One row per day: the station's actual high and low in bold, and beside each what every source said. Pick a goal and a cut-off; a green cell is a forecast that landed within the goal, and the top row counts them. The cut-offs are the Scoreboard's: the last forecast each source had issued before the period began, the one it had out the day before, and two days before. The main model is the candidate on trial (L-038).
| The high | The low that night | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Day | Actual | Ours | NWS | WxSim | Avg | Main | Actual | Ours | NWS | WxSim | Avg | Main |
| Within the goal | ||||||||||||
| Tue Nov 10 | 66.0° | 34.2° | ||||||||||
| Mon Nov 9 | 61.2° | 40.5° | ||||||||||
| Sun Nov 8 | 59.7° | 36.9° | ||||||||||
| Sat Nov 7 | 51.3° | 31.6° | ||||||||||
| Fri Nov 6 | 50.2° | 36.9° | ||||||||||
| Thu Nov 5 | 56.5° | 45.0° | ||||||||||
| Wed Nov 4 | 68.5° | 50.9° | ||||||||||
| Tue Nov 3 | 76.5° | 61.5° | ||||||||||
| Mon Nov 2 | 58.8° | 58.1° | ||||||||||
| Sun Nov 1 | 68.2° | 55.0° | ||||||||||
| Sat Oct 31 | 66.7° | 51.4° | ||||||||||
| Fri Oct 30 | 64.8° | 52.3° | ||||||||||
| Thu Oct 29 | 79.3° | 39.6° | ||||||||||
| Wed Oct 28 | 75.4° | 50.9° | ||||||||||
| Tue Oct 27 | 73.2° | 46.8° | ||||||||||
| Mon Oct 26 | 73.9° | 48.9° | ||||||||||
| Sun Oct 25 | 75.6° | 40.8° | ||||||||||
| Sat Oct 24 | 67.5° | 38.8° | ||||||||||
| Fri Oct 23 | 69.1° | 41.4° | ||||||||||
| Thu Oct 22 | 70.0° | 37.8° | ||||||||||
| Wed Oct 21 | 71.8° | 38.3° | ||||||||||
| Tue Oct 20 | 70.7° | 38.8° | ||||||||||
| Mon Oct 19 | 71.2° | 35.4° | ||||||||||
| Sun Oct 18 | 68.4° | 34.0° | ||||||||||
| Sat Oct 17 | 55.8° | 31.5° | ||||||||||
| Fri Oct 16 | 54.0° | 36.0° | ||||||||||
| Thu Oct 15 | 66.9° | 52.9° | ||||||||||
| Wed Oct 14 | 57.6° | |||||||||||
What the columns are
- Actual
- What the station measured: the day's high in its window (7 AM to 7 PM) and the night's low in its window (7 PM to 8 AM), the same windows the Scoreboard grades.
- Ours
- The forecast the site publishes: the global models blended and then corrected by what the station's own record says about days like the one ahead (the "statistical forecast" on How it was made).
- NWS
- The National Weather Service's gridpoint forecast for this spot, in whole degrees.
- WxSim
- The WxSim run on hand at the cut-off; blank when no run had been collected in time.
- Avg
- The four-model average: the GFS, ECMWF, ICON and GEM global models' own highs and lows for the station's grid point, averaged, with nothing corrected. Ours and the main model are both built on it.
- Main
- The main model, the candidate on trial (L-038): our own physics column run on the station's readings, corrected by a line fitted on the station's record that also leans on the four-model average. If it beats Ours by May 2027 it becomes the forecast. It issues the day and the night ahead only, so it has no two-days-before column.
- Cut-off
- Which forecast is graded: the last one each source had out before the period began, the one it had out a day before that, or two days before. Every source is held to the same clock.
- Goal and colour
- A green cell landed within the goal of the actual; a grey one did not; a blank is a source with nothing out at that cut-off. The top row counts the green cells per column. "Forecast minus actual" shows each cell's miss instead, plus warm, minus cold; hover a cell for the other.
Last 45 days with a finished period. The NWS forecasts whole degrees; a blank cell is a source that had no forecast out at that cut-off (WxSim's runs are not always on hand, and the main model issues the day and the night ahead only, so it has no two-days-before forecast). Ours is the forecast the site publishes. "Within the goal" counts the days each source landed inside the chosen goal at the chosen cut-off. The Scoreboard has the averages and the year-long test.