Spike removal bug

[perlamit]

Hi.
In recent days I experience unstable temperature reading from my weather station. I don't know what is the root cause for this (replaced battery, clean connectors, reset to transmitter and receiver etc.) so I start using the spike removal tool and set it first to 10ºC and when the spike continue I set it to 5ºC.
I found bug using this tool (Spike removal):
1st scenario is when the system detects a Spike and did remove it off from the data that written (tables) but not remove it from the graph.
2nd scenario is that the Apparent temperature, the heat index and the dew point not been remove from data and graph.
Please see attached file for example (you can see the day before only in the graph, day with large spikes).
So I hope that you can fix those bugs.
Regards,
Amit.

I am using: Cumulus 1.9.4-1076

[steve]

It's not possible for the values removed by the spike removal to appear on the graphs; if they are removed, they are completely ignored. The graphs are plotted using the same values that are logged. I can't tell which lines are which on your graph because you haven't shown the legend, but it appears that the red line is the outside temperature, and I can't see anywhere on the graph with a change of more than 5 degrees. The value of 30.3 that was removed at 0245 does not appear on the graph, which is contrary to what you are suggesting.

Similarly, if a temperature is ignored because of spike removal, it is not possible for that value to be used in the calculations of dew point etc. I can't see any evidence on the graph you have shown where this is not the case.

Perhaps I am missing something - can you explain further what makes you think there is a problem?

[perlamit]

Thank you for your response.
Could you please tell me what are the spike in the below graph.
I translate the legend for you hope that you can read it now.
you can see that the blue (Heat index) and the Purple ) Appear temperature are not related to reality.
now can you see the problem?

[mcrossley]

What does your humidity graph show?

[steve]

If the outside temperature increase/decrease is less than the spike removal setting, then the value is ignored and not used in the calculation of heat index etc (the last good value is used). There are no spike removals in the log for a sudden increase in temperature; it's hard to see exactly what's going on in the graph you've shown, but assuming there was not a sudden increase in outside temperature of more than 10 degrees, then the reading would be accepted and used in the calculation of heat index (etc). Similarly for your humidity reading (which you haven't shown the graph for), which may also have increased at the same time (but still have been within your spike removal setting) and hence contributed to the increase in heat index.

There is no separate spike removal setting for heat index, as it is calculated from temperature and humidity. If the temperature and humidity are accepted (i.e. not ignored because of spike removal) then whatever those values are will be used in the calculation of heat index. Are you saying that the heat index calculation is wrong? Or are you expecting the outside temperature spike removal value to be applied to heat index etc?

[perlamit]

yes In my opinionת I am saying that the calculate base on wrong data that have been remove by the spike removal.
attached Humidity graph.
example at ~10:30: temperature was: ~30ºC , Humidity was: ~75% -> heat index supposed to be : 36ºC but it: ~47ºC.
If I try to calculate it from the dew point (25ºC) I get the same value, 36ºC. so why it is written 47ºC?

[steve]

Turn off everything except temperature and heat index on the temperature graph, and zoom it to within a few minutes of that peak heat index value at around 1045, then attach the image here, please.

[perlamit]

I made a new graph using select a graph tool.
As you can see @graph below, it is not relative to the first graph. where is the 45ºC-50ºC spike (time interval 10:30 - 11:00)?

[steve]

The main graphs are plotted at 1-minute intervals. The logs are at whatever interval you have configured, in your case it looks like 15 minutes. It was the main graph I was asking for; the same one you posted earlier, but so it's possible to see what the actual values were.

[perlamit]

hope that this one will be good...
I add Values

[steve]

At 1045, your temp was about 34, and your RH 76. That's a heat index of 50 C.

At about 1330 the temp was about 37, RH 66. That's a heat index of 55 C.

At about 1345, your temp was just over 35, say 35.5, and your RH was about 75%. That's a heat index of 55 C.

All of those calculations agree approximately with the graphs.

[perlamit]

Ok thank you Steve.
SO I have to find what is the root cause of the temperature spikes.
i have ws1080 weather staion.
BTW you are welcome to visit my web site that base on your program link : http://hanaton.tk/
Many thanks,
Amit