Heat waves distort a picture of airplanes waiting to take off from Reagan National Airport. (Jim Lo Scalzo/European Pressphoto Agency)

Responding to concerns that Reagan National Airport’s temperature sensor was reporting erroneously warm temperature readings, the National Weather Service (NWS) installed a new sensor on Monday.

“We have a new temperature sensor out there,” said Jim Lee, meteorologist-in-charge at the NWS forecast office in Sterling, Va., which serves the Washington and Baltimore metro regions. “We want to see if it helps anything.”

In recent weeks, a number of local constituents — including the Capital Weather Gang, former Virginia state climatologist Pat Michaels, and WUSA9 meteorologist Howard Bernstein — had voiced concerns that the airport readings were too warm compared to neighboring weather stations. For example, Reagan National had registered more days at or above 90 degrees than any location in the region through early August.

[Is D.C.’s weather station reading way too hot?]

90-degree days through Aug. 3. This map shows WBAN stations. It includes a number that are not long-term climatology stations, but also all main climate sites. (Author analysis of NOAA data)

“Through the comments we were seeing [from scientists and the media], we definitely wanted to send a team there,” Lee said.

Technicians arrived at Reagan National on Monday and discovered the temperature sensor within the airport’s Automated Surface Observation System (ASOS) was reading 1.7 degrees warmer than a reference temperature sensor, known as a psychro-dyne (an instrument roughly the size of a small suitcase).

“It looked like it [the sensor] was reading 1.5-2 degrees too high,” Lee said, noting the ASOS sensor was reporting 78.7 degrees while the psychro-dyne registered 77 degrees during initial testing.

Lee said he authorized technicians to replace the sensor given the difference along with the fact “a couple of  its [the sensor’s] cooling fins were bent” (though he doesn’t think that was obstructing airflow or affecting readings). The sensor was last replaced in 2010, he said.

Lee said the National Weather Service inspects ASOS sites every 90 days to conduct preventative maintenance. Prior to Monday, Reagan National’s last check was on June 9 when there was only a 0.4-degree difference between the temperature sensor and the psychro-dyne reference measurement.

Lee explained that the temperature measurement at an ASOS site is within technical specification as long as its temperature readings are within (plus or minus) 5 degrees of the reference measurement. Both the ASOS and the psychro-dyne sensors have an error of plus or minus 1.8 degrees. Lee noted that, even though he decided it was time to install a new sensor Monday, Reagan National’s temperature readings had not strayed from specification then or on any prior inspection.

After the new sensor was installed Monday, the new temperature reported at Reagan National dropped to 76.5 degrees compared to the 78.7 degrees reported prior – although Lee said some of that change was due to changing sky conditions and a later time, rather than solely the change in equipment.

The installation of a new sensor that reads cooler than the previous sensor introduces complicated questions about the integrity of the weather records maintained at the airport. If the temperature sensor was previously registering values anywhere between about 0.5 and 2 degrees too warm, does that invalidate records that were set in previous months or even years?

For example, May 2015 was said to be the warmest on record at Reagan National, edging out May 1991 by just 0.2 degrees. Should a correction be applied and May 1991 reclaim its spot as warmest May ever recorded?

Lee said no — that corrections will not be applied to past data and that doing so would not be scientifically defensible. “It boils down to the accuracy of the sensor which is only [plus or minus] 1.8 degrees,” Lee said. “There’s inherent error in these readings. That’s science. These sensors have limitations.”

In other words, whenever a new record is set that is within a couple degrees of a previous record, it may or may not actually be a record. Take records with a grain of salt.