• Stormtrack's forum runs on Xenforo forum software, which will be undergoing a major update the evening of Wednesday, Feb 28th. The site may be down for a period while that update takes place.

First frost......no accurate temp!!!


I was expecting it to get down into the mid-30s here in central AR over night. The NWS had forecasted 35oF and then 38oF for our location overnight tonight.

I turn on TWC this morning to see what their cable display says it is for us. 37oF. OK - fair enough I think. Then - TWC puts on a different "Current Conditions" map which shows us at 34oF. OK. So there's a disparity.

Well - I just stuck my head out of the front door while waiting on my tea steeping - and lo and behold all the cars and topw of our apartments are COVERED in pretty thick white frost! :eek: Not only did this cold front bring us our first true cold snap - but also our first real, tangible FROST. On October 13th. I mean - it's bad enough to where you'd really be advised to go scrape the car......if any of us actually had any time for that! :p

So how do I sort through this Jackson Pollock of temperature readings and make my own semi-intelligent guess as to what our *actual* low was this morning in Conway, AR? I mean - without consulting my own home indoor/outdoor temperature sensor......which I don't have......yet? Can this relatively thick frost form without it being that close to freezing? If not - I'm thinking that in localized areas (of which ours is one) we really got down to 32oF - at least! :confused:

Not necessarily. You can get frost with official temps as warm as 37 or 38. This morning we had our first frost as well and all the local thermometers were in the 36 to 38 range.

Temperatures are recorded at 6 feet, while the frost is occuring at ground level. Also, there are significant effects from the humidity, wet bulb, etc.
TWC does not use actual temperature reports, they modify them based on surrounding obs. Many TV stations are using software that does the same thing. I hate it, but viewers demand local information...

Your best source of accurate temps would be the local METAR report, available from the NWS website.
i was looking up the average date for first frost on the nws little rock website and it said "the dates shown are based on climatological normals, with a temperature of 36° for frost and 32° for a freeze." looks like conway beat their average by 8 days for first frost (oct 21). it got to 31 here... (1650 feet - 7 miles northeast of Ozark, AR). I never really noticed before that you could get frost with temp so high.
As noted above, the official temp obs are taken at 2m. When conditions are favorable (light/calm winds and clear skies), radiational cooling can result in a very strong vertical temperature gradient, sometimes on the order of 5-6 degrees F over 2m. Therefore, it's quite possible (and not entirely uncommon) to see frost on the ground when the official temperature is in the 32-36F range.

In addition, there's another problem, one that affects more than just frost potential. The observations that we get are point observations, which we must assume, to varying degrees, are representative of the entire area. Unfortunately, there is plenty of small-scale variability that affects the representativeness of a single point observation. For example, you'll most likely see different T and Td if you measure right next to a lake and compare it with an observation taken away from the lake (even just 500 feet away from the lake). One of my professors a couple years ago showed us the T and Td trace he took while driving along a local highway (Hwy 9 in OUN, FWIW) -- the temperature varied 2-3 degrees over the course of just several hundred feet in some cases! A single point observation is more representative of a larger area if the area is the same elevation, land type (forest, marsh, urban, etc), etc, of the area where the ob was taken. In a few months (when it gets colder), compare the OKC observation with nearby obs on a night with northeast winds (and, preferably, following a day with full sunshine). The OKC METAR, taken at the airport on the southwest side of the city, can be several degrees higher than nearby observations, mainly because the northeast winds bring urban-heated air right through the sensor location. If you are forecasting for a point location like OKC, it's important to remember such quirks.

Local effects tend to become very important when winds go to calm and skies are clear. Check out the OK Mesonet the next time we see a calm winds, clear skies night, and you'll see that the T will vary 5-6+ degrees over the distance of 15-20 miles. Nearby micronets would also shows 4+ degrees variability over several hundred feet as well.
Last edited by a moderator:
This also helps explain the difference between a frost and a freeze. Frost warnings may go up when temps are as warm as 38 (under the right conditions). There may be scattered or widespread frost, yet no 2 meter observations measured sub freezing temeperatures.

A freeze warning means that temperatures are likely to drop below 32, and 2 meter obs will likely see readings less than 32.
Back in 2001 during a very cold, calm night in Norman I drove all around parts of Norman and the surrounding countryside with an temperature sensor kit designed for aircraft use. I got very surprising variations... down to 18F in gulleys, 24F on ridges, and a homogenous 27F in town. Location makes a tremendous difference. I'll see if I can find the notes from that experiment... doubtful, but if it turns up I'll post it.