done on the raw data in the meanwhile (e.g. the deletion of some false records).
So, all these mean, I suppose, that I cannot do anything to solve this issue except changing weather station? Any other suggestion?

Hi,

Changing to a different weather station will not make a difference, as Meteohub will apply this process to any weatherstation.

I don't think you will be very happy if the graph was directly generated from the raw data, as this would force it to go through all the data, which will cost a lot of resources.

Might be a better idea to evaluate the reason of the "bad" data, and how to prevent this from happening in the future.

Ysbrand