>  but any such loss will be minor.

But often unnecessary and easily avoidable.

> Bottom line is users should not really need to use the --reweight option

I use the rebuild option when I have bad values for a single obs_type and 
don't want to lose any precision on min/max values and their dateTime, 
especially for other obs_types, when fixing them. I just have to ensure 
that I take care of the to-be-fixed abs_type min/max values correctly. I 
don't know any better way for a scenarios like I just described. I assume 
this doesn't apply on the user's scenario since he is missing entire 
archive_intervals, but anyway. 

As long as there is no option for wee_database, that manipulates min/max 
values and their dateTime only when it is necessary, I will not 
--rebuild-daily in such a scenario.

My bottom line: if you have power outages on a regular basis or even more 
than once the one or the other year, fiddling around correcting your 
database is not a good approach anyway. You should consider using 
a uninterruptible power supply (UPS) for the device weewx is running on 
(and, depending on the hardware and network setup/configuration, for the 
hardware, DHCP device and switch/access point) and hardware/driver that 
supports backfilling to the weewx database.





gjr80 schrieb am Mittwoch, 26. Juli 2023 um 10:28:31 UTC+2:

> Just a few points here. I'm sure it's just a typo that has perpetuated 
> itself through the thread, but the default WeeWX database is weewx.sdb 
> not weewx.sbd. Also, case matters here.
>
> The recommended approach for correcting spurious data in the database is 
> to locate and fix the spurious data in the archive table and then rebuild 
> the daily summaries for the period concerned using wee_database. This 
> procedure is covered in the Cleaning up old 'bad' data 
> <https://github.com/weewx/weewx/wiki/Cleaning-up-old-'bad'-data> wiki 
> page (note this page details the use of sqlite3 to manipulate the 
> archive, these steps can equally be completed using other tools such as DB 
> Browser for SQLite etc, also the page currently does not mention the use 
> of --from, --to or --date to limit the period of any daily summary 
> rebuilds). This may give some loss of max/min values and times but any such 
> loss will be minor.
>
> Fixing any spurious data in the daily summary tables by direct editing is 
> also possible, but it will likely be time consuming and is prone to error 
> and hence not recommended. Consider if there is one anomalous outTemp value 
> in a day, first you need to edit and remove the spurious data in the 
> archive. Then you would need to locate the day concerned in the outTemp 
> daily summary table and edit the max and maxtime fields replacing the 
> data with the new max outTemp and the time it occurred for that day 
> obtained by looking through that days temperature values in the archive, 
> perhaps looking through hundreds of records. If there are other days with 
> spurious data you would need to repeat this process for each day. Likewise 
> for other obs (remember some obs are derived from others so an anomaly in 
> outTemp may cause anomalies in heatindex, windchill, dewpoint etc rain 
> anomalies 
> may cause anomalies in rainRate). Thus just a few days of anomalous data 
> can result in many time consuming changes being required.
>
> Also, the --reweight option was implemented to fix an issue in earlier 
> WeeWX versions (post v3.0.0) where archive records in databases with 
> multiple different archive intervals were not correctly weighted in the 
> daily summary tables. This issue was fixed in v4.3.0 and providing you are 
> running v4.3.0 or later the reweighing of the database should only be 
> needed as a one-off 'patch' which should have been performed automatically 
> on startup of v 4.3.0 or later, all archive records subsequently 
> added/modified etc by WeeWX or wee_database are automatically correctly 
> weighted. Bottom line is users should not really need to use the 
> --reweight option. Note the --reweight option does not perform the same 
> function as --rebuild-daily.
>
> Gary
> On Wednesday, 26 July 2023 at 08:31:42 UTC+1 [email protected] wrote:
>
>> 1. Stop service on Raspberry PI. 
>>
>> 2. Copy WEEWX.SBD to your PC and open it with DB Browser for SQLite. 
>>
>>
>> 3. Using the website https://www.epochconverter.com/ find the 
>> approximate dateTime to find the position in the database and change the 
>> number in the table "archive". 
>>
>> 4. Adjust the maximum value and time in the table "archive_day_xx". 
>>
>>
>> 5. Write the changes to the file and copy it back to Raspberry PI.
>>
>> *6. Run sudo wee_database --reweight --date=YYYY-MM-DD (this affects only 
>> the particular day you changed values for)*
>> 7. Start the Weewx service. 
>>
>> The "reweight" option doesn't touch the min/max values and their 
>> timestamp, you took care of that yourself in (4.)
>>
>> [email protected] schrieb am Mittwoch, 26. Juli 2023 um 06:36:17 UTC+2:
>>
>>>
>>> How I hate Google Groups. It looks like I answer only to you Michael. 
>>> To be short - your explanation helps me a lot. Thank you. 
>>>
>>> I have one more question. Here is how I do it now when I want to change 
>>> something manually (for example the rain rate - which is quite often a 
>>> problem). 
>>>
>>> 1. Stop service on Raspberry PI. 
>>> 2. Copy WEEWX.SBD to my PC and open it with DB Browser for SQLite. 
>>> 3. Using the website https://www.epochconverter.com/ find the 
>>> approximate dateTime to find the position in the database and change the 
>>> number in the table "archive". 
>>> 4. After your explanation from yesterday change the number in the table 
>>> "archive_day_rainRate". 
>>> 5. Write the changes to the file and copy it back to Raspberry PI. 
>>> 6. Start the Weewx service. 
>>>
>>> Is there any better way?
>>>
>>> On Tuesday, 25 July 2023 at 07:45:43 UTC+2 [email protected] wrote:
>>>
>>>> The values are in the "archive_day_*" tables in the database. These 
>>>> tables contain a dateTime values,  min and max values as well as their 
>>>> exact time (depending on the source), sum and weighted sum values and a 
>>>> count. When you change a value in the archive table by hand, it won't show 
>>>> up there, thus not show up in any statistics in the front end, as long as 
>>>> you don't fix this in the proper way. 
>>>> See https://weewx.com/docs/latest/utilities.htm#wee_database_utility 
>>>> for a description of a tool that does it the proper way. Be aware, that 
>>>> some actions might lead to loss in precision. For instance, the "exact 
>>>> time" I mentioned before is then changed to the dateTime of the archive 
>>>> value, e.g. when you maximum gust occurred at 8:33pm using a 5 min 
>>>> archive_interval, it might happen, that after running the tool you maximum 
>>>> gust will be shown at 8:35pm. Or, if your maximum temp was at 4:32pm 
>>>> showing 30,1°C with a single loop value of 30,1°C, and all other loop 
>>>> values were 30,0°C in that archive_interval, after running the tool, it 
>>>> might be, that your maximum will read as 30,0°C at 1:15pm, if that was the 
>>>> first interval that day, that had 30,0°C.
>>>>
>>>> [email protected] schrieb am Dienstag, 25. Juli 2023 um 06:37:12 UTC+2:
>>>>
>>>>> I had some problems with connections and also physical shocks on 
>>>>> sensors. 
>>>>> So, I stopped the Weewx service, copy the file 
>>>>> /var/lib/weewx/weewx.sdb to my PC. I opened it with SQLiteDatabaseBrowser 
>>>>> and change some extreme (unreal) results. For example wind and also 
>>>>> rainRate which was caused by the son's ball :-)
>>>>> After that, I copy the file back onto my server and start the Weewx 
>>>>> service again. 
>>>>> Now some changes are properly shown (for example in graphs), but the 
>>>>> max results are old (unreal). I don't understand how this is calculated. 
>>>>> I 
>>>>> even tried "xstats" and it also shows old (unreal) max results. I checked 
>>>>> again in the database and there are no old results there. 
>>>>>
>>>>> Any idea how to get rid of "unreal" results which should not apear in 
>>>>> the database (and AFAIK they are not there anymore)?
>>>>>
>>>>> Here is the monthly graph where (for example) where we can see wind 
>>>>> (VETER) max on the 18th and 19th of July. It was 65 km/h:
>>>>> https://izo.amebis.si/belchertown/graphs/?graph=month
>>>>>
>>>>> Here is a summary where the max wind for July 2023 is 50 km/h:
>>>>> https://izo.amebis.si/belchertown/
>>>>>
>>>>> It is similar to the rain rate. Max is 13107.0 mm/hr (unreal and I 
>>>>> can't find this result in the database (I sorted the rain rate column, 
>>>>> tried to check for the particular time ... there is no such data 
>>>>> anymore). 
>>>>>
>>>>> Regards.
>>>>
>>>>

-- 
You received this message because you are subscribed to the Google Groups 
"weewx-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/weewx-user/738e62eb-9cd4-4dd7-9a10-13113d1e7e1en%40googlegroups.com.

Reply via email to