I have been capturing a lot of data regarding the sun lately…

… and a pattern has formed that some of you may find interesting…

In an earlier thread of mine, we observed how the “dayLength” changes dramatically around the Equinoxes, and it slows to a crawl around the Solstices…

To gather this new data, I used the new information found in:

`$twcweather.conditions.sunriseTimeLocal`

and

`$twcweather.conditions.sunsetTimeLocal`

This allows us to capture the SECOND the sun rises/sets.

*(If we use $sunset, it drops down to the nearest minute)*

So this new $twc option is much more accurate when talking about sunrise/set times.

The graph below shows (yesterday’s dayLength) minus (today’s dayLength).

*(101 days… one dataPoint per day)*

It is important to note that, because we are rounding four times to the nearest second…

*(yesterday’s sunset - yesterdays sunrise) - (today’s sunset - today’s sunrise)*

… then the final results may be up to two seconds off. *(which is why the graph is jagged)*

From the chart above, I would guess that day 58 *(the highest point)* was the Equinox… but it actually occurred 4 days earlier on day 54. Perhaps the green “trend” line helps us to see that:

*(the arrow points to the Equinox)*

If we zoom in, it is a bit easier to see…

Here is 68 days, centered around the Equinox, with each number *(on the Y axis)* representing one sec:

*(Notice the highest point in the trend line is on the Equinox, even though some dataPoints may be higher)*

I find it especially interesting that there are **SIX** days with a bigger change than the Equinox…

For the really observant, the slope of the angle changes *gradually* before the event, and a bit more *aggressive* afterwards…

Anyways, even with the analog data, the new $twc times for sunset still surpasses $sunset.

When I used $sunset, my dayLengthPercent would vary up to 0.43% on any given day…

Now, using $twc, it can only vary 0.007%. *(or within a half second of each event)*

Edit:

Updated images to show the recent data