Changelog
v2.1.13
Added
-
Sun metric for history charts. Solar radiation, UV, and brightness each get their own view under the Sun metric, available on the standard chart and as heatmaps. A new Sun card also joins the Records page, showing the sunniest day, the highest UV, and the brightest moment your station has recorded.
-
Lightning distance in history. Your station reports how far away it judged each strike to be, but until now that only ever appeared on the panel for the most recent strike. The Lightning metric can now show average distance over time — the difference between a storm that passed twenty kilometres away and one that came directly overhead, two nights that otherwise look identical on strike count alone.
Changed
- Historical backfill complete. Version 2.1.11 introduced recording of solar radiation, UV, brightness, and lightning distance, noted that nothing charted them yet, and promised a way to recover this data for already-imported history. That backfill has now run for every station, rebuilding these readings from the raw observations already stored and reaching better than 99.9% of every hour on record. The remaining gaps are hours your station genuinely reported nothing at all, and stay empty because they should. As a result, the new Sun and lightning-distance charts — and the records drawn from them — cover your history from its beginning, not just from when recording started. No action is required: no re-import, no refresh, no settings to change.
Fixed
-
Fullscreen radar freezing on open. The radar fetched imagery once when opened and never requested more, so a radar left open through a storm kept showing the sky as it looked when first opened, with no indication anything was stale. (In the case that surfaced this, the view was over an hour behind, showing clear air above a station with heavy rain falling on it at the time.) Radar now refreshes every five minutes while open, and holds the current frame until the replacement has loaded instead of blanking in between. Scrubbing back to an older frame now stays put. This also fixes animation playback coming up empty after the radar had been left open for more than fifteen minutes.
-
Max gust timestamp corruption. Today’s strongest gust ran into its arrival time in the display, so a peak of 13 km/h at 1:30 PM read as “Max 131:30 PM.” Introduced by the timestamp added in 2.1.12.
Removed
- Hail indicator on the Rainfall panel (added in 2.1.11). It was wrong, and the claim made for it at the time — that the Tempest can distinguish hail from rain by impact signature — doesn’t hold up against the data. It’s marked as experimental anyway from Weather Flow. Checking a week of one station’s readings, it flagged hail on three separate August days, twenty-five minutes in total, and every one of those minutes fell during that day’s heaviest rain. It never once used the reading that indicates rain and hail together, even while those same minutes were putting down half a millimetre of water per minute — which is rain by definition. What the sensor actually reports is how hard it’s being hit, and heavy rain hits hard; there’s no delay or threshold that separates the two, because there’s nothing in the reading to separate. Rather than leave an indicator that turns red during any downpour, it’s been removed. The underlying sensor readings are still recorded, unchanged, in case the sensor’s judgment improves.