Faulty thermometers exaggerated western U.S. mountain warming
Incorrect temperatures were used in snowpack and ecology research
TAINTED TEMPERATURES High-elevation warming in the western United States was artificially inflated by flawed temperature sensors at stations such as this one on Mount Eyak in Alaska, new research suggests.
Daniel Fisher/USDA
Observations of sharply rising high-elevation temperatures in the western United States were caused by faulty equipment, not climate change, new research suggests.
From 1991 to 2012, the National Water and Climate Center’s Snow Telemetry network, or SNOTEL, reported a 1.16 degree Celsius per decade climb in minimum temperatures at high elevations throughout the mountainous American West. Correcting for flawed temperature sensors, which overstated temperatures by as much as 2 degrees Celsius, new research published January 13 in Geophysical Research Letters reduces this decadal increase to roughly 0.106 degrees. That’s roughly in line with warming at lower altitudes. While not used in global climate research, the flawed data were used by ecologists, hydrologists and regional climate researchers, says lead author Jared Oyler, an earth scientist at the University of Montana in Missoula.
“This is definitely the biggest networkwide bias I’ve ever seen,” Oyler says. “To have it be one-half to two degrees across the whole network — that’s a significant change.”
SNOTEL’s operators were aware of a possible temperature discrepancy, but didn’t know the full extent of the issue until now,says Michael Strobel, director of the National Water and Climate Center in Portland, Ore. The agency provides documentation and attempts to warn users about the potential problem, he says, but many researchers download the publicly available dataset and use it “without doing their due diligence and checking its validity.”
The SNOTEL network measures snow volume and conditions on mountains throughout the 12 westernmost states. Its primary goal is to forecast water availability for agriculture in the spring and summer, not to record long-term climate trends. For much of the area covered by the network’s 885 remotely operated stations, though, SNOTEL is the only available data source at high elevations.
In 2013, Oyler noticed something odd as he scrutinized raw historical temperature data collected by SNOTEL: The minimum observed temperature at stations would suddenly jump a degree or more compared with measurements taken outside the network. Investigating further, he found that the temperature jumps coincided with a systemwide switch to new temperature sensors in the late 1990s and early 2000s.
Comparing measurements from the two generations of sensors, Oyler calculated that the new equipment recorded relatively higher temperatures in colder conditions. For instance, at zero degrees the new sensors read around 1.4°. Because the upgraded equipment rolled out over several years, the change created an artificial warming trend across the entire network amounting to about 1 degree a decade.
This increase seemed realistic given a prediction that higher elevations would warm faster during climate change as reflective snow melted and newly exposed rock absorbed more sunlight, Oyler says. He estimates that the flawed SNOTEL data inflated the perceived increase in warming for every 500 meter rise in elevation in the United States by as much as 562 percent.
The inaccurate measurements were not used by global climate scientists, who Oyler says would have quickly flagged the data for inconsistencies. The SNOTEL dataset was, however, used by ecologists and hydrologists studying the impacts of climate change, he says, and by climate researchers producing detailed regional climate predictions based on results from global climate simulations. Many of those studies may have to be reexamined for erroneous conclusions.
Oyler says the agency plans to test the old and new sensors in a controlled laboratory environment to find the precise actual temperatures corresponding to the network’s measurements and will provide that corrected data. Properly calibrated temperatures will become increasingly important as SNOTEL’s dataset matures, says climate scientist John Abatzoglou of the University of Idaho in Moscow. “SNOTEL is nearing 30 years of temperature data,” he says. “In climatology, 30 years is a magic number where people start using data more and more for climate change analysis, so this is coming at a good time.”
While Oyler’s work suggests higher elevations are warming in step with lower elevations, that might not be true, says Henry Diaz, a climate scientist at the National Oceanic and Atmospheric Administration’s Earth System Research Laboratory in Boulder, Colo. In the interior western United States, the minimum altitude for freezing temperatures has risen by about 100 meters since the early 1980s, he says. This corresponds to a faster rate of overall warming at higher elevations of about 0.15 degree per decade, though this is still significantly less than projected using the uncorrected SNOTEL temperature data.
