Somerset Weather · Hyper-local Forecast

Our watches the next 7 days

Today
low 38° tonight
Frost13%
small chance
Freeze0%
unlikely
Friday
high 69°, low 39° that night
Frost10%
small chance
Freeze0%
unlikely
Saturday
high 68°, low 44° that night
Frost0%
unlikely
Freeze0%
unlikely
Sunday
high 63°, low 37° that night
Frost36%
good chance
Freeze4%
unlikely
Heavy rainElevated
daytime; happened 1 in 6 times at this level
Monday
high 47°, low 29° that night
Frost99%
very likely
Freeze85%
very likely
Tuesday
high 43°, low 28° that night
Frost99%
very likely
Freeze87%
very likely
Wednesday
high 40°, low 32° that night
Frost79%
likely
Freeze54%
good chance
Snowpossible
afternoon, night; from the models' ingredients

Each tile is a day and the night that follows it. Frost and freeze are our tested odds for the night (a night given a 70% chance has frost about 70% of the time); the heavy-rain and gust levels are tested too. Storms, fog, snow, ice, cold and heat come from the models' ingredients and are not tested. The National Weather Service's own watches and warnings show at the top of every page; the ingredients are on Watch.

The next 7 days

DayHighLow that nightRainFrost on the grass that night
Today was 66° ± 3° now 38° very unlikely (7%) unlikely (13%)
Friday ± 3° ± 3° very unlikely (7%) unlikely (10%)
Saturday ± 3° ± 3° very unlikely (7%)
Sunday ± 5° ± 4° likely (63%) possible (36%)
Monday ± 5° ± 4° unlikely (11%) very likely (99%)
Tuesday ± 5° ± 5° unlikely (13%) very likely (99%)
Wednesday ± 6° ± 7° possible (39%) likely (79%)

The ± after each temperature is the range the real temperature lands inside 8 times in 10: 72° ± 3° means somewhere from 69° to 75°. Rain is the higher of the day's and the night's chances of at least a hundredth of an inch in the station's gauge. What a chance means: a 30% chance of rain means that on days with a forecast like this one, it rains on about 3 out of every 10. It does not mean rain for 30% of the day, or over 30% of the area. Tap any high or low to see how we got it, or see them all under Why we say this below.

Highs and lows at a glance

The blue line is our forecast, and the shaded band is its ± range. The green squares are the National Weather Service forecast.

The forecast in words

Tonight

Clear, with a low around . Southwest wind around 5 miles per hour.

Friday

Sunny, with a high near . South wind 5 to 10 miles per hour.

Friday night

Clear, with a low around . South wind around 10 miles per hour. Wind chill values as low as 36.

Saturday

Sunny, with a high near . South wind around 10 miles per hour.

Saturday night

Mostly clear, with a low around . South wind around 10 miles per hour.

Sunday

Rain likely, mainly after 2 PM. Mostly cloudy, with a high near . South wind around 10 miles per hour. Chance of rain 63%. Rain amounts between a tenth and a quarter of an inch possible.

Sunday night

A slight chance of rain, mainly before 11 PM. Mostly cloudy, with a low around . North wind around 15 miles per hour. Patchy frost possible. Chance of rain 24%.

Monday

Mostly sunny, with a high near . North wind around 10 miles per hour.

Monday night

Mostly clear, with a low around . Northwest wind around 5 miles per hour, becoming northeast. A freeze is likely.

Tuesday

Partly sunny, with a high near . Northeast wind around 10 miles per hour.

Tuesday night

Mostly cloudy, with a low around . East wind around 10 miles per hour. A freeze is likely.

Wednesday

A chance of rain. Cloudy, with a high near . East wind around 10 miles per hour. Chance of rain 37%.

Wednesday night

A chance of rain. Partly cloudy, with a low around . East wind around 10 miles per hour. A freeze is likely. Chance of rain 39%.

The highs, lows and rain chances are our forecast, the same numbers as the table above. The sky and wind words come from the four weather models' average, so they are only as good as the models' clouds. Wind is at the models' standard height of 33 feet; the station's sheltered sensor usually reads less.

Why we say this

Every high and low we forecast starts from the average of four big weather models. Then it is adjusted for how forecasts like this one have actually turned out at this station over about 900 past days. Here is that working for the next three highs and lows: where the models started, how far we moved it, and the biggest reasons.

Tonight

The low temperature
38°

The four models averaged 39.7°. We moved it 1.4 degrees colder.

The biggest reasons:

  • A clear sky with light wind: 1.0° colder.
  • This station's usual correction to the models: 0.6° colder.
  • The cloud cover expected: 0.4° warmer.

This low has already begun. The number here is the latest hourly build; the forecast of record was made just before it began, and the number updated with the station's readings since says 38°.

Every adjustment, step by step
Models' average (the blend) 39.7°start
Clear & calm (radiational cooling) −1.0°
Station baseline −0.6°
Expected cloud cover +0.4°
Season −0.3°
Expected wind +0.1°
Dryness (dew-point depression) +0.1°
How warm or cold the blend runs +0.1°
Model disagreement +0.1°
Models' recent error here +0.0°
Our forecast38.3°= blend −1.4°, shown as 38°

Start with the models' average, add each row (blue pushed it warmer, orange cooler), and you arrive at our forecast. Each row compares this period with an ordinary one: a factor that is typical adds nothing. The rows are rounded, so adding them by hand can be off by a tenth.

This forecast, run by run
3 days ago2 days agoa day agolatest run
Ours41°39°36°38°
National Weather Service41°40°40°39°

Our forecast as each run issued it, before any update from the station.

Friday

The high temperature
69°

The four models averaged 66.7°. We moved it 2.2 degrees warmer.

The biggest reasons:

  • The sunshine expected: 1.9° warmer.
  • This station's usual correction to the models: 1.5° warmer.
  • The time of year: 1.1° colder.
Every adjustment, step by step
Models' average (the blend) 66.7°start
Forecast sunshine +1.9°
Station baseline +1.5°
Season −1.1°
Dryness (dew-point depression) −0.3°
Instability (CAPE) +0.2°
Clear & calm (radiational cooling) +0.1°
Model disagreement −0.1°
Models' recent error here +0.1°
Expected cloud cover −0.1°
How warm or cold the blend runs +0.0°
Expected wind +0.0°
Our forecast68.9°= blend +2.2°, shown as 69°

Start with the models' average, add each row (blue pushed it warmer, orange cooler), and you arrive at our forecast. Each row compares this period with an ordinary one: a factor that is typical adds nothing. The rows are rounded, so adding them by hand can be off by a tenth.

This forecast, run by run
3 days ago2 days agoa day agolatest run
Ours67°68°68°69°
National Weather Service65°67°67°68°

Our forecast as each run issued it, before any update from the station.

Friday night

The low temperature
39°

The four models averaged 40.8°. We moved it 1.6 degrees colder.

The biggest reasons:

  • The time of year: 1.1° colder.
  • This station's usual correction to the models: 0.5° colder.
  • How warm or cold the models' own number is: 0.4° warmer.
Every adjustment, step by step
Models' average (the blend) 40.8°start
Season −1.1°
Station baseline −0.5°
How warm or cold the blend runs +0.4°
Clear & calm (radiational cooling) −0.3°
Expected cloud cover −0.3°
Dryness (dew-point depression) +0.1°
Models' recent error here +0.1°
Model disagreement +0.0°
Expected wind +0.0°
Our forecast39.2°= blend −1.6°, shown as 39°

Start with the models' average, add each row (blue pushed it warmer, orange cooler), and you arrive at our forecast. Each row compares this period with an ordinary one: a factor that is typical adds nothing. The rows are rounded, so adding them by hand can be off by a tenth.

This forecast, run by run
3 days ago2 days agoa day agolatest run
Ours44°41°39°39°
National Weather Service42°43°42°41°

Our forecast as each run issued it, before any update from the station.

"Warmer" and "colder" compare this day or night with an ordinary one: a factor that is typical for the time of year adds nothing. The adjustments were all learned together from the station's own record, and each one was tested on a year of days the model had never seen before it was used. Tap the small "i" beside any step for what it means.

The discussion written Thursday, November 19, 7:05 PM

SynopsisOver the next week, highs range from 40° on Wednesday to 69° on Friday. The biggest change comes Monday, when highs fall about 16° from Sunday's. A drop that sharp usually means a cold front has come through. Rain is most likely Sunday.

TemperaturesTonight's low so far is 52°. Going by how the station is running, we now expect about 38°.

Friday's high: we say 69°, 2° warmer than the models' average (67°). The main reason: it should be a sunny day, and the models tend to undersell how warm sunny days get here. It's partly offset by the time of year; the models usually run a little warm here this season.

Friday's low: we say 39°, 2° colder than the models' average (41°). The main reason: the time of year; the models usually run a little warm here this season. It's partly offset by a habit of the models, which at this temperature usually end up a little too low here.

Saturday's high: we say 68°, 3° warmer than the models' average (66°). The main reason: it should be a sunny day, and the models tend to undersell how warm sunny days get here. It's partly offset by the time of year; the models usually run a little warm here this season.

Saturday's low: we say 44°, 3° colder than the models' average (47°). The main reason: the air will be dry, and dry air cools off quickly at night. A smaller push in the same direction: the time of year; the models usually run a little warm here this season.

We differ from the National Weather Service on Saturday's low: we say 44°, they say 48°.

PrecipitationSunday, there's a fair chance of rain: a 63% chance your gauge catches at least a hundredth of an inch. 89% of the 122 ensemble runs show some rain, but almost none reach a quarter inch, so any rain should be light. Our chance is lower than the ensemble's because, in this gauge's history, model runs showing only a little rain have often left it dry.

Sunday night, rain is possible: a 24% chance your gauge catches at least a hundredth of an inch. 66% of the 122 ensemble runs show some rain, but almost none reach a quarter inch, so any rain should be light. Our chance is again lower than the ensemble's, for the same reason.

HazardsHeavy rain (a quarter inch or more in 12 hours): elevated Sunday. Over the past year, when we've said elevated, it happened about 1 time in 6.

Tonight: a small chance of frost (13%). Low near 38°.

Friday night: a small chance of frost (10%). Low near 39°.

Sunday night: a good chance of frost (36%). Low near 37°. Frost forms most readily when the sky clears and the wind goes calm.

Monday night: frost is very likely (99%), and a freeze is very likely (85%). Low near 29°.

Tuesday night: frost is very likely (99%), and a freeze is very likely (87%). Low near 28°.

Wednesday night: frost is likely (79%), with a good chance of a freeze (54%). Low near 32°.

ConfidenceThe ensemble is the same forecast run 122 times from slightly different starting points; the closer the runs agree, the more the forecast can be trusted. For the daytime highs, here is where 8 out of 10 runs land: Friday, very sure (64° to 68°); Saturday, very sure (63° to 68°); Sunday, fairly sure (56° to 65°); Monday, very sure (43° to 49°).

FrontsA front is probably coming through around 9 PM Sunday. All 4 models show it, and calls like this have verified 63% of the time here. Behind it, expect noticeably drier air (dew point down about 12°) and a wind shift to the northwest. A front is possible around 6 AM Tuesday. 2 of the 4 models show it, and calls like this have verified 46% of the time here. Behind it, expect somewhat drier air (dew point down about 6°) and a wind shift to the northeast.

Written automatically from the numbers each hour: where our forecast differs from the weather models and the National Weather Service and why, any hazards, fronts, and how the forecast has changed. It says only what the numbers show.

On this page:The next 7 daysThe forecast in wordsWhy we say thisThe discussion Ours next to the National Weather Service and WxSimMidnight to midnightHow the forecast is made
More detail

Ours next to the National Weather Service and WxSim made Thursday, November 19, 7:05 PM

Daytime high
7 AM to 7 PM
Overnight low
that night, 7 PM to 8 AM
Chance of rain, dayChance of rain, night
Day OursNational Weather ServiceWxSim OursNational Weather ServiceWxSim OursNational Weather ServiceOursNational Weather Service
Thursday, November 19 was 66°
± 3°
made at 7 PM; now 38° ± 2°
so far: 52°
39° 34° 0.00 inch0.00 inch 7%0%
Friday, November 20
± 3°
68° 64°
± 3°
41° 34° 6%0% 7%0%
Saturday, November 21
± 3°
68° 68°
± 3°
48° 38° 6%0% 7%4%
Sunday, November 22
± 5°
63° 66°
± 4°
36° 39° 63% 0.13 inch73% 0.02 inch 24%31%
Monday, November 23
± 5°
46° 47°
± 4°
27° 30° 11%4% 8%0%
Tuesday, November 24
± 5°
41° -
± 5°
27° - 10%0% 13%5%
Wednesday, November 25
± 6°
40° -
± 7°
27° - 37% 0.07 inch22% 39% 0.10 inch24%

Once a high or low period has begun, the big number is the forecast of record, made just before it began: that is the one How we're doing grades, and it does not change. Under it is the number updated every hour with how warm or cool the station is running against the forecast (it never goes below the high so far, or above the low so far), and the high or low so far. The periods are fixed in universal time, the way the National Weather Service counts them, so they run an hour later during daylight saving time.

The chance of rain means at least a hundredth of an inch in the station's gauge during that 12-hour period; amounts show when the chance is 30% or more. The National Weather Service forecast is its point forecast for this spot.

The high and low from midnight to midnight

DayHighLow
Thursday, November 19
± 3°
made at midnight; now 66°
so far: 66°
± 3°
made at midnight; now 37°
so far: 37°
Friday, November 20
± 2°
± 3°
Saturday, November 21
± 3°
± 4°
Sunday, November 22
± 5°
± 3°
Monday, November 23
± 4°
± 4°
Tuesday, November 24
± 6°
± 4°
Wednesday, November 25
± 6°
± 6°

The same day measured the way the station's records and the NOAA monthly report count it. On days a front comes through, the high can land just after midnight, so these can differ from the daytime high and overnight low above.

How the forecast is made

Four global weather models are averaged: the American model (GFS), the European model (ECMWF), the German model (ICON) and the Canadian model (GEM). That average is then corrected with this station's own record: how each kind of day and night has actually turned out here. The ± after each number is the range the real temperature lands inside 8 times in 10. It is wider when the models disagree.