Temperature Readings Just Improved!

Custom Davis radiation shield

There was a Problem?

So perhaps you’ve noticed lately on sunny days forecasted highs have been lower than the actual highs by several degrees. Well… it turns out that’s not entirely true.

During the deepest part of winter when the polar vortex was regularly occurring (late December into January), we were having equipment issues. Normally, we measure temperatures inside an enclosure that has a fan running on solar power during the day and batteries at night. Unfortunately, the weather destroyed the electronics that controlled the battery charging, among other casualties.

As a quick fix, we installed a passive (fan-less) shield for our sensor. Passive shelters are fine during our normally very cloudy, windy winters. Wouldn’t you know it, the sun decided to show up and, along with it, gentle breezes! That makes for beautiful winter days in Marquette. It is also a recipe for solar-induced sensor errors.

The snow acts like a mirror to the sun and the reflected radiation gets into the sensor and heats it up. How much? A rough estimate, based on before and after tests of running with and without a fan, is 3-5 degrees F. Studies have shown that, in fact, on completely calm, sunny days over a fresh blanket of snow, errors can reach up to 18 deg F!

The Solution

Those same studies also prescribe running a fan to generate air movement across the sensor to combat sensor warming.

We had planned to resurrect the fan-aspirated radiation shield during warmer weather. Seems that weather just arrived!

We needed something reliable. Mechatronics makes an AC fan that’s rated down to -40F, designed to resist moisture, and boasts 50,000 hours mean time between failures. It draws only 11W of power.  Perfect!

A big shout out goes to Randy for field testing this solution in the hot Nebraska summers (90s are common) and freezing cold winters (frequently below zero).  He even put a heat gun to the fan to be sure it wasn’t generating heat into the sensor chamber.

With 24 cubic feet per minute of air movement, radiational heating and cooling of the sensor should be negligible. We look forward to better alignment between our measured highs and lows and forecasted numbers.* We’re already seeing improvements over the last 24 hours.

* You can always check the accuracy of forecasts that appear on the home page.

Is Marquette Snowier Than We Think?

weather instrumentation facing northwest
Snow board looking NW

Over the last 30 years, the city has averaged 117.2” per year of snow. That’s based on measurements taken every morning close to the lakeshore (at the Waste Water Treatment Plant which houses the official COOP weather station for Marquette, MI).

A few days ago we added a “Seasonal Snow Total” entry to the Almanac section on the home page. This should provide some insight into the kind of winter we are having. Currently, the official weather station is seeing below average snowfall so far (55.1″) while we are experiencing above normal totals (83.6″).

Here at, located 1/4 mile south of downtown and approximately 100 ft above the lakeshore, we measure up to 4x daily.  According to the Community Collaborative Rain, Hail & Snow Network (CoCoRaHS), if snow stops falling midday, measuring the next morning can mean up to a 50% reduction in measurement (see below).  Continue reading “Is Marquette Snowier Than We Think?”