Question on heat index

[Pat O'Brien]

I thought heat index wasn't calculated unless the temperature was over 80F? Today it's about 49 F and I'm seeing heat index in my charts (which is reporting lower than temperature - I thought that was wind chill?). I don't recall seeing it before at this temperature range. 

I just want to make sure this is expected for temperatures so low to have a heat index reading?

Top graph is Seasons skin and bottom is Belchertown skin. The values match across skins

[gjr80]

Pat,

v4.2.0 saw a new heatindex formula implemented, refer issue 601. The new formula gives some results in some circumstances that may appear counterintuitive. v4.3.0 added back the ‘old’ heatindex formula and gives the user the ability to choose between the two algorithms . This is covered here in the User’s Guide under [[[heatindex]]].

By the looks of it you are seeing the ‘new’ algorithm.

Gary

[Pat O'Brien]

Thanks Gary! I've got some catching up to do it seems

[Les Niles]

It looks strange to see the heat index suddenly dropping a few degrees below the temperature when the temperature gets above 40ºF. (I just saw this this morning.) After looking through the thread on weewx’s new heat index calculation, I noticed an oddity at the National Weather Service: while their calculator calculates a heat index down to 40ºF, on the page where they describe the HI equation they say the “Rothfusz regression is not valid for extreme temperature and relative humidity conditions beyond the range of data considered by Steadman.” Looking at Steadman’s sultriness papers from 1979  he never looked at temperatures below 68ºF. I couldn’t find any reference to indicate why they arbitrarily extended the javascript calculator down to 40ºF. Other authors, and the NWS tables, suggest that the heat index isn’t really of value below 80ºF.  

So I tweaked dewpointF() in my installation to return None if the temperature is below 68, and now I feel better.  

  -Les

--
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 weewx-user+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/weewx-user/f46bd022-500d-414e-b96b-c2bc382d37b9n%40googlegroups.com.

[n7uv...@gmail.com]

Hi there Les - less than two days ago I updated my weewx from an older 4.xx version to the current one, and suddenly i was getting the same weird plot. Finally typed in "heat index wind chill" into the forum and out popped this thread. I want to feel better as well - where is the spot where you tweaked DewpointF()?

Cheers - Jon

[Les Niles]

First, I misspoke — the change is in heatindexF(), not dewpointF() (duh!)

In the weewx software directory (/usr/share/weewx when installed as a Debian package), there is a file weewx/wxformulas.py.  Look for the line "if T is None or R is None:” — it’s line 164 in the 4.4.0 version.  Add “ or T < 68”, so that the line becomes “if T is None or R is None or T < 68:”  You could change 68 to 80 or whatever you want the low-temperature threshold to be.  

WARNING: this is the wrong way to maintain weewx software.  The change will get overwritten when weewx is updated.  There are better ways to implement this, but that would’ve taken me more time than I wanted to spend so I’m OK with the hack for my installation. But use at your own risk. 

  -Les

To view this discussion on the web visit https://groups.google.com/d/msgid/weewx-user/c9ac3f2e-8a1b-4dc5-8168-74fff1cc44e9n%40googlegroups.com.

[Pat O'Brien]

You don't have to modify any code, just change the calculation formula by setting this option in weewx.conf

Look for the [StdWXCalculate] section, then beneath add

    [[WXXTypes]]

        [[[heatindex]]]

            algorithm = old

That put it back to the "old" method for me where heat index is calculated above 80F.

[n7uv...@gmail.com]

Thanks so much, Les. I recognize that it's ephemeral, but your guidance helped me learn a few things about weewx that might help me do some other calcs!

Cheers - Jon

[n7uv...@gmail.com]

that makes sense also. I will also play with that!

[Les Niles]

That would be the sensible thing to do.  But it’s a different algorithm, not just a change in the low-temperature limit.  The new algorithm is closer to Steadman’s original concept of giving a perceived temperature relative to an average humidity, so low humidities can result in a heat index that’s lower than the ambient temperature, which will never happen with the old algorithm.  

The other subtle difference is that my modification returns None when the temperature is below threshold, rather than returning a heat index equal to the temperature.  That means there are no abrupt jumps in the value as temperature changes slightly; the heat index just isn’t reported.  

  -Les

To view this discussion on the web visit https://groups.google.com/d/msgid/weewx-user/cd966cfa-772f-4190-b177-3180fbe8b0d3n%40googlegroups.com.

[n7uv...@gmail.com]

After a few days of running with the heatindexF(), and with relatively low temps here in PHX, all was fine. Then yesterday it spiked up to 69+ deg F, and the ~ 2F difference came back for the duration of the heat wave %^)

Given how dry it is here, I think that I'll switch to the old method - at least *I'll* feel better %^)