Details from the National Weather Service
* WHAT...Temperatures as low as 35 will result in frost formation. * WHERE...Portions of northern Indiana and northwest Ohio. * WHEN...From 1 AM EDT /midnight CDT/ to 9 AM EDT /8 AM CDT/ Monday. * IMPACTS...Frost could harm sensitive outdoor vegetation. Sensitive outdoor plants may be killed if left uncovered. Take steps now to protect tender plants from the cold.
How this forecast works
The temperature correction is a straight-line formula, so every degree can be traced (see "Why ours is…" in the latest run in detail). Rain chances come from a probability model of the same inputs, and a small AI model on the station computer reads the National Weather Service's written forecast discussion and adds what it says about rain. Every piece has to pass a fair test before it is called tested.
How we test
- Practice and grading are kept apart. A formula is built from one stretch of the station's past, then graded on a later stretch it was never built from, like a student who practices on one set of problems and is graded on another. A formula that only looks good on its own practice data is usually memorizing noise, so only the grade on set-aside data counts.
- The passing mark is written down first. Before grading, the Lab Notebook records exactly what counts as a pass: for example, at least 0.05° better than what we use now. The result is published either way, pass or fail.
- Luck is checked. A small gain can be luck. So the graded days are re-drawn at random thousands of times, and a gain has to hold in at least 9 of 10 re-draws.
- New days settle it. Some features then run "on trial": logged every day as the weather happens and judged again once enough new days have passed.
How accurate it is
| Days ahead | High: our forecast | Model average | Improvement | Low: our forecast | Model average | Improvement |
|---|---|---|---|---|---|---|
| 1 (tomorrow) | 1.6° | 2.0° | 21% | 1.7° | 1.9° | 12% |
| 2 | 2.0° | 2.5° | 22% | 2.0° | 2.3° | 13% |
| 3 | 2.4° | 2.8° | 15% | 2.3° | 2.6° | 10% |
| 4 | 2.8° | 3.2° | 13% | 2.5° | 2.8° | 9% |
| 5 | 3.2° | 3.6° | 10% | 3.1° | 3.4° | 7% |
| 6 | 4.0° | 4.3° | 7% | 3.8° | 4.0° | 5% |
| 7 | 4.8° | 4.9° | 3% | 4.7° | 4.9° | 4% |
"Average error" is how far off the forecast was, on average, in degrees. Day 1 is the next day's high and the coming night's low. The biggest gains are 1 to 3 days out, where local detail matters most; by day 7 the big-picture pattern dominates and there is less for a local correction to fix.
| Days ahead | Daytime | Overnight |
|---|---|---|
| 1 | 38% | 32% |
| 2 | 36% | 37% |
| 3 | 31% | 34% |
| 4 | 28% | 34% |
| 5 | 26% | 31% |
| 6 | 25% | 22% |
| 7 | 11% | 14% |
0% would be no better than simply knowing how often it rains at this time of year; 100% would be perfect. Since October 2026 the rain chances also use the NWS forecast discussion, which tested 3.4% better again over the next two days.
| Next day's | Our forecast | Model average | NWS | WxSim | Days scored |
|---|---|---|---|---|---|
| High | 2.1° | 2.6° | 3.0° | 5.5° | 3 |
| Low | 0.2° | 1.2° | 1.5° | 2.1° | 4 |
Average error, in degrees, of the last forecast each source had out before the period began: before the daytime high's period starts in the morning, and before the overnight low's starts in the evening. Every source is held to the same moment, so no one gets credit for updating once the answer is showing. With only a handful of days so far these numbers swing a lot; they become meaningful after a few months.
Once a period has begun, the forecast keeps updating every hour with how warm or cool the station is running. That update passed a test on a year of days set aside for grading; these checkpoints keep checking it on real days, against the forecast made before the period began.
Collecting since October 19, 2026. The first numbers appear after the first full day.
Every finished high and low, graded on the forecast made before it began, with the conditions of the day, so big misses can be studied as a group. One day is mostly bad luck; a pattern across many days is something to fix, and it gets tested in the Lab Notebook before anything changes.
| Periods | Typical miss | On average came in | Outside the ± range | Cloudy or wet days outside | 5°+ misses | |
|---|---|---|---|---|---|---|
| Highs | 3 | 2.1° | -1.4° colder | 33% | 50% of 2 | 0 |
| Lows | 4 | 0.2° | -0.1° colder | 0% | 0% of 2 | 0 |
The ± is an 80% range, so about 1 in 5 should land outside it. Much more than that means the range is too narrow.
| Period | Our forecast | Actual | Miss | Model average | NWS | Models disagreed by | Forecast cloud | Sunshine measured | Wind | Dew point swing | Rain |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Fri Oct 16 high | 56.6° ±2 | 54.0° | -2.6° | 56.4° | 57.0° | 0.6° | 90% | 17% of clear sky | 7 mph | 3.1° | 0.02″ |
Miss = actual minus our forecast (+ means it came in warmer). "Models disagreed by" is how far apart the four models were (a typical day is about 2°). Sunshine is what the station's sunlight sensor measured, as a share of a cloudless day.
Looking back over 2½ years of forecasts made the same way (research note R-001): the ± for the high is honest on ordinary days but too narrow on cloudy and wet days (about 1 in 3 outside instead of 1 in 5), and most misses of 5° or more were misses all four weather models made too, usually when the clouds came or went at a different time than forecast.
What it's built from
| What the station computer saves | Saved | Since | Latest |
|---|---|---|---|
| Global model runs | 132 | Oct 13, 2026 | Oct 18, 6:43 PM |
| Upper air & sunshine runs | 50 | Oct 14, 2026 | Oct 18, 6:43 PM |
| Ensemble runs (122 members) | 21 | Oct 14, 2026 | Oct 18, 2:08 PM |
| HRRR runs | 91 | Oct 14, 2026 | Oct 18, 6:43 PM |
| NWS gridpoint forecasts | 72 | Oct 13, 2026 | Oct 18, 3:25 PM |
| NWS text forecasts | 97 | Oct 13, 2026 | Oct 18, 6:43 PM |
| WxSim runs | 32 | Oct 13, 2026 | Oct 18, 1:37 PM |
| National surface analyses | 33 | Oct 14, 2026 | Oct 18, 6:27 PM |
| NWS alert checks | 9 | Oct 16, 2026 | Oct 18, 6:43 PM |
| Storm outlooks (day 1) | 15 | Oct 16, 2026 | Oct 18, 6:43 PM |
A new copy is saved only when the content changes, so the counts are real updates, not repeats.
Ensemble archive: 4 days of the full 122-run ensemble saved, of about 180 needed before calibrating it to this station can be tested fairly.
What's proven, what's on trial
Every feature on this site, and how far it has been tested. "Proven" means it passed a fair test. "On trial" means it is being checked on new days as they happen. "Experimental" means it is shown for interest and not yet tested. Badges link to the test in the Lab Notebook.
| Feature | Status | Where |
|---|---|---|
| Highs and lows, with their ± ranges | Proven | Today, Forecast |
| Rain chances, with the NWS discussion | Proven | Today, Forecast |
| Today's high and tonight's low, adjusted live | Proven | Today |
| The 6-hour nowcast (the station's error and the upwind airports) | Proven | Forecast |
| Today's high and tonight's low updated with the station once the period has begun | Proven | Forecast, Today |
| Frost and freeze chances; gust and heavy-rain levels | Proven | Today, Weather Watch |
| The hour-to-hour shape between high and low | Models' own | Forecast |
| Physics model: night low, blended with our low | Proven On trial | Physics |
| Physics model: day high (with the crop calendar) | Proven | Physics |
| Physics model: day high blended with ours | On trial | Physics |
| Afternoon update from measured sunshine | On trial | Physics |
| Rain, snow and freezing rain in the physics model | Experimental | Physics |
| Grass temperature estimate at night | Experimental | Forecast, Farm |
| Harvest hours, drying, spray windows, your-corn drydown | Experimental | Farm |
| Snow on the ground estimate | Experimental | Ground & Snow |
Limits worth knowing
- Clouds come from the big models. Nothing here forecasts clouds on its own. When the models get the clouds wrong, so does this forecast: on days much darker than forecast the high misses by about 2.5° on average, against about 1.4° when the sunshine is right.
- One station, one spot. Everything is tuned to this station's mast, beside a field that is corn or soybeans by year. Its surroundings (crop height, harvest) change its readings, and it can run warmer or cooler than airports a few miles away.
- The rain gauge is unheated. Snow counts only when it melts, and the gauge's record has a known glitch (September 20 to 25, 2026) that is replaced from a reanalysis where it matters.
- Short records are less certain. The ground sensor went in on October 17, 2026; anything built on it (grass frost, dew timing, inversions) is still being checked.
- Ranges are honest, not guarantees. An 80% range misses one time in five, by design.
Going deeper: the latest run in detail (every input, the forecast discussion, why each number is what it is, run-to-run changes, recent verification) and the model charts (hour by hour, and the upper air).
Retraining
Four times a year the station computer refreshes everything (the station's minute logs, the models' past forecasts, the forecast discussions), checks the station history for glitches like September 2026's false "rain days", re-runs the tests and trains a new model. The new model only replaces the current one if it does at least as well on the days since the last training, days neither model has seen. Otherwise the current one stays. The old model is always kept, so going back is one step.
- Friday, January 1, 2027 at 4 AM (next)
- Thursday, April 1, 2027 at 4 AM
- Thursday, July 1, 2027 at 4 AM
- Friday, October 1, 2027 at 4 AM
| When | Outcome | Test on new days | Why |
|---|---|---|---|
| Oct 16, 2026 | Kept current | – | not enough new days yet for a fair test (3 of 45) |
| Oct 16, 2026 | Kept current | – | not enough new days yet for a fair test (3 of 45) |
Glossary
80% range (±). The range the actual value fell inside four times out of five when tested.
Average error. How far off a forecast was, on average, in degrees, ignoring whether it was too warm or too cold.
Model average. The four global models (GFS, ECMWF, ICON, GEM) averaged at the station. "Our forecast" is that average after the station correction.
Ensemble. A model run many times with slightly different starting points; the spread of the runs shows the uncertainty.
Nowcast. A forecast for the next few hours that leans on what the station and nearby airports are reading now.
Frost. Ice forming on grass, roofs and windshields. It needs the grass at 32° or colder, which on a clear, calm night can happen with the 5-foot air at 36° or warmer.
Freeze. The air at 5 feet reaching 32°. A killing freeze is about 28° for a few hours, enough to end the season for tender crops.
Grass temperature. The air 2 inches above the grass, measured by the station's ground sensor (or estimated at night).
Dew point. The temperature at which the air becomes saturated; dew or frost forms when the grass cools to it.
Wet-bulb temperature. How cold the air gets when water evaporates into it. Falling snow can survive to the ground in air a few degrees above freezing when the wet-bulb is near 32°.
Temperature inversion. Air near the ground colder than the air above, common on calm, clear nights; it traps smoke, fog and spray drift.
Delta T. The air temperature minus the wet-bulb: how quickly spray droplets evaporate.
925 mb, 850 mb, 500 mb. Pressure levels about 2,500, 5,000 and 18,000 feet up, where forecasters watch the air mass.
CAPE, lifted index, K index, Total Totals. Measures of how unstable the air is, which is how much energy is available for thunderstorms.
Growing degree days (86/50). Each day's average temperature above 50°, with highs capped at 86°: the heat units crops develop on.
Physics model. Our own model of the heat budget at the station mast (sun, sky, ground, air), run beside the forecast.
Reanalysis (ERA5). A best reconstruction of past weather from all observations, used for normals and for checking.
HRRR, NBM, MOS. A fast-updating high-resolution American model; the National Blend of Models; and Model Output Statistics, the National Weather Service's statistical corrections of model output.